pugixml.cpp 348 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427242824292430243124322433243424352436243724382439244024412442244324442445244624472448244924502451245224532454245524562457245824592460246124622463246424652466246724682469247024712472247324742475247624772478247924802481248224832484248524862487248824892490249124922493249424952496249724982499250025012502250325042505250625072508250925102511251225132514251525162517251825192520252125222523252425252526252725282529253025312532253325342535253625372538253925402541254225432544254525462547254825492550255125522553255425552556255725582559256025612562256325642565256625672568256925702571257225732574257525762577257825792580258125822583258425852586258725882589259025912592259325942595259625972598259926002601260226032604260526062607260826092610261126122613261426152616261726182619262026212622262326242625262626272628262926302631263226332634263526362637263826392640264126422643264426452646264726482649265026512652265326542655265626572658265926602661266226632664266526662667266826692670267126722673267426752676267726782679268026812682268326842685268626872688268926902691269226932694269526962697269826992700270127022703270427052706270727082709271027112712271327142715271627172718271927202721272227232724272527262727272827292730273127322733273427352736273727382739274027412742274327442745274627472748274927502751275227532754275527562757275827592760276127622763276427652766276727682769277027712772277327742775277627772778277927802781278227832784278527862787278827892790279127922793279427952796279727982799280028012802280328042805280628072808280928102811281228132814281528162817281828192820282128222823282428252826282728282829283028312832283328342835283628372838283928402841284228432844284528462847284828492850285128522853285428552856285728582859286028612862286328642865286628672868286928702871287228732874287528762877287828792880288128822883288428852886288728882889289028912892289328942895289628972898289929002901290229032904290529062907290829092910291129122913291429152916291729182919292029212922292329242925292629272928292929302931293229332934293529362937293829392940294129422943294429452946294729482949295029512952295329542955295629572958295929602961296229632964296529662967296829692970297129722973297429752976297729782979298029812982298329842985298629872988298929902991299229932994299529962997299829993000300130023003300430053006300730083009301030113012301330143015301630173018301930203021302230233024302530263027302830293030303130323033303430353036303730383039304030413042304330443045304630473048304930503051305230533054305530563057305830593060306130623063306430653066306730683069307030713072307330743075307630773078307930803081308230833084308530863087308830893090309130923093309430953096309730983099310031013102310331043105310631073108310931103111311231133114311531163117311831193120312131223123312431253126312731283129313031313132313331343135313631373138313931403141314231433144314531463147314831493150315131523153315431553156315731583159316031613162316331643165316631673168316931703171317231733174317531763177317831793180318131823183318431853186318731883189319031913192319331943195319631973198319932003201320232033204320532063207320832093210321132123213321432153216321732183219322032213222322332243225322632273228322932303231323232333234323532363237323832393240324132423243324432453246324732483249325032513252325332543255325632573258325932603261326232633264326532663267326832693270327132723273327432753276327732783279328032813282328332843285328632873288328932903291329232933294329532963297329832993300330133023303330433053306330733083309331033113312331333143315331633173318331933203321332233233324332533263327332833293330333133323333333433353336333733383339334033413342334333443345334633473348334933503351335233533354335533563357335833593360336133623363336433653366336733683369337033713372337333743375337633773378337933803381338233833384338533863387338833893390339133923393339433953396339733983399340034013402340334043405340634073408340934103411341234133414341534163417341834193420342134223423342434253426342734283429343034313432343334343435343634373438343934403441344234433444344534463447344834493450345134523453345434553456345734583459346034613462346334643465346634673468346934703471347234733474347534763477347834793480348134823483348434853486348734883489349034913492349334943495349634973498349935003501350235033504350535063507350835093510351135123513351435153516351735183519352035213522352335243525352635273528352935303531353235333534353535363537353835393540354135423543354435453546354735483549355035513552355335543555355635573558355935603561356235633564356535663567356835693570357135723573357435753576357735783579358035813582358335843585358635873588358935903591359235933594359535963597359835993600360136023603360436053606360736083609361036113612361336143615361636173618361936203621362236233624362536263627362836293630363136323633363436353636363736383639364036413642364336443645364636473648364936503651365236533654365536563657365836593660366136623663366436653666366736683669367036713672367336743675367636773678367936803681368236833684368536863687368836893690369136923693369436953696369736983699370037013702370337043705370637073708370937103711371237133714371537163717371837193720372137223723372437253726372737283729373037313732373337343735373637373738373937403741374237433744374537463747374837493750375137523753375437553756375737583759376037613762376337643765376637673768376937703771377237733774377537763777377837793780378137823783378437853786378737883789379037913792379337943795379637973798379938003801380238033804380538063807380838093810381138123813381438153816381738183819382038213822382338243825382638273828382938303831383238333834383538363837383838393840384138423843384438453846384738483849385038513852385338543855385638573858385938603861386238633864386538663867386838693870387138723873387438753876387738783879388038813882388338843885388638873888388938903891389238933894389538963897389838993900390139023903390439053906390739083909391039113912391339143915391639173918391939203921392239233924392539263927392839293930393139323933393439353936393739383939394039413942394339443945394639473948394939503951395239533954395539563957395839593960396139623963396439653966396739683969397039713972397339743975397639773978397939803981398239833984398539863987398839893990399139923993399439953996399739983999400040014002400340044005400640074008400940104011401240134014401540164017401840194020402140224023402440254026402740284029403040314032403340344035403640374038403940404041404240434044404540464047404840494050405140524053405440554056405740584059406040614062406340644065406640674068406940704071407240734074407540764077407840794080408140824083408440854086408740884089409040914092409340944095409640974098409941004101410241034104410541064107410841094110411141124113411441154116411741184119412041214122412341244125412641274128412941304131413241334134413541364137413841394140414141424143414441454146414741484149415041514152415341544155415641574158415941604161416241634164416541664167416841694170417141724173417441754176417741784179418041814182418341844185418641874188418941904191419241934194419541964197419841994200420142024203420442054206420742084209421042114212421342144215421642174218421942204221422242234224422542264227422842294230423142324233423442354236423742384239424042414242424342444245424642474248424942504251425242534254425542564257425842594260426142624263426442654266426742684269427042714272427342744275427642774278427942804281428242834284428542864287428842894290429142924293429442954296429742984299430043014302430343044305430643074308430943104311431243134314431543164317431843194320432143224323432443254326432743284329433043314332433343344335433643374338433943404341434243434344434543464347434843494350435143524353435443554356435743584359436043614362436343644365436643674368436943704371437243734374437543764377437843794380438143824383438443854386438743884389439043914392439343944395439643974398439944004401440244034404440544064407440844094410441144124413441444154416441744184419442044214422442344244425442644274428442944304431443244334434443544364437443844394440444144424443444444454446444744484449445044514452445344544455445644574458445944604461446244634464446544664467446844694470447144724473447444754476447744784479448044814482448344844485448644874488448944904491449244934494449544964497449844994500450145024503450445054506450745084509451045114512451345144515451645174518451945204521452245234524452545264527452845294530453145324533453445354536453745384539454045414542454345444545454645474548454945504551455245534554455545564557455845594560456145624563456445654566456745684569457045714572457345744575457645774578457945804581458245834584458545864587458845894590459145924593459445954596459745984599460046014602460346044605460646074608460946104611461246134614461546164617461846194620462146224623462446254626462746284629463046314632463346344635463646374638463946404641464246434644464546464647464846494650465146524653465446554656465746584659466046614662466346644665466646674668466946704671467246734674467546764677467846794680468146824683468446854686468746884689469046914692469346944695469646974698469947004701470247034704470547064707470847094710471147124713471447154716471747184719472047214722472347244725472647274728472947304731473247334734473547364737473847394740474147424743474447454746474747484749475047514752475347544755475647574758475947604761476247634764476547664767476847694770477147724773477447754776477747784779478047814782478347844785478647874788478947904791479247934794479547964797479847994800480148024803480448054806480748084809481048114812481348144815481648174818481948204821482248234824482548264827482848294830483148324833483448354836483748384839484048414842484348444845484648474848484948504851485248534854485548564857485848594860486148624863486448654866486748684869487048714872487348744875487648774878487948804881488248834884488548864887488848894890489148924893489448954896489748984899490049014902490349044905490649074908490949104911491249134914491549164917491849194920492149224923492449254926492749284929493049314932493349344935493649374938493949404941494249434944494549464947494849494950495149524953495449554956495749584959496049614962496349644965496649674968496949704971497249734974497549764977497849794980498149824983498449854986498749884989499049914992499349944995499649974998499950005001500250035004500550065007500850095010501150125013501450155016501750185019502050215022502350245025502650275028502950305031503250335034503550365037503850395040504150425043504450455046504750485049505050515052505350545055505650575058505950605061506250635064506550665067506850695070507150725073507450755076507750785079508050815082508350845085508650875088508950905091509250935094509550965097509850995100510151025103510451055106510751085109511051115112511351145115511651175118511951205121512251235124512551265127512851295130513151325133513451355136513751385139514051415142514351445145514651475148514951505151515251535154515551565157515851595160516151625163516451655166516751685169517051715172517351745175517651775178517951805181518251835184518551865187518851895190519151925193519451955196519751985199520052015202520352045205520652075208520952105211521252135214521552165217521852195220522152225223522452255226522752285229523052315232523352345235523652375238523952405241524252435244524552465247524852495250525152525253525452555256525752585259526052615262526352645265526652675268526952705271527252735274527552765277527852795280528152825283528452855286528752885289529052915292529352945295529652975298529953005301530253035304530553065307530853095310531153125313531453155316531753185319532053215322532353245325532653275328532953305331533253335334533553365337533853395340534153425343534453455346534753485349535053515352535353545355535653575358535953605361536253635364536553665367536853695370537153725373537453755376537753785379538053815382538353845385538653875388538953905391539253935394539553965397539853995400540154025403540454055406540754085409541054115412541354145415541654175418541954205421542254235424542554265427542854295430543154325433543454355436543754385439544054415442544354445445544654475448544954505451545254535454545554565457545854595460546154625463546454655466546754685469547054715472547354745475547654775478547954805481548254835484548554865487548854895490549154925493549454955496549754985499550055015502550355045505550655075508550955105511551255135514551555165517551855195520552155225523552455255526552755285529553055315532553355345535553655375538553955405541554255435544554555465547554855495550555155525553555455555556555755585559556055615562556355645565556655675568556955705571557255735574557555765577557855795580558155825583558455855586558755885589559055915592559355945595559655975598559956005601560256035604560556065607560856095610561156125613561456155616561756185619562056215622562356245625562656275628562956305631563256335634563556365637563856395640564156425643564456455646564756485649565056515652565356545655565656575658565956605661566256635664566556665667566856695670567156725673567456755676567756785679568056815682568356845685568656875688568956905691569256935694569556965697569856995700570157025703570457055706570757085709571057115712571357145715571657175718571957205721572257235724572557265727572857295730573157325733573457355736573757385739574057415742574357445745574657475748574957505751575257535754575557565757575857595760576157625763576457655766576757685769577057715772577357745775577657775778577957805781578257835784578557865787578857895790579157925793579457955796579757985799580058015802580358045805580658075808580958105811581258135814581558165817581858195820582158225823582458255826582758285829583058315832583358345835583658375838583958405841584258435844584558465847584858495850585158525853585458555856585758585859586058615862586358645865586658675868586958705871587258735874587558765877587858795880588158825883588458855886588758885889589058915892589358945895589658975898589959005901590259035904590559065907590859095910591159125913591459155916591759185919592059215922592359245925592659275928592959305931593259335934593559365937593859395940594159425943594459455946594759485949595059515952595359545955595659575958595959605961596259635964596559665967596859695970597159725973597459755976597759785979598059815982598359845985598659875988598959905991599259935994599559965997599859996000600160026003600460056006600760086009601060116012601360146015601660176018601960206021602260236024602560266027602860296030603160326033603460356036603760386039604060416042604360446045604660476048604960506051605260536054605560566057605860596060606160626063606460656066606760686069607060716072607360746075607660776078607960806081608260836084608560866087608860896090609160926093609460956096609760986099610061016102610361046105610661076108610961106111611261136114611561166117611861196120612161226123612461256126612761286129613061316132613361346135613661376138613961406141614261436144614561466147614861496150615161526153615461556156615761586159616061616162616361646165616661676168616961706171617261736174617561766177617861796180618161826183618461856186618761886189619061916192619361946195619661976198619962006201620262036204620562066207620862096210621162126213621462156216621762186219622062216222622362246225622662276228622962306231623262336234623562366237623862396240624162426243624462456246624762486249625062516252625362546255625662576258625962606261626262636264626562666267626862696270627162726273627462756276627762786279628062816282628362846285628662876288628962906291629262936294629562966297629862996300630163026303630463056306630763086309631063116312631363146315631663176318631963206321632263236324632563266327632863296330633163326333633463356336633763386339634063416342634363446345634663476348634963506351635263536354635563566357635863596360636163626363636463656366636763686369637063716372637363746375637663776378637963806381638263836384638563866387638863896390639163926393639463956396639763986399640064016402640364046405640664076408640964106411641264136414641564166417641864196420642164226423642464256426642764286429643064316432643364346435643664376438643964406441644264436444644564466447644864496450645164526453645464556456645764586459646064616462646364646465646664676468646964706471647264736474647564766477647864796480648164826483648464856486648764886489649064916492649364946495649664976498649965006501650265036504650565066507650865096510651165126513651465156516651765186519652065216522652365246525652665276528652965306531653265336534653565366537653865396540654165426543654465456546654765486549655065516552655365546555655665576558655965606561656265636564656565666567656865696570657165726573657465756576657765786579658065816582658365846585658665876588658965906591659265936594659565966597659865996600660166026603660466056606660766086609661066116612661366146615661666176618661966206621662266236624662566266627662866296630663166326633663466356636663766386639664066416642664366446645664666476648664966506651665266536654665566566657665866596660666166626663666466656666666766686669667066716672667366746675667666776678667966806681668266836684668566866687668866896690669166926693669466956696669766986699670067016702670367046705670667076708670967106711671267136714671567166717671867196720672167226723672467256726672767286729673067316732673367346735673667376738673967406741674267436744674567466747674867496750675167526753675467556756675767586759676067616762676367646765676667676768676967706771677267736774677567766777677867796780678167826783678467856786678767886789679067916792679367946795679667976798679968006801680268036804680568066807680868096810681168126813681468156816681768186819682068216822682368246825682668276828682968306831683268336834683568366837683868396840684168426843684468456846684768486849685068516852685368546855685668576858685968606861686268636864686568666867686868696870687168726873687468756876687768786879688068816882688368846885688668876888688968906891689268936894689568966897689868996900690169026903690469056906690769086909691069116912691369146915691669176918691969206921692269236924692569266927692869296930693169326933693469356936693769386939694069416942694369446945694669476948694969506951695269536954695569566957695869596960696169626963696469656966696769686969697069716972697369746975697669776978697969806981698269836984698569866987698869896990699169926993699469956996699769986999700070017002700370047005700670077008700970107011701270137014701570167017701870197020702170227023702470257026702770287029703070317032703370347035703670377038703970407041704270437044704570467047704870497050705170527053705470557056705770587059706070617062706370647065706670677068706970707071707270737074707570767077707870797080708170827083708470857086708770887089709070917092709370947095709670977098709971007101710271037104710571067107710871097110711171127113711471157116711771187119712071217122712371247125712671277128712971307131713271337134713571367137713871397140714171427143714471457146714771487149715071517152715371547155715671577158715971607161716271637164716571667167716871697170717171727173717471757176717771787179718071817182718371847185718671877188718971907191719271937194719571967197719871997200720172027203720472057206720772087209721072117212721372147215721672177218721972207221722272237224722572267227722872297230723172327233723472357236723772387239724072417242724372447245724672477248724972507251725272537254725572567257725872597260726172627263726472657266726772687269727072717272727372747275727672777278727972807281728272837284728572867287728872897290729172927293729472957296729772987299730073017302730373047305730673077308730973107311731273137314731573167317731873197320732173227323732473257326732773287329733073317332733373347335733673377338733973407341734273437344734573467347734873497350735173527353735473557356735773587359736073617362736373647365736673677368736973707371737273737374737573767377737873797380738173827383738473857386738773887389739073917392739373947395739673977398739974007401740274037404740574067407740874097410741174127413741474157416741774187419742074217422742374247425742674277428742974307431743274337434743574367437743874397440744174427443744474457446744774487449745074517452745374547455745674577458745974607461746274637464746574667467746874697470747174727473747474757476747774787479748074817482748374847485748674877488748974907491749274937494749574967497749874997500750175027503750475057506750775087509751075117512751375147515751675177518751975207521752275237524752575267527752875297530753175327533753475357536753775387539754075417542754375447545754675477548754975507551755275537554755575567557755875597560756175627563756475657566756775687569757075717572757375747575757675777578757975807581758275837584758575867587758875897590759175927593759475957596759775987599760076017602760376047605760676077608760976107611761276137614761576167617761876197620762176227623762476257626762776287629763076317632763376347635763676377638763976407641764276437644764576467647764876497650765176527653765476557656765776587659766076617662766376647665766676677668766976707671767276737674767576767677767876797680768176827683768476857686768776887689769076917692769376947695769676977698769977007701770277037704770577067707770877097710771177127713771477157716771777187719772077217722772377247725772677277728772977307731773277337734773577367737773877397740774177427743774477457746774777487749775077517752775377547755775677577758775977607761776277637764776577667767776877697770777177727773777477757776777777787779778077817782778377847785778677877788778977907791779277937794779577967797779877997800780178027803780478057806780778087809781078117812781378147815781678177818781978207821782278237824782578267827782878297830783178327833783478357836783778387839784078417842784378447845784678477848784978507851785278537854785578567857785878597860786178627863786478657866786778687869787078717872787378747875787678777878787978807881788278837884788578867887788878897890789178927893789478957896789778987899790079017902790379047905790679077908790979107911791279137914791579167917791879197920792179227923792479257926792779287929793079317932793379347935793679377938793979407941794279437944794579467947794879497950795179527953795479557956795779587959796079617962796379647965796679677968796979707971797279737974797579767977797879797980798179827983798479857986798779887989799079917992799379947995799679977998799980008001800280038004800580068007800880098010801180128013801480158016801780188019802080218022802380248025802680278028802980308031803280338034803580368037803880398040804180428043804480458046804780488049805080518052805380548055805680578058805980608061806280638064806580668067806880698070807180728073807480758076807780788079808080818082808380848085808680878088808980908091809280938094809580968097809880998100810181028103810481058106810781088109811081118112811381148115811681178118811981208121812281238124812581268127812881298130813181328133813481358136813781388139814081418142814381448145814681478148814981508151815281538154815581568157815881598160816181628163816481658166816781688169817081718172817381748175817681778178817981808181818281838184818581868187818881898190819181928193819481958196819781988199820082018202820382048205820682078208820982108211821282138214821582168217821882198220822182228223822482258226822782288229823082318232823382348235823682378238823982408241824282438244824582468247824882498250825182528253825482558256825782588259826082618262826382648265826682678268826982708271827282738274827582768277827882798280828182828283828482858286828782888289829082918292829382948295829682978298829983008301830283038304830583068307830883098310831183128313831483158316831783188319832083218322832383248325832683278328832983308331833283338334833583368337833883398340834183428343834483458346834783488349835083518352835383548355835683578358835983608361836283638364836583668367836883698370837183728373837483758376837783788379838083818382838383848385838683878388838983908391839283938394839583968397839883998400840184028403840484058406840784088409841084118412841384148415841684178418841984208421842284238424842584268427842884298430843184328433843484358436843784388439844084418442844384448445844684478448844984508451845284538454845584568457845884598460846184628463846484658466846784688469847084718472847384748475847684778478847984808481848284838484848584868487848884898490849184928493849484958496849784988499850085018502850385048505850685078508850985108511851285138514851585168517851885198520852185228523852485258526852785288529853085318532853385348535853685378538853985408541854285438544854585468547854885498550855185528553855485558556855785588559856085618562856385648565856685678568856985708571857285738574857585768577857885798580858185828583858485858586858785888589859085918592859385948595859685978598859986008601860286038604860586068607860886098610861186128613861486158616861786188619862086218622862386248625862686278628862986308631863286338634863586368637863886398640864186428643864486458646864786488649865086518652865386548655865686578658865986608661866286638664866586668667866886698670867186728673867486758676867786788679868086818682868386848685868686878688868986908691869286938694869586968697869886998700870187028703870487058706870787088709871087118712871387148715871687178718871987208721872287238724872587268727872887298730873187328733873487358736873787388739874087418742874387448745874687478748874987508751875287538754875587568757875887598760876187628763876487658766876787688769877087718772877387748775877687778778877987808781878287838784878587868787878887898790879187928793879487958796879787988799880088018802880388048805880688078808880988108811881288138814881588168817881888198820882188228823882488258826882788288829883088318832883388348835883688378838883988408841884288438844884588468847884888498850885188528853885488558856885788588859886088618862886388648865886688678868886988708871887288738874887588768877887888798880888188828883888488858886888788888889889088918892889388948895889688978898889989008901890289038904890589068907890889098910891189128913891489158916891789188919892089218922892389248925892689278928892989308931893289338934893589368937893889398940894189428943894489458946894789488949895089518952895389548955895689578958895989608961896289638964896589668967896889698970897189728973897489758976897789788979898089818982898389848985898689878988898989908991899289938994899589968997899889999000900190029003900490059006900790089009901090119012901390149015901690179018901990209021902290239024902590269027902890299030903190329033903490359036903790389039904090419042904390449045904690479048904990509051905290539054905590569057905890599060906190629063906490659066906790689069907090719072907390749075907690779078907990809081908290839084908590869087908890899090909190929093909490959096909790989099910091019102910391049105910691079108910991109111911291139114911591169117911891199120912191229123912491259126912791289129913091319132913391349135913691379138913991409141914291439144914591469147914891499150915191529153915491559156915791589159916091619162916391649165916691679168916991709171917291739174917591769177917891799180918191829183918491859186918791889189919091919192919391949195919691979198919992009201920292039204920592069207920892099210921192129213921492159216921792189219922092219222922392249225922692279228922992309231923292339234923592369237923892399240924192429243924492459246924792489249925092519252925392549255925692579258925992609261926292639264926592669267926892699270927192729273927492759276927792789279928092819282928392849285928692879288928992909291929292939294929592969297929892999300930193029303930493059306930793089309931093119312931393149315931693179318931993209321932293239324932593269327932893299330933193329333933493359336933793389339934093419342934393449345934693479348934993509351935293539354935593569357935893599360936193629363936493659366936793689369937093719372937393749375937693779378937993809381938293839384938593869387938893899390939193929393939493959396939793989399940094019402940394049405940694079408940994109411941294139414941594169417941894199420942194229423942494259426942794289429943094319432943394349435943694379438943994409441944294439444944594469447944894499450945194529453945494559456945794589459946094619462946394649465946694679468946994709471947294739474947594769477947894799480948194829483948494859486948794889489949094919492949394949495949694979498949995009501950295039504950595069507950895099510951195129513951495159516951795189519952095219522952395249525952695279528952995309531953295339534953595369537953895399540954195429543954495459546954795489549955095519552955395549555955695579558955995609561956295639564956595669567956895699570957195729573957495759576957795789579958095819582958395849585958695879588958995909591959295939594959595969597959895999600960196029603960496059606960796089609961096119612961396149615961696179618961996209621962296239624962596269627962896299630963196329633963496359636963796389639964096419642964396449645964696479648964996509651965296539654965596569657965896599660966196629663966496659666966796689669967096719672967396749675967696779678967996809681968296839684968596869687968896899690969196929693969496959696969796989699970097019702970397049705970697079708970997109711971297139714971597169717971897199720972197229723972497259726972797289729973097319732973397349735973697379738973997409741974297439744974597469747974897499750975197529753975497559756975797589759976097619762976397649765976697679768976997709771977297739774977597769777977897799780978197829783978497859786978797889789979097919792979397949795979697979798979998009801980298039804980598069807980898099810981198129813981498159816981798189819982098219822982398249825982698279828982998309831983298339834983598369837983898399840984198429843984498459846984798489849985098519852985398549855985698579858985998609861986298639864986598669867986898699870987198729873987498759876987798789879988098819882988398849885988698879888988998909891989298939894989598969897989898999900990199029903990499059906990799089909991099119912991399149915991699179918991999209921992299239924992599269927992899299930993199329933993499359936993799389939994099419942994399449945994699479948994999509951995299539954995599569957995899599960996199629963996499659966996799689969997099719972997399749975997699779978997999809981998299839984998599869987998899899990999199929993999499959996999799989999100001000110002100031000410005100061000710008100091001010011100121001310014100151001610017100181001910020100211002210023100241002510026100271002810029100301003110032100331003410035100361003710038100391004010041100421004310044100451004610047100481004910050100511005210053100541005510056100571005810059100601006110062100631006410065100661006710068100691007010071100721007310074100751007610077100781007910080100811008210083100841008510086100871008810089100901009110092100931009410095100961009710098100991010010101101021010310104101051010610107101081010910110101111011210113101141011510116101171011810119101201012110122101231012410125101261012710128101291013010131101321013310134101351013610137101381013910140101411014210143101441014510146101471014810149101501015110152101531015410155101561015710158101591016010161101621016310164101651016610167101681016910170101711017210173101741017510176101771017810179101801018110182101831018410185101861018710188101891019010191101921019310194101951019610197101981019910200102011020210203102041020510206102071020810209102101021110212102131021410215102161021710218102191022010221102221022310224102251022610227102281022910230102311023210233102341023510236102371023810239102401024110242102431024410245102461024710248102491025010251102521025310254102551025610257102581025910260102611026210263102641026510266102671026810269102701027110272102731027410275102761027710278102791028010281102821028310284102851028610287102881028910290102911029210293102941029510296102971029810299103001030110302103031030410305103061030710308103091031010311103121031310314103151031610317103181031910320103211032210323103241032510326103271032810329103301033110332103331033410335103361033710338103391034010341103421034310344103451034610347103481034910350103511035210353103541035510356103571035810359103601036110362103631036410365103661036710368103691037010371103721037310374103751037610377103781037910380103811038210383103841038510386103871038810389103901039110392103931039410395103961039710398103991040010401104021040310404104051040610407104081040910410104111041210413104141041510416104171041810419104201042110422104231042410425104261042710428104291043010431104321043310434104351043610437104381043910440104411044210443104441044510446104471044810449104501045110452104531045410455104561045710458104591046010461104621046310464104651046610467104681046910470104711047210473104741047510476104771047810479104801048110482104831048410485104861048710488104891049010491104921049310494104951049610497104981049910500105011050210503105041050510506105071050810509105101051110512105131051410515105161051710518105191052010521105221052310524105251052610527105281052910530105311053210533105341053510536105371053810539105401054110542105431054410545105461054710548105491055010551105521055310554105551055610557105581055910560105611056210563105641056510566105671056810569105701057110572105731057410575105761057710578105791058010581105821058310584105851058610587105881058910590105911059210593105941059510596105971059810599106001060110602106031060410605106061060710608106091061010611106121061310614106151061610617106181061910620106211062210623106241062510626106271062810629106301063110632106331063410635106361063710638106391064010641106421064310644106451064610647106481064910650106511065210653106541065510656106571065810659106601066110662106631066410665106661066710668106691067010671106721067310674106751067610677106781067910680106811068210683106841068510686106871068810689106901069110692106931069410695106961069710698106991070010701107021070310704107051070610707107081070910710107111071210713107141071510716107171071810719107201072110722107231072410725107261072710728107291073010731107321073310734107351073610737107381073910740107411074210743107441074510746107471074810749107501075110752107531075410755107561075710758107591076010761107621076310764107651076610767107681076910770107711077210773107741077510776107771077810779107801078110782107831078410785107861078710788107891079010791107921079310794107951079610797107981079910800108011080210803108041080510806108071080810809108101081110812108131081410815108161081710818108191082010821108221082310824108251082610827108281082910830108311083210833108341083510836108371083810839108401084110842108431084410845108461084710848108491085010851108521085310854108551085610857108581085910860108611086210863108641086510866108671086810869108701087110872108731087410875108761087710878108791088010881108821088310884108851088610887108881088910890108911089210893108941089510896108971089810899109001090110902109031090410905109061090710908109091091010911109121091310914109151091610917109181091910920109211092210923109241092510926109271092810929109301093110932109331093410935109361093710938109391094010941109421094310944109451094610947109481094910950109511095210953109541095510956109571095810959109601096110962109631096410965109661096710968109691097010971109721097310974109751097610977109781097910980109811098210983109841098510986109871098810989109901099110992109931099410995109961099710998109991100011001110021100311004110051100611007110081100911010110111101211013110141101511016110171101811019110201102111022110231102411025110261102711028110291103011031110321103311034110351103611037110381103911040110411104211043110441104511046110471104811049110501105111052110531105411055110561105711058110591106011061110621106311064110651106611067110681106911070110711107211073110741107511076110771107811079110801108111082110831108411085110861108711088110891109011091110921109311094110951109611097110981109911100111011110211103111041110511106111071110811109111101111111112111131111411115111161111711118111191112011121111221112311124111251112611127111281112911130111311113211133111341113511136111371113811139111401114111142111431114411145111461114711148111491115011151111521115311154111551115611157111581115911160111611116211163111641116511166111671116811169111701117111172111731117411175111761117711178111791118011181111821118311184111851118611187111881118911190111911119211193111941119511196111971119811199112001120111202112031120411205112061120711208112091121011211112121121311214112151121611217112181121911220112211122211223112241122511226112271122811229112301123111232112331123411235112361123711238112391124011241112421124311244112451124611247112481124911250112511125211253112541125511256112571125811259112601126111262112631126411265112661126711268112691127011271112721127311274112751127611277112781127911280112811128211283112841128511286112871128811289112901129111292112931129411295112961129711298112991130011301113021130311304113051130611307113081130911310113111131211313113141131511316113171131811319113201132111322113231132411325113261132711328113291133011331113321133311334113351133611337113381133911340113411134211343113441134511346113471134811349113501135111352113531135411355113561135711358113591136011361113621136311364113651136611367113681136911370113711137211373113741137511376113771137811379113801138111382113831138411385113861138711388113891139011391113921139311394113951139611397113981139911400114011140211403114041140511406114071140811409114101141111412114131141411415114161141711418114191142011421114221142311424114251142611427114281142911430114311143211433114341143511436114371143811439114401144111442114431144411445114461144711448114491145011451114521145311454114551145611457114581145911460114611146211463114641146511466114671146811469114701147111472114731147411475114761147711478114791148011481114821148311484114851148611487114881148911490114911149211493114941149511496114971149811499115001150111502115031150411505115061150711508115091151011511115121151311514115151151611517115181151911520115211152211523115241152511526115271152811529115301153111532115331153411535115361153711538115391154011541115421154311544115451154611547115481154911550115511155211553115541155511556115571155811559115601156111562115631156411565115661156711568115691157011571115721157311574115751157611577115781157911580115811158211583115841158511586115871158811589115901159111592115931159411595115961159711598115991160011601116021160311604116051160611607116081160911610116111161211613116141161511616116171161811619116201162111622116231162411625116261162711628116291163011631116321163311634116351163611637116381163911640116411164211643116441164511646116471164811649116501165111652116531165411655116561165711658116591166011661116621166311664116651166611667116681166911670116711167211673116741167511676116771167811679116801168111682116831168411685116861168711688116891169011691116921169311694116951169611697116981169911700117011170211703117041170511706117071170811709117101171111712117131171411715117161171711718117191172011721117221172311724117251172611727117281172911730117311173211733117341173511736117371173811739117401174111742117431174411745117461174711748117491175011751117521175311754117551175611757117581175911760117611176211763117641176511766117671176811769117701177111772117731177411775117761177711778117791178011781117821178311784117851178611787117881178911790117911179211793117941179511796117971179811799118001180111802118031180411805118061180711808118091181011811118121181311814118151181611817118181181911820118211182211823118241182511826118271182811829118301183111832118331183411835118361183711838118391184011841118421184311844118451184611847118481184911850118511185211853118541185511856118571185811859118601186111862118631186411865118661186711868118691187011871118721187311874118751187611877118781187911880118811188211883118841188511886118871188811889118901189111892118931189411895118961189711898118991190011901119021190311904119051190611907119081190911910119111191211913119141191511916119171191811919119201192111922119231192411925119261192711928119291193011931119321193311934119351193611937119381193911940119411194211943119441194511946119471194811949119501195111952119531195411955119561195711958119591196011961119621196311964119651196611967119681196911970119711197211973119741197511976119771197811979119801198111982119831198411985119861198711988119891199011991119921199311994119951199611997119981199912000120011200212003120041200512006120071200812009120101201112012120131201412015120161201712018120191202012021120221202312024120251202612027120281202912030120311203212033120341203512036120371203812039120401204112042120431204412045120461204712048120491205012051120521205312054120551205612057120581205912060120611206212063120641206512066120671206812069120701207112072120731207412075120761207712078120791208012081120821208312084120851208612087120881208912090120911209212093120941209512096120971209812099121001210112102121031210412105121061210712108121091211012111121121211312114121151211612117121181211912120121211212212123121241212512126121271212812129121301213112132121331213412135121361213712138121391214012141121421214312144121451214612147121481214912150121511215212153121541215512156121571215812159121601216112162121631216412165121661216712168121691217012171121721217312174121751217612177121781217912180121811218212183121841218512186121871218812189121901219112192121931219412195121961219712198121991220012201122021220312204122051220612207122081220912210122111221212213122141221512216122171221812219122201222112222122231222412225122261222712228122291223012231122321223312234122351223612237122381223912240122411224212243122441224512246122471224812249122501225112252122531225412255122561225712258122591226012261122621226312264122651226612267122681226912270122711227212273122741227512276122771227812279122801228112282122831228412285122861228712288122891229012291122921229312294122951229612297122981229912300123011230212303123041230512306123071230812309123101231112312123131231412315123161231712318123191232012321123221232312324123251232612327123281232912330123311233212333123341233512336123371233812339123401234112342123431234412345123461234712348123491235012351123521235312354123551235612357123581235912360123611236212363123641236512366123671236812369123701237112372123731237412375123761237712378123791238012381123821238312384123851238612387123881238912390123911239212393123941239512396123971239812399124001240112402124031240412405124061240712408124091241012411124121241312414124151241612417124181241912420124211242212423124241242512426124271242812429124301243112432124331243412435124361243712438124391244012441124421244312444124451244612447124481244912450124511245212453124541245512456124571245812459124601246112462124631246412465124661246712468124691247012471124721247312474124751247612477124781247912480124811248212483124841248512486124871248812489124901249112492124931249412495124961249712498124991250012501125021250312504125051250612507125081250912510125111251212513125141251512516125171251812519125201252112522125231252412525125261252712528125291253012531125321253312534125351253612537125381253912540125411254212543125441254512546125471254812549125501255112552125531255412555125561255712558125591256012561125621256312564125651256612567125681256912570125711257212573125741257512576125771257812579125801258112582125831258412585125861258712588125891259012591125921259312594125951259612597125981259912600126011260212603126041260512606126071260812609126101261112612126131261412615126161261712618126191262012621126221262312624126251262612627126281262912630126311263212633126341263512636126371263812639126401264112642126431264412645126461264712648126491265012651126521265312654126551265612657126581265912660126611266212663126641266512666126671266812669126701267112672126731267412675126761267712678126791268012681126821268312684126851268612687126881268912690126911269212693126941269512696126971269812699127001270112702127031270412705127061270712708127091271012711127121271312714127151271612717127181271912720127211272212723127241272512726127271272812729127301273112732127331273412735127361273712738127391274012741127421274312744127451274612747127481274912750127511275212753127541275512756127571275812759127601276112762127631276412765127661276712768127691277012771127721277312774127751277612777127781277912780127811278212783127841278512786127871278812789127901279112792127931279412795127961279712798127991280012801128021280312804128051280612807128081280912810128111281212813128141281512816128171281812819128201282112822128231282412825128261282712828128291283012831128321283312834128351283612837128381283912840128411284212843128441284512846128471284812849128501285112852128531285412855128561285712858128591286012861128621286312864128651286612867128681286912870128711287212873128741287512876128771287812879128801288112882128831288412885128861288712888128891289012891128921289312894128951289612897128981289912900129011290212903129041290512906129071290812909129101291112912129131291412915129161291712918129191292012921129221292312924129251292612927129281292912930129311293212933129341293512936129371293812939129401294112942129431294412945129461294712948129491295012951129521295312954129551295612957129581295912960129611296212963129641296512966129671296812969129701297112972129731297412975129761297712978129791298012981129821298312984129851298612987129881298912990129911299212993129941299512996129971299812999130001300113002130031300413005130061300713008130091301013011130121301313014130151301613017130181301913020130211302213023130241302513026130271302813029130301303113032130331303413035130361303713038130391304013041130421304313044130451304613047130481304913050130511305213053130541305513056130571305813059130601306113062130631306413065130661306713068130691307013071130721307313074130751307613077130781307913080130811308213083130841308513086130871308813089130901309113092130931309413095130961309713098130991310013101131021310313104131051310613107131081310913110131111311213113131141311513116131171311813119131201312113122131231312413125131261312713128131291313013131131321313313134131351313613137131381313913140131411314213143131441314513146131471314813149131501315113152131531315413155131561315713158131591316013161131621316313164131651316613167131681316913170131711317213173131741317513176131771317813179131801318113182131831318413185131861318713188131891319013191131921319313194131951319613197131981319913200132011320213203132041320513206132071320813209132101321113212132131321413215132161321713218132191322013221132221322313224132251322613227132281322913230132311323213233132341323513236132371323813239132401324113242132431324413245132461324713248132491325013251132521325313254132551325613257132581325913260132611326213263132641326513266132671326813269132701327113272132731327413275132761327713278132791328013281132821328313284132851328613287132881328913290132911329213293132941329513296132971329813299133001330113302133031330413305133061330713308133091331013311133121331313314133151331613317133181331913320133211332213323133241332513326133271332813329133301333113332133331333413335133361333713338133391334013341133421334313344133451334613347133481334913350133511335213353133541335513356133571335813359133601336113362133631336413365133661336713368133691337013371133721337313374133751337613377133781337913380133811338213383133841338513386133871338813389133901339113392133931339413395133961339713398133991340013401134021340313404134051340613407134081340913410134111341213413134141341513416134171341813419134201342113422134231342413425134261342713428134291343013431134321343313434134351343613437134381343913440134411344213443134441344513446134471344813449134501345113452134531345413455134561345713458134591346013461134621346313464134651346613467134681346913470134711347213473134741347513476134771347813479134801348113482134831348413485134861348713488134891349013491134921349313494134951349613497134981349913500135011350213503135041350513506135071350813509135101351113512135131351413515135161351713518135191352013521135221352313524135251352613527135281352913530135311353213533135341353513536135371353813539135401354113542135431354413545135461354713548135491355013551135521355313554135551355613557
  1. /**
  2. * pugixml parser - version 1.15
  3. * --------------------------------------------------------
  4. * Report bugs and download new versions at https://pugixml.org/
  5. *
  6. * SPDX-FileCopyrightText: Copyright (C) 2006-2026, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com)
  7. * SPDX-License-Identifier: MIT
  8. *
  9. * See LICENSE.md or notice at the end of this file.
  10. */
  11. #ifndef SOURCE_PUGIXML_CPP
  12. #define SOURCE_PUGIXML_CPP
  13. #include "pugixml.hpp"
  14. #include <stdlib.h>
  15. #include <stdio.h>
  16. #include <string.h>
  17. #include <assert.h>
  18. #include <limits.h>
  19. #ifdef PUGIXML_WCHAR_MODE
  20. # include <wchar.h>
  21. #endif
  22. #ifndef PUGIXML_NO_XPATH
  23. # include <math.h>
  24. # include <float.h>
  25. #endif
  26. #ifndef PUGIXML_NO_STL
  27. # include <istream>
  28. # include <ostream>
  29. # include <string>
  30. #endif
  31. // For placement new
  32. #include <new>
  33. // For load_file
  34. #if defined(__linux__) || defined(__APPLE__)
  35. #include <sys/stat.h>
  36. #endif
  37. #ifdef _MSC_VER
  38. # pragma warning(push)
  39. # pragma warning(disable: 4127) // conditional expression is constant
  40. # pragma warning(disable: 4324) // structure was padded due to __declspec(align())
  41. # pragma warning(disable: 4702) // unreachable code
  42. # pragma warning(disable: 4996) // this function or variable may be unsafe
  43. #endif
  44. #if defined(__clang__)
  45. # pragma clang diagnostic push
  46. # pragma clang diagnostic ignored "-Wzero-as-null-pointer-constant" // NULL as null pointer constant
  47. #endif
  48. #if defined(_MSC_VER) && defined(__c2__)
  49. # pragma clang diagnostic push
  50. # pragma clang diagnostic ignored "-Wdeprecated" // this function or variable may be unsafe
  51. #endif
  52. #ifdef __INTEL_COMPILER
  53. # pragma warning(disable: 177) // function was declared but never referenced
  54. # pragma warning(disable: 279) // controlling expression is constant
  55. # pragma warning(disable: 1478 1786) // function was declared "deprecated"
  56. # pragma warning(disable: 1684) // conversion from pointer to same-sized integral type
  57. #endif
  58. #if defined(__BORLANDC__) && defined(PUGIXML_HEADER_ONLY)
  59. # pragma warn -8080 // symbol is declared but never used; disabling this inside push/pop bracket does not make the warning go away
  60. #endif
  61. #ifdef __BORLANDC__
  62. # pragma option push
  63. # pragma warn -8008 // condition is always false
  64. # pragma warn -8066 // unreachable code
  65. #endif
  66. #ifdef __SNC__
  67. // Using diag_push/diag_pop does not disable the warnings inside templates due to a compiler bug
  68. # pragma diag_suppress=178 // function was declared but never referenced
  69. # pragma diag_suppress=237 // controlling expression is constant
  70. #endif
  71. #ifdef __TI_COMPILER_VERSION__
  72. # pragma diag_suppress 179 // function was declared but never referenced
  73. #endif
  74. // Inlining controls
  75. #if defined(_MSC_VER) && _MSC_VER >= 1300
  76. # define PUGI_IMPL_NO_INLINE __declspec(noinline)
  77. #elif defined(__GNUC__)
  78. # define PUGI_IMPL_NO_INLINE __attribute__((noinline))
  79. #else
  80. # define PUGI_IMPL_NO_INLINE
  81. #endif
  82. // Branch weight controls
  83. #if defined(__GNUC__) && !defined(__c2__)
  84. # define PUGI_IMPL_UNLIKELY(cond) __builtin_expect(cond, 0)
  85. #else
  86. # define PUGI_IMPL_UNLIKELY(cond) (cond)
  87. #endif
  88. // Simple static assertion
  89. #define PUGI_IMPL_STATIC_ASSERT(cond) { static const char condition_failed[(cond) ? 1 : -1] = {0}; (void)condition_failed[0]; }
  90. // Digital Mars C++ bug workaround for passing char loaded from memory via stack
  91. #ifdef __DMC__
  92. # define PUGI_IMPL_DMC_VOLATILE volatile
  93. #else
  94. # define PUGI_IMPL_DMC_VOLATILE
  95. #endif
  96. // Integer sanitizer workaround; we only apply this for clang since gcc8 has no_sanitize but not unsigned-integer-overflow and produces "attribute directive ignored" warnings
  97. #if defined(__clang__) && defined(__has_attribute)
  98. # if __has_attribute(no_sanitize)
  99. # define PUGI_IMPL_UNSIGNED_OVERFLOW __attribute__((no_sanitize("unsigned-integer-overflow")))
  100. # else
  101. # define PUGI_IMPL_UNSIGNED_OVERFLOW
  102. # endif
  103. #else
  104. # define PUGI_IMPL_UNSIGNED_OVERFLOW
  105. #endif
  106. // Borland C++ bug workaround for not defining ::memcpy depending on header include order (can't always use std::memcpy because some compilers don't have it at all)
  107. #if defined(__BORLANDC__) && !defined(__MEM_H_USING_LIST)
  108. using std::memcpy;
  109. using std::memmove;
  110. using std::memset;
  111. #endif
  112. // Old versions of GCC do not define ::malloc and ::free depending on header include order
  113. #if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 4))
  114. using std::malloc;
  115. using std::free;
  116. #endif
  117. // Some MinGW/GCC versions have headers that erroneously omit LLONG_MIN/LLONG_MAX/ULLONG_MAX definitions from limits.h in some configurations
  118. #if defined(PUGIXML_HAS_LONG_LONG) && defined(__GNUC__) && !defined(LLONG_MAX) && !defined(LLONG_MIN) && !defined(ULLONG_MAX)
  119. # define LLONG_MIN (-LLONG_MAX - 1LL)
  120. # define LLONG_MAX __LONG_LONG_MAX__
  121. # define ULLONG_MAX (LLONG_MAX * 2ULL + 1ULL)
  122. #endif
  123. // In some environments MSVC is a compiler but the CRT lacks certain MSVC-specific features
  124. #if defined(_MSC_VER) && !defined(__S3E__) && !defined(_WIN32_WCE)
  125. # define PUGI_IMPL_MSVC_CRT_VERSION _MSC_VER
  126. #elif defined(_WIN32_WCE)
  127. # define PUGI_IMPL_MSVC_CRT_VERSION 1310 // MSVC7.1
  128. #endif
  129. // Not all platforms have snprintf; we define a wrapper that uses snprintf if possible. This only works with buffers with a known size.
  130. #if __cplusplus >= 201103
  131. # define PUGI_IMPL_SNPRINTF(buf, ...) snprintf(buf, sizeof(buf), __VA_ARGS__)
  132. #elif defined(PUGI_IMPL_MSVC_CRT_VERSION) && PUGI_IMPL_MSVC_CRT_VERSION >= 1400
  133. # define PUGI_IMPL_SNPRINTF(buf, ...) _snprintf_s(buf, _countof(buf), _TRUNCATE, __VA_ARGS__)
  134. #elif defined(__APPLE__) && __clang_major__ >= 14 // Xcode 14 marks sprintf as deprecated while still using C++98 by default
  135. # define PUGI_IMPL_SNPRINTF(buf, fmt, arg1, arg2) snprintf(buf, sizeof(buf), fmt, arg1, arg2)
  136. #else
  137. # define PUGI_IMPL_SNPRINTF sprintf
  138. #endif
  139. // We put implementation details into an anonymous namespace in source mode, but have to keep it in non-anonymous namespace in header-only mode to prevent binary bloat.
  140. #ifdef PUGIXML_HEADER_ONLY
  141. # define PUGI_IMPL_NS_BEGIN namespace pugi { namespace impl {
  142. # define PUGI_IMPL_NS_END } }
  143. # define PUGI_IMPL_FN inline
  144. # define PUGI_IMPL_FN_NO_INLINE inline
  145. #else
  146. # if defined(_MSC_VER) && _MSC_VER < 1300 // MSVC6 seems to have an amusing bug with anonymous namespaces inside namespaces
  147. # define PUGI_IMPL_NS_BEGIN namespace pugi { namespace impl {
  148. # define PUGI_IMPL_NS_END } }
  149. # else
  150. # define PUGI_IMPL_NS_BEGIN namespace pugi { namespace impl { namespace {
  151. # define PUGI_IMPL_NS_END } } }
  152. # endif
  153. # define PUGI_IMPL_FN
  154. # define PUGI_IMPL_FN_NO_INLINE PUGI_IMPL_NO_INLINE
  155. #endif
  156. // uintptr_t
  157. #if (defined(_MSC_VER) && _MSC_VER < 1600) || (defined(__BORLANDC__) && __BORLANDC__ < 0x561)
  158. namespace pugi
  159. {
  160. # ifndef _UINTPTR_T_DEFINED
  161. typedef size_t uintptr_t;
  162. # endif
  163. typedef unsigned __int8 uint8_t;
  164. typedef unsigned __int16 uint16_t;
  165. typedef unsigned __int32 uint32_t;
  166. }
  167. #else
  168. # include <stdint.h>
  169. #endif
  170. // Memory allocation
  171. PUGI_IMPL_NS_BEGIN
  172. PUGI_IMPL_FN void* default_allocate(size_t size)
  173. {
  174. return malloc(size);
  175. }
  176. PUGI_IMPL_FN void default_deallocate(void* ptr)
  177. {
  178. free(ptr);
  179. }
  180. template <typename T>
  181. struct xml_memory_management_function_storage
  182. {
  183. static allocation_function allocate;
  184. static deallocation_function deallocate;
  185. };
  186. // Global allocation functions are stored in class statics so that in header mode linker deduplicates them
  187. // Without a template<> we'll get multiple definitions of the same static
  188. template <typename T> allocation_function xml_memory_management_function_storage<T>::allocate = default_allocate;
  189. template <typename T> deallocation_function xml_memory_management_function_storage<T>::deallocate = default_deallocate;
  190. typedef xml_memory_management_function_storage<int> xml_memory;
  191. PUGI_IMPL_NS_END
  192. // String utilities
  193. PUGI_IMPL_NS_BEGIN
  194. // Get string length
  195. PUGI_IMPL_FN size_t strlength(const char_t* s)
  196. {
  197. assert(s);
  198. #ifdef PUGIXML_WCHAR_MODE
  199. return wcslen(s);
  200. #else
  201. return strlen(s);
  202. #endif
  203. }
  204. // Compare two strings
  205. PUGI_IMPL_FN bool strequal(const char_t* src, const char_t* dst)
  206. {
  207. assert(src && dst);
  208. #ifdef PUGIXML_WCHAR_MODE
  209. return wcscmp(src, dst) == 0;
  210. #else
  211. return strcmp(src, dst) == 0;
  212. #endif
  213. }
  214. #ifdef PUGIXML_HAS_STRING_VIEW
  215. // Check if the null-terminated dst string is equal to the entire contents of srcview
  216. PUGI_IMPL_FN bool stringview_equal(string_view_t srcview, const char_t* dst)
  217. {
  218. // std::basic_string_view::compare(const char*) has the right behavior, but it performs an
  219. // extra traversal of dst to compute its length.
  220. assert(dst);
  221. const char_t* src = srcview.data();
  222. size_t srclen = srcview.size();
  223. while (srclen && *dst && *src == *dst)
  224. {
  225. --srclen; ++dst; ++src;
  226. }
  227. return srclen == 0 && *dst == 0;
  228. }
  229. #endif
  230. // Compare lhs with [rhs_begin, rhs_end)
  231. PUGI_IMPL_FN bool strequalrange(const char_t* lhs, const char_t* rhs, size_t count)
  232. {
  233. for (size_t i = 0; i < count; ++i)
  234. if (lhs[i] != rhs[i])
  235. return false;
  236. return lhs[count] == 0;
  237. }
  238. // Get length of wide string, even if CRT lacks wide character support
  239. PUGI_IMPL_FN size_t strlength_wide(const wchar_t* s)
  240. {
  241. assert(s);
  242. #ifdef PUGIXML_WCHAR_MODE
  243. return wcslen(s);
  244. #else
  245. const wchar_t* end = s;
  246. while (*end) end++;
  247. return static_cast<size_t>(end - s);
  248. #endif
  249. }
  250. PUGI_IMPL_NS_END
  251. // auto_ptr-like object for exception recovery
  252. PUGI_IMPL_NS_BEGIN
  253. template <typename T> struct auto_deleter
  254. {
  255. typedef void (*D)(T*);
  256. T* data;
  257. D deleter;
  258. auto_deleter(T* data_, D deleter_): data(data_), deleter(deleter_)
  259. {
  260. }
  261. ~auto_deleter()
  262. {
  263. if (data) deleter(data);
  264. }
  265. T* release()
  266. {
  267. T* result = data;
  268. data = NULL;
  269. return result;
  270. }
  271. };
  272. PUGI_IMPL_NS_END
  273. #ifdef PUGIXML_COMPACT
  274. PUGI_IMPL_NS_BEGIN
  275. class compact_hash_table
  276. {
  277. public:
  278. compact_hash_table(): _items(NULL), _capacity(0), _count(0)
  279. {
  280. }
  281. void clear()
  282. {
  283. if (_items)
  284. {
  285. xml_memory::deallocate(_items);
  286. _items = NULL;
  287. _capacity = 0;
  288. _count = 0;
  289. }
  290. }
  291. void* find(const void* key)
  292. {
  293. if (_capacity == 0) return NULL;
  294. item_t* item = get_item(key);
  295. assert(item);
  296. assert(item->key == key || (item->key == NULL && item->value == NULL));
  297. return item->value;
  298. }
  299. void insert(const void* key, void* value)
  300. {
  301. assert(_capacity != 0 && _count < _capacity - _capacity / 4);
  302. item_t* item = get_item(key);
  303. assert(item);
  304. if (item->key == NULL)
  305. {
  306. _count++;
  307. item->key = key;
  308. }
  309. item->value = value;
  310. }
  311. bool reserve(size_t extra = 16)
  312. {
  313. if (_count + extra >= _capacity - _capacity / 4)
  314. return rehash(_count + extra);
  315. return true;
  316. }
  317. private:
  318. struct item_t
  319. {
  320. const void* key;
  321. void* value;
  322. };
  323. item_t* _items;
  324. size_t _capacity;
  325. size_t _count;
  326. bool rehash(size_t count);
  327. item_t* get_item(const void* key)
  328. {
  329. assert(key);
  330. assert(_capacity > 0);
  331. size_t hashmod = _capacity - 1;
  332. size_t bucket = hash(key) & hashmod;
  333. for (size_t probe = 0; probe <= hashmod; ++probe)
  334. {
  335. item_t& probe_item = _items[bucket];
  336. if (probe_item.key == key || probe_item.key == NULL)
  337. return &probe_item;
  338. // hash collision, quadratic probing
  339. bucket = (bucket + probe + 1) & hashmod;
  340. }
  341. assert(false && "Hash table is full"); // unreachable
  342. return NULL;
  343. }
  344. static PUGI_IMPL_UNSIGNED_OVERFLOW unsigned int hash(const void* key)
  345. {
  346. unsigned int h = static_cast<unsigned int>(reinterpret_cast<uintptr_t>(key) & 0xffffffff);
  347. // MurmurHash3 32-bit finalizer
  348. h ^= h >> 16;
  349. h *= 0x85ebca6bu;
  350. h ^= h >> 13;
  351. h *= 0xc2b2ae35u;
  352. h ^= h >> 16;
  353. return h;
  354. }
  355. };
  356. PUGI_IMPL_FN_NO_INLINE bool compact_hash_table::rehash(size_t count)
  357. {
  358. size_t capacity = 32;
  359. while (count >= capacity - capacity / 4)
  360. capacity *= 2;
  361. compact_hash_table rt;
  362. rt._capacity = capacity;
  363. rt._items = static_cast<item_t*>(xml_memory::allocate(sizeof(item_t) * capacity));
  364. if (!rt._items)
  365. return false;
  366. memset(rt._items, 0, sizeof(item_t) * capacity);
  367. for (size_t i = 0; i < _capacity; ++i)
  368. if (_items[i].key)
  369. rt.insert(_items[i].key, _items[i].value);
  370. if (_items)
  371. xml_memory::deallocate(_items);
  372. _capacity = capacity;
  373. _items = rt._items;
  374. assert(_count == rt._count);
  375. return true;
  376. }
  377. PUGI_IMPL_NS_END
  378. #endif
  379. PUGI_IMPL_NS_BEGIN
  380. #ifdef PUGIXML_COMPACT
  381. static const uintptr_t xml_memory_block_alignment = 4;
  382. #else
  383. static const uintptr_t xml_memory_block_alignment = sizeof(void*);
  384. #endif
  385. // extra metadata bits
  386. static const uintptr_t xml_memory_page_contents_shared_mask = 64;
  387. static const uintptr_t xml_memory_page_name_allocated_mask = 32;
  388. static const uintptr_t xml_memory_page_value_allocated_mask = 16;
  389. static const uintptr_t xml_memory_page_type_mask = 15;
  390. // combined masks for string uniqueness
  391. static const uintptr_t xml_memory_page_name_allocated_or_shared_mask = xml_memory_page_name_allocated_mask | xml_memory_page_contents_shared_mask;
  392. static const uintptr_t xml_memory_page_value_allocated_or_shared_mask = xml_memory_page_value_allocated_mask | xml_memory_page_contents_shared_mask;
  393. #ifdef PUGIXML_COMPACT
  394. #define PUGI_IMPL_GETHEADER_IMPL(object, page, flags) // unused
  395. #define PUGI_IMPL_GETPAGE_IMPL(header) (header).get_page()
  396. #else
  397. #define PUGI_IMPL_GETHEADER_IMPL(object, page, flags) (((reinterpret_cast<char*>(object) - reinterpret_cast<char*>(page)) << 8) | (flags))
  398. // this macro casts pointers through void* to avoid 'cast increases required alignment of target type' warnings
  399. #define PUGI_IMPL_GETPAGE_IMPL(header) static_cast<impl::xml_memory_page*>(const_cast<void*>(static_cast<const void*>(reinterpret_cast<const char*>(&header) - (header >> 8))))
  400. #endif
  401. #define PUGI_IMPL_GETPAGE(n) PUGI_IMPL_GETPAGE_IMPL((n)->header)
  402. #define PUGI_IMPL_NODETYPE(n) static_cast<xml_node_type>((n)->header & impl::xml_memory_page_type_mask)
  403. struct xml_allocator;
  404. struct xml_memory_page
  405. {
  406. static xml_memory_page* construct(void* memory)
  407. {
  408. xml_memory_page* result = static_cast<xml_memory_page*>(memory);
  409. result->allocator = NULL;
  410. result->prev = NULL;
  411. result->next = NULL;
  412. result->busy_size = 0;
  413. result->freed_size = 0;
  414. #ifdef PUGIXML_COMPACT
  415. result->compact_string_base = NULL;
  416. result->compact_shared_parent = NULL;
  417. result->compact_page_marker = NULL;
  418. #endif
  419. return result;
  420. }
  421. xml_allocator* allocator;
  422. xml_memory_page* prev;
  423. xml_memory_page* next;
  424. size_t busy_size;
  425. size_t freed_size;
  426. #ifdef PUGIXML_COMPACT
  427. char_t* compact_string_base;
  428. void* compact_shared_parent;
  429. uint32_t* compact_page_marker;
  430. #endif
  431. };
  432. static const size_t xml_memory_page_size =
  433. #ifdef PUGIXML_MEMORY_PAGE_SIZE
  434. (PUGIXML_MEMORY_PAGE_SIZE)
  435. #else
  436. 32768
  437. #endif
  438. - sizeof(xml_memory_page);
  439. struct xml_memory_string_header
  440. {
  441. uint16_t page_offset; // offset from page->data
  442. uint16_t full_size; // 0 if string occupies whole page
  443. };
  444. struct xml_allocator
  445. {
  446. xml_allocator(xml_memory_page* root): _root(root), _busy_size(root->busy_size)
  447. {
  448. #ifdef PUGIXML_COMPACT
  449. _hash = NULL;
  450. #endif
  451. }
  452. xml_memory_page* allocate_page(size_t data_size)
  453. {
  454. size_t size = sizeof(xml_memory_page) + data_size;
  455. // allocate block with some alignment, leaving memory for worst-case padding
  456. void* memory = xml_memory::allocate(size);
  457. if (!memory) return NULL;
  458. // prepare page structure
  459. xml_memory_page* page = xml_memory_page::construct(memory);
  460. assert(page);
  461. assert(this == _root->allocator);
  462. page->allocator = this;
  463. return page;
  464. }
  465. static void deallocate_page(xml_memory_page* page)
  466. {
  467. xml_memory::deallocate(page);
  468. }
  469. void* allocate_memory_oob(size_t size, xml_memory_page*& out_page);
  470. void* allocate_memory(size_t size, xml_memory_page*& out_page)
  471. {
  472. if (PUGI_IMPL_UNLIKELY(_busy_size + size > xml_memory_page_size))
  473. return allocate_memory_oob(size, out_page);
  474. void* buf = reinterpret_cast<char*>(_root) + sizeof(xml_memory_page) + _busy_size;
  475. _busy_size += size;
  476. out_page = _root;
  477. return buf;
  478. }
  479. #ifdef PUGIXML_COMPACT
  480. void* allocate_object(size_t size, xml_memory_page*& out_page)
  481. {
  482. void* result = allocate_memory(size + sizeof(uint32_t), out_page);
  483. if (!result) return NULL;
  484. // adjust for marker
  485. ptrdiff_t offset = static_cast<char*>(result) - reinterpret_cast<char*>(out_page->compact_page_marker);
  486. if (PUGI_IMPL_UNLIKELY(static_cast<uintptr_t>(offset) >= 256 * xml_memory_block_alignment))
  487. {
  488. // insert new marker
  489. uint32_t* marker = static_cast<uint32_t*>(result);
  490. *marker = static_cast<uint32_t>(reinterpret_cast<char*>(marker) - reinterpret_cast<char*>(out_page));
  491. out_page->compact_page_marker = marker;
  492. // since we don't reuse the page space until we reallocate it, we can just pretend that we freed the marker block
  493. // this will make sure deallocate_memory correctly tracks the size
  494. out_page->freed_size += sizeof(uint32_t);
  495. return marker + 1;
  496. }
  497. else
  498. {
  499. // roll back uint32_t part
  500. _busy_size -= sizeof(uint32_t);
  501. return result;
  502. }
  503. }
  504. #else
  505. void* allocate_object(size_t size, xml_memory_page*& out_page)
  506. {
  507. return allocate_memory(size, out_page);
  508. }
  509. #endif
  510. void deallocate_memory(void* ptr, size_t size, xml_memory_page* page)
  511. {
  512. if (page == _root) page->busy_size = _busy_size;
  513. assert(ptr >= reinterpret_cast<char*>(page) + sizeof(xml_memory_page) && ptr < reinterpret_cast<char*>(page) + sizeof(xml_memory_page) + page->busy_size);
  514. (void)!ptr;
  515. page->freed_size += size;
  516. assert(page->freed_size <= page->busy_size);
  517. if (page->freed_size == page->busy_size)
  518. {
  519. if (page->next == NULL)
  520. {
  521. assert(_root == page);
  522. // top page freed, just reset sizes
  523. page->busy_size = 0;
  524. page->freed_size = 0;
  525. #ifdef PUGIXML_COMPACT
  526. // reset compact state to maximize efficiency
  527. page->compact_string_base = NULL;
  528. page->compact_shared_parent = NULL;
  529. page->compact_page_marker = NULL;
  530. #endif
  531. _busy_size = 0;
  532. }
  533. else
  534. {
  535. assert(_root != page);
  536. assert(page->prev);
  537. // remove from the list
  538. page->prev->next = page->next;
  539. page->next->prev = page->prev;
  540. // deallocate
  541. deallocate_page(page);
  542. }
  543. }
  544. }
  545. char_t* allocate_string(size_t length)
  546. {
  547. static const size_t max_encoded_offset = (1 << 16) * xml_memory_block_alignment;
  548. PUGI_IMPL_STATIC_ASSERT(xml_memory_page_size <= max_encoded_offset);
  549. // allocate memory for string and header block
  550. size_t size = sizeof(xml_memory_string_header) + length * sizeof(char_t);
  551. // round size up to block alignment boundary
  552. size_t full_size = (size + (xml_memory_block_alignment - 1)) & ~(xml_memory_block_alignment - 1);
  553. xml_memory_page* page;
  554. xml_memory_string_header* header = static_cast<xml_memory_string_header*>(allocate_memory(full_size, page));
  555. if (!header) return NULL;
  556. // setup header
  557. ptrdiff_t page_offset = reinterpret_cast<char*>(header) - reinterpret_cast<char*>(page) - sizeof(xml_memory_page);
  558. assert(page_offset % xml_memory_block_alignment == 0);
  559. assert(page_offset >= 0 && static_cast<size_t>(page_offset) < max_encoded_offset);
  560. header->page_offset = static_cast<uint16_t>(static_cast<size_t>(page_offset) / xml_memory_block_alignment);
  561. // full_size == 0 for large strings that occupy the whole page
  562. assert(full_size % xml_memory_block_alignment == 0);
  563. assert(full_size < max_encoded_offset || (page->busy_size == full_size && page_offset == 0));
  564. header->full_size = static_cast<uint16_t>(full_size < max_encoded_offset ? full_size / xml_memory_block_alignment : 0);
  565. // round-trip through void* to avoid 'cast increases required alignment of target type' warning
  566. // header is guaranteed a pointer-sized alignment, which should be enough for char_t
  567. return static_cast<char_t*>(static_cast<void*>(header + 1));
  568. }
  569. void deallocate_string(char_t* string)
  570. {
  571. // this function casts pointers through void* to avoid 'cast increases required alignment of target type' warnings
  572. // we're guaranteed the proper (pointer-sized) alignment on the input string if it was allocated via allocate_string
  573. // get header
  574. xml_memory_string_header* header = static_cast<xml_memory_string_header*>(static_cast<void*>(string)) - 1;
  575. assert(header);
  576. // deallocate
  577. size_t page_offset = sizeof(xml_memory_page) + header->page_offset * xml_memory_block_alignment;
  578. xml_memory_page* page = reinterpret_cast<xml_memory_page*>(static_cast<void*>(reinterpret_cast<char*>(header) - page_offset));
  579. // if full_size == 0 then this string occupies the whole page
  580. size_t full_size = header->full_size == 0 ? page->busy_size : header->full_size * xml_memory_block_alignment;
  581. deallocate_memory(header, full_size, page);
  582. }
  583. bool reserve()
  584. {
  585. #ifdef PUGIXML_COMPACT
  586. return _hash->reserve();
  587. #else
  588. return true;
  589. #endif
  590. }
  591. xml_memory_page* _root;
  592. size_t _busy_size;
  593. #ifdef PUGIXML_COMPACT
  594. compact_hash_table* _hash;
  595. #endif
  596. };
  597. PUGI_IMPL_FN_NO_INLINE void* xml_allocator::allocate_memory_oob(size_t size, xml_memory_page*& out_page)
  598. {
  599. const size_t large_allocation_threshold = xml_memory_page_size / 4;
  600. xml_memory_page* page = allocate_page(size <= large_allocation_threshold ? xml_memory_page_size : size);
  601. out_page = page;
  602. if (!page) return NULL;
  603. if (size <= large_allocation_threshold)
  604. {
  605. _root->busy_size = _busy_size;
  606. // insert page at the end of linked list
  607. page->prev = _root;
  608. _root->next = page;
  609. _root = page;
  610. _busy_size = size;
  611. }
  612. else
  613. {
  614. // insert page before the end of linked list, so that it is deleted as soon as possible
  615. // the last page is not deleted even if it's empty (see deallocate_memory)
  616. assert(_root->prev);
  617. page->prev = _root->prev;
  618. page->next = _root;
  619. _root->prev->next = page;
  620. _root->prev = page;
  621. page->busy_size = size;
  622. }
  623. return reinterpret_cast<char*>(page) + sizeof(xml_memory_page);
  624. }
  625. PUGI_IMPL_NS_END
  626. #ifdef PUGIXML_COMPACT
  627. PUGI_IMPL_NS_BEGIN
  628. static const uintptr_t compact_alignment_log2 = 2;
  629. static const uintptr_t compact_alignment = 1 << compact_alignment_log2;
  630. class compact_header
  631. {
  632. public:
  633. compact_header(xml_memory_page* page, unsigned int flags)
  634. {
  635. PUGI_IMPL_STATIC_ASSERT(xml_memory_block_alignment == compact_alignment);
  636. ptrdiff_t offset = (reinterpret_cast<char*>(this) - reinterpret_cast<char*>(page->compact_page_marker));
  637. assert(offset % compact_alignment == 0 && static_cast<uintptr_t>(offset) < 256 * compact_alignment);
  638. _page = static_cast<unsigned char>(offset >> compact_alignment_log2);
  639. _flags = static_cast<unsigned char>(flags);
  640. }
  641. void operator&=(uintptr_t mod)
  642. {
  643. _flags &= static_cast<unsigned char>(mod);
  644. }
  645. void operator|=(uintptr_t mod)
  646. {
  647. _flags |= static_cast<unsigned char>(mod);
  648. }
  649. uintptr_t operator&(uintptr_t mod) const
  650. {
  651. return _flags & mod;
  652. }
  653. xml_memory_page* get_page() const
  654. {
  655. // round-trip through void* to silence 'cast increases required alignment of target type' warnings
  656. const char* page_marker = reinterpret_cast<const char*>(this) - (_page << compact_alignment_log2);
  657. const char* page = page_marker - *reinterpret_cast<const uint32_t*>(static_cast<const void*>(page_marker));
  658. return const_cast<xml_memory_page*>(reinterpret_cast<const xml_memory_page*>(static_cast<const void*>(page)));
  659. }
  660. private:
  661. unsigned char _page;
  662. unsigned char _flags;
  663. };
  664. PUGI_IMPL_FN xml_memory_page* compact_get_page(const void* object, int header_offset)
  665. {
  666. const compact_header* header = reinterpret_cast<const compact_header*>(static_cast<const char*>(object) - header_offset);
  667. return header->get_page();
  668. }
  669. template <int header_offset, typename T> PUGI_IMPL_FN_NO_INLINE T* compact_get_value(const void* object)
  670. {
  671. return static_cast<T*>(compact_get_page(object, header_offset)->allocator->_hash->find(object));
  672. }
  673. template <int header_offset, typename T> PUGI_IMPL_FN_NO_INLINE void compact_set_value(const void* object, T* value)
  674. {
  675. compact_get_page(object, header_offset)->allocator->_hash->insert(object, value);
  676. }
  677. template <typename T, int header_offset, int start = -126> class compact_pointer
  678. {
  679. public:
  680. compact_pointer(): _data(0)
  681. {
  682. }
  683. void operator=(const compact_pointer& rhs)
  684. {
  685. *this = rhs + 0;
  686. }
  687. void operator=(T* value)
  688. {
  689. if (value)
  690. {
  691. // value is guaranteed to be compact-aligned; 'this' is not
  692. // our decoding is based on 'this' aligned to compact alignment downwards (see operator T*)
  693. // so for negative offsets (e.g. -3) we need to adjust the diff by compact_alignment - 1 to
  694. // compensate for arithmetic shift rounding for negative values
  695. ptrdiff_t diff = reinterpret_cast<char*>(value) - reinterpret_cast<char*>(this);
  696. ptrdiff_t offset = ((diff + int(compact_alignment - 1)) >> compact_alignment_log2) - start;
  697. if (static_cast<uintptr_t>(offset) <= 253)
  698. _data = static_cast<unsigned char>(offset + 1);
  699. else
  700. {
  701. compact_set_value<header_offset>(this, value);
  702. _data = 255;
  703. }
  704. }
  705. else
  706. _data = 0;
  707. }
  708. operator T*() const
  709. {
  710. if (_data)
  711. {
  712. if (_data < 255)
  713. {
  714. uintptr_t base = reinterpret_cast<uintptr_t>(this) & ~(compact_alignment - 1);
  715. return reinterpret_cast<T*>(base + (_data - 1 + start) * compact_alignment);
  716. }
  717. else
  718. return compact_get_value<header_offset, T>(this);
  719. }
  720. else
  721. return NULL;
  722. }
  723. T* operator->() const
  724. {
  725. return *this;
  726. }
  727. private:
  728. unsigned char _data;
  729. };
  730. template <typename T, int header_offset> class compact_pointer_parent
  731. {
  732. public:
  733. compact_pointer_parent(): _data(0)
  734. {
  735. }
  736. void operator=(const compact_pointer_parent& rhs)
  737. {
  738. *this = rhs + 0;
  739. }
  740. void operator=(T* value)
  741. {
  742. if (value)
  743. {
  744. // value is guaranteed to be compact-aligned; 'this' is not
  745. // our decoding is based on 'this' aligned to compact alignment downwards (see operator T*)
  746. // so for negative offsets (e.g. -3) we need to adjust the diff by compact_alignment - 1 to
  747. // compensate for arithmetic shift behavior for negative values
  748. ptrdiff_t diff = reinterpret_cast<char*>(value) - reinterpret_cast<char*>(this);
  749. ptrdiff_t offset = ((diff + int(compact_alignment - 1)) >> compact_alignment_log2) + 65533;
  750. if (static_cast<uintptr_t>(offset) <= 65533)
  751. {
  752. _data = static_cast<unsigned short>(offset + 1);
  753. }
  754. else
  755. {
  756. xml_memory_page* page = compact_get_page(this, header_offset);
  757. if (PUGI_IMPL_UNLIKELY(page->compact_shared_parent == NULL))
  758. page->compact_shared_parent = value;
  759. if (page->compact_shared_parent == value)
  760. {
  761. _data = 65534;
  762. }
  763. else
  764. {
  765. compact_set_value<header_offset>(this, value);
  766. _data = 65535;
  767. }
  768. }
  769. }
  770. else
  771. {
  772. _data = 0;
  773. }
  774. }
  775. operator T*() const
  776. {
  777. if (_data)
  778. {
  779. if (_data < 65534)
  780. {
  781. uintptr_t base = reinterpret_cast<uintptr_t>(this) & ~(compact_alignment - 1);
  782. return reinterpret_cast<T*>(base + (_data - 1 - 65533) * compact_alignment);
  783. }
  784. else if (_data == 65534)
  785. return static_cast<T*>(compact_get_page(this, header_offset)->compact_shared_parent);
  786. else
  787. return compact_get_value<header_offset, T>(this);
  788. }
  789. else
  790. return NULL;
  791. }
  792. T* operator->() const
  793. {
  794. return *this;
  795. }
  796. private:
  797. uint16_t _data;
  798. };
  799. template <int header_offset, int base_offset> class compact_string
  800. {
  801. public:
  802. compact_string(): _data(0)
  803. {
  804. }
  805. void operator=(const compact_string& rhs)
  806. {
  807. *this = rhs + 0;
  808. }
  809. void operator=(char_t* value)
  810. {
  811. if (value)
  812. {
  813. xml_memory_page* page = compact_get_page(this, header_offset);
  814. if (PUGI_IMPL_UNLIKELY(page->compact_string_base == NULL))
  815. page->compact_string_base = value;
  816. ptrdiff_t offset = value - page->compact_string_base;
  817. if (static_cast<uintptr_t>(offset) < (65535 << 7))
  818. {
  819. // round-trip through void* to silence 'cast increases required alignment of target type' warnings
  820. uint16_t* base = reinterpret_cast<uint16_t*>(static_cast<void*>(reinterpret_cast<char*>(this) - base_offset));
  821. if (*base == 0)
  822. {
  823. *base = static_cast<uint16_t>((offset >> 7) + 1);
  824. _data = static_cast<unsigned char>((offset & 127) + 1);
  825. }
  826. else
  827. {
  828. ptrdiff_t remainder = offset - ((*base - 1) << 7);
  829. if (static_cast<uintptr_t>(remainder) <= 253)
  830. {
  831. _data = static_cast<unsigned char>(remainder + 1);
  832. }
  833. else
  834. {
  835. compact_set_value<header_offset>(this, value);
  836. _data = 255;
  837. }
  838. }
  839. }
  840. else
  841. {
  842. compact_set_value<header_offset>(this, value);
  843. _data = 255;
  844. }
  845. }
  846. else
  847. {
  848. _data = 0;
  849. }
  850. }
  851. operator char_t*() const
  852. {
  853. if (_data)
  854. {
  855. if (_data < 255)
  856. {
  857. xml_memory_page* page = compact_get_page(this, header_offset);
  858. // round-trip through void* to silence 'cast increases required alignment of target type' warnings
  859. const uint16_t* base = reinterpret_cast<const uint16_t*>(static_cast<const void*>(reinterpret_cast<const char*>(this) - base_offset));
  860. assert(*base);
  861. ptrdiff_t offset = ((*base - 1) << 7) + (_data - 1);
  862. return page->compact_string_base + offset;
  863. }
  864. else
  865. {
  866. return compact_get_value<header_offset, char_t>(this);
  867. }
  868. }
  869. else
  870. return NULL;
  871. }
  872. private:
  873. unsigned char _data;
  874. };
  875. PUGI_IMPL_NS_END
  876. #endif
  877. #ifdef PUGIXML_COMPACT
  878. namespace pugi
  879. {
  880. struct xml_attribute_struct
  881. {
  882. xml_attribute_struct(impl::xml_memory_page* page): header(page, 0), namevalue_base(0)
  883. {
  884. PUGI_IMPL_STATIC_ASSERT(sizeof(xml_attribute_struct) == 8);
  885. }
  886. impl::compact_header header;
  887. uint16_t namevalue_base;
  888. impl::compact_string<4, 2> name;
  889. impl::compact_string<5, 3> value;
  890. impl::compact_pointer<xml_attribute_struct, 6> prev_attribute_c;
  891. impl::compact_pointer<xml_attribute_struct, 7, 0> next_attribute;
  892. };
  893. struct xml_node_struct
  894. {
  895. xml_node_struct(impl::xml_memory_page* page, xml_node_type type): header(page, type), namevalue_base(0)
  896. {
  897. PUGI_IMPL_STATIC_ASSERT(sizeof(xml_node_struct) == 12);
  898. }
  899. impl::compact_header header;
  900. uint16_t namevalue_base;
  901. impl::compact_string<4, 2> name;
  902. impl::compact_string<5, 3> value;
  903. impl::compact_pointer_parent<xml_node_struct, 6> parent;
  904. impl::compact_pointer<xml_node_struct, 8, 0> first_child;
  905. impl::compact_pointer<xml_node_struct, 9> prev_sibling_c;
  906. impl::compact_pointer<xml_node_struct, 10, 0> next_sibling;
  907. impl::compact_pointer<xml_attribute_struct, 11, 0> first_attribute;
  908. };
  909. }
  910. #else
  911. namespace pugi
  912. {
  913. struct xml_attribute_struct
  914. {
  915. xml_attribute_struct(impl::xml_memory_page* page): name(NULL), value(NULL), prev_attribute_c(NULL), next_attribute(NULL)
  916. {
  917. header = PUGI_IMPL_GETHEADER_IMPL(this, page, 0);
  918. }
  919. uintptr_t header;
  920. char_t* name;
  921. char_t* value;
  922. xml_attribute_struct* prev_attribute_c;
  923. xml_attribute_struct* next_attribute;
  924. };
  925. struct xml_node_struct
  926. {
  927. xml_node_struct(impl::xml_memory_page* page, xml_node_type type): name(NULL), value(NULL), parent(NULL), first_child(NULL), prev_sibling_c(NULL), next_sibling(NULL), first_attribute(NULL)
  928. {
  929. header = PUGI_IMPL_GETHEADER_IMPL(this, page, type);
  930. }
  931. uintptr_t header;
  932. char_t* name;
  933. char_t* value;
  934. xml_node_struct* parent;
  935. xml_node_struct* first_child;
  936. xml_node_struct* prev_sibling_c;
  937. xml_node_struct* next_sibling;
  938. xml_attribute_struct* first_attribute;
  939. };
  940. }
  941. #endif
  942. PUGI_IMPL_NS_BEGIN
  943. struct xml_extra_buffer
  944. {
  945. char_t* buffer;
  946. xml_extra_buffer* next;
  947. };
  948. struct xml_document_struct: public xml_node_struct, public xml_allocator
  949. {
  950. xml_document_struct(xml_memory_page* page): xml_node_struct(page, node_document), xml_allocator(page), buffer(NULL), extra_buffers(NULL)
  951. {
  952. }
  953. const char_t* buffer;
  954. xml_extra_buffer* extra_buffers;
  955. #ifdef PUGIXML_COMPACT
  956. compact_hash_table hash;
  957. #endif
  958. };
  959. template <typename Object> inline xml_allocator& get_allocator(const Object* object)
  960. {
  961. assert(object);
  962. return *PUGI_IMPL_GETPAGE(object)->allocator;
  963. }
  964. template <typename Object> inline xml_document_struct& get_document(const Object* object)
  965. {
  966. assert(object);
  967. return *static_cast<xml_document_struct*>(PUGI_IMPL_GETPAGE(object)->allocator);
  968. }
  969. PUGI_IMPL_NS_END
  970. // Low-level DOM operations
  971. PUGI_IMPL_NS_BEGIN
  972. inline xml_attribute_struct* allocate_attribute(xml_allocator& alloc)
  973. {
  974. xml_memory_page* page;
  975. void* memory = alloc.allocate_object(sizeof(xml_attribute_struct), page);
  976. if (!memory) return NULL;
  977. return new (memory) xml_attribute_struct(page);
  978. }
  979. inline xml_node_struct* allocate_node(xml_allocator& alloc, xml_node_type type)
  980. {
  981. xml_memory_page* page;
  982. void* memory = alloc.allocate_object(sizeof(xml_node_struct), page);
  983. if (!memory) return NULL;
  984. return new (memory) xml_node_struct(page, type);
  985. }
  986. inline void destroy_attribute(xml_attribute_struct* a, xml_allocator& alloc)
  987. {
  988. if (a->header & impl::xml_memory_page_name_allocated_mask)
  989. alloc.deallocate_string(a->name);
  990. if (a->header & impl::xml_memory_page_value_allocated_mask)
  991. alloc.deallocate_string(a->value);
  992. alloc.deallocate_memory(a, sizeof(xml_attribute_struct), PUGI_IMPL_GETPAGE(a));
  993. }
  994. inline void destroy_node(xml_node_struct* n, xml_allocator& alloc)
  995. {
  996. if (n->header & impl::xml_memory_page_name_allocated_mask)
  997. alloc.deallocate_string(n->name);
  998. if (n->header & impl::xml_memory_page_value_allocated_mask)
  999. alloc.deallocate_string(n->value);
  1000. for (xml_attribute_struct* attr = n->first_attribute; attr; )
  1001. {
  1002. xml_attribute_struct* next = attr->next_attribute;
  1003. destroy_attribute(attr, alloc);
  1004. attr = next;
  1005. }
  1006. for (xml_node_struct* child = n->first_child; child; )
  1007. {
  1008. xml_node_struct* next = child->next_sibling;
  1009. destroy_node(child, alloc);
  1010. child = next;
  1011. }
  1012. alloc.deallocate_memory(n, sizeof(xml_node_struct), PUGI_IMPL_GETPAGE(n));
  1013. }
  1014. inline void append_node(xml_node_struct* child, xml_node_struct* node)
  1015. {
  1016. child->parent = node;
  1017. xml_node_struct* head = node->first_child;
  1018. if (head)
  1019. {
  1020. xml_node_struct* tail = head->prev_sibling_c;
  1021. tail->next_sibling = child;
  1022. child->prev_sibling_c = tail;
  1023. head->prev_sibling_c = child;
  1024. }
  1025. else
  1026. {
  1027. node->first_child = child;
  1028. child->prev_sibling_c = child;
  1029. }
  1030. }
  1031. inline void prepend_node(xml_node_struct* child, xml_node_struct* node)
  1032. {
  1033. child->parent = node;
  1034. xml_node_struct* head = node->first_child;
  1035. if (head)
  1036. {
  1037. child->prev_sibling_c = head->prev_sibling_c;
  1038. head->prev_sibling_c = child;
  1039. }
  1040. else
  1041. child->prev_sibling_c = child;
  1042. child->next_sibling = head;
  1043. node->first_child = child;
  1044. }
  1045. inline void insert_node_after(xml_node_struct* child, xml_node_struct* node)
  1046. {
  1047. xml_node_struct* parent = node->parent;
  1048. child->parent = parent;
  1049. xml_node_struct* next = node->next_sibling;
  1050. if (next)
  1051. next->prev_sibling_c = child;
  1052. else
  1053. parent->first_child->prev_sibling_c = child;
  1054. child->next_sibling = next;
  1055. child->prev_sibling_c = node;
  1056. node->next_sibling = child;
  1057. }
  1058. inline void insert_node_before(xml_node_struct* child, xml_node_struct* node)
  1059. {
  1060. xml_node_struct* parent = node->parent;
  1061. child->parent = parent;
  1062. xml_node_struct* prev = node->prev_sibling_c;
  1063. if (prev->next_sibling)
  1064. prev->next_sibling = child;
  1065. else
  1066. parent->first_child = child;
  1067. child->prev_sibling_c = prev;
  1068. child->next_sibling = node;
  1069. node->prev_sibling_c = child;
  1070. }
  1071. inline void remove_node(xml_node_struct* node)
  1072. {
  1073. xml_node_struct* parent = node->parent;
  1074. xml_node_struct* next = node->next_sibling;
  1075. xml_node_struct* prev = node->prev_sibling_c;
  1076. if (next)
  1077. next->prev_sibling_c = prev;
  1078. else
  1079. parent->first_child->prev_sibling_c = prev;
  1080. if (prev->next_sibling)
  1081. prev->next_sibling = next;
  1082. else
  1083. parent->first_child = next;
  1084. node->parent = NULL;
  1085. node->prev_sibling_c = NULL;
  1086. node->next_sibling = NULL;
  1087. }
  1088. inline void append_attribute(xml_attribute_struct* attr, xml_node_struct* node)
  1089. {
  1090. xml_attribute_struct* head = node->first_attribute;
  1091. if (head)
  1092. {
  1093. xml_attribute_struct* tail = head->prev_attribute_c;
  1094. tail->next_attribute = attr;
  1095. attr->prev_attribute_c = tail;
  1096. head->prev_attribute_c = attr;
  1097. }
  1098. else
  1099. {
  1100. node->first_attribute = attr;
  1101. attr->prev_attribute_c = attr;
  1102. }
  1103. }
  1104. inline void prepend_attribute(xml_attribute_struct* attr, xml_node_struct* node)
  1105. {
  1106. xml_attribute_struct* head = node->first_attribute;
  1107. if (head)
  1108. {
  1109. attr->prev_attribute_c = head->prev_attribute_c;
  1110. head->prev_attribute_c = attr;
  1111. }
  1112. else
  1113. attr->prev_attribute_c = attr;
  1114. attr->next_attribute = head;
  1115. node->first_attribute = attr;
  1116. }
  1117. inline void insert_attribute_after(xml_attribute_struct* attr, xml_attribute_struct* place, xml_node_struct* node)
  1118. {
  1119. xml_attribute_struct* next = place->next_attribute;
  1120. if (next)
  1121. next->prev_attribute_c = attr;
  1122. else
  1123. node->first_attribute->prev_attribute_c = attr;
  1124. attr->next_attribute = next;
  1125. attr->prev_attribute_c = place;
  1126. place->next_attribute = attr;
  1127. }
  1128. inline void insert_attribute_before(xml_attribute_struct* attr, xml_attribute_struct* place, xml_node_struct* node)
  1129. {
  1130. xml_attribute_struct* prev = place->prev_attribute_c;
  1131. if (prev->next_attribute)
  1132. prev->next_attribute = attr;
  1133. else
  1134. node->first_attribute = attr;
  1135. attr->prev_attribute_c = prev;
  1136. attr->next_attribute = place;
  1137. place->prev_attribute_c = attr;
  1138. }
  1139. inline void remove_attribute(xml_attribute_struct* attr, xml_node_struct* node)
  1140. {
  1141. xml_attribute_struct* next = attr->next_attribute;
  1142. xml_attribute_struct* prev = attr->prev_attribute_c;
  1143. if (next)
  1144. next->prev_attribute_c = prev;
  1145. else
  1146. node->first_attribute->prev_attribute_c = prev;
  1147. if (prev->next_attribute)
  1148. prev->next_attribute = next;
  1149. else
  1150. node->first_attribute = next;
  1151. attr->prev_attribute_c = NULL;
  1152. attr->next_attribute = NULL;
  1153. }
  1154. PUGI_IMPL_FN_NO_INLINE xml_node_struct* append_new_node(xml_node_struct* node, xml_allocator& alloc, xml_node_type type = node_element)
  1155. {
  1156. if (!alloc.reserve()) return NULL;
  1157. xml_node_struct* child = allocate_node(alloc, type);
  1158. if (!child) return NULL;
  1159. append_node(child, node);
  1160. return child;
  1161. }
  1162. PUGI_IMPL_FN_NO_INLINE xml_attribute_struct* append_new_attribute(xml_node_struct* node, xml_allocator& alloc)
  1163. {
  1164. if (!alloc.reserve()) return NULL;
  1165. xml_attribute_struct* attr = allocate_attribute(alloc);
  1166. if (!attr) return NULL;
  1167. append_attribute(attr, node);
  1168. return attr;
  1169. }
  1170. PUGI_IMPL_NS_END
  1171. // Helper classes for code generation
  1172. PUGI_IMPL_NS_BEGIN
  1173. struct opt_false
  1174. {
  1175. enum { value = 0 };
  1176. };
  1177. struct opt_true
  1178. {
  1179. enum { value = 1 };
  1180. };
  1181. PUGI_IMPL_NS_END
  1182. // Unicode utilities
  1183. PUGI_IMPL_NS_BEGIN
  1184. inline uint16_t endian_swap(uint16_t value)
  1185. {
  1186. return static_cast<uint16_t>(((value & 0xff) << 8) | (value >> 8));
  1187. }
  1188. inline uint32_t endian_swap(uint32_t value)
  1189. {
  1190. return ((value & 0xff) << 24) | ((value & 0xff00) << 8) | ((value & 0xff0000) >> 8) | (value >> 24);
  1191. }
  1192. struct utf8_counter
  1193. {
  1194. typedef size_t value_type;
  1195. static value_type low(value_type result, uint32_t ch)
  1196. {
  1197. // U+0000..U+007F
  1198. if (ch < 0x80) return result + 1;
  1199. // U+0080..U+07FF
  1200. else if (ch < 0x800) return result + 2;
  1201. // U+0800..U+FFFF
  1202. else return result + 3;
  1203. }
  1204. static value_type high(value_type result, uint32_t)
  1205. {
  1206. // U+10000..U+10FFFF
  1207. return result + 4;
  1208. }
  1209. };
  1210. struct utf8_writer
  1211. {
  1212. typedef uint8_t* value_type;
  1213. static value_type low(value_type result, uint32_t ch)
  1214. {
  1215. // U+0000..U+007F
  1216. if (ch < 0x80)
  1217. {
  1218. *result = static_cast<uint8_t>(ch);
  1219. return result + 1;
  1220. }
  1221. // U+0080..U+07FF
  1222. else if (ch < 0x800)
  1223. {
  1224. result[0] = static_cast<uint8_t>(0xC0 | (ch >> 6));
  1225. result[1] = static_cast<uint8_t>(0x80 | (ch & 0x3F));
  1226. return result + 2;
  1227. }
  1228. // U+0800..U+FFFF
  1229. else
  1230. {
  1231. result[0] = static_cast<uint8_t>(0xE0 | (ch >> 12));
  1232. result[1] = static_cast<uint8_t>(0x80 | ((ch >> 6) & 0x3F));
  1233. result[2] = static_cast<uint8_t>(0x80 | (ch & 0x3F));
  1234. return result + 3;
  1235. }
  1236. }
  1237. static value_type high(value_type result, uint32_t ch)
  1238. {
  1239. // U+10000..U+10FFFF
  1240. result[0] = static_cast<uint8_t>(0xF0 | (ch >> 18));
  1241. result[1] = static_cast<uint8_t>(0x80 | ((ch >> 12) & 0x3F));
  1242. result[2] = static_cast<uint8_t>(0x80 | ((ch >> 6) & 0x3F));
  1243. result[3] = static_cast<uint8_t>(0x80 | (ch & 0x3F));
  1244. return result + 4;
  1245. }
  1246. static value_type any(value_type result, uint32_t ch)
  1247. {
  1248. return (ch < 0x10000) ? low(result, ch) : high(result, ch);
  1249. }
  1250. };
  1251. struct utf16_counter
  1252. {
  1253. typedef size_t value_type;
  1254. static value_type low(value_type result, uint32_t)
  1255. {
  1256. return result + 1;
  1257. }
  1258. static value_type high(value_type result, uint32_t)
  1259. {
  1260. return result + 2;
  1261. }
  1262. };
  1263. struct utf16_writer
  1264. {
  1265. typedef uint16_t* value_type;
  1266. static value_type low(value_type result, uint32_t ch)
  1267. {
  1268. *result = static_cast<uint16_t>(ch);
  1269. return result + 1;
  1270. }
  1271. static value_type high(value_type result, uint32_t ch)
  1272. {
  1273. uint32_t msh = (ch - 0x10000U) >> 10;
  1274. uint32_t lsh = (ch - 0x10000U) & 0x3ff;
  1275. result[0] = static_cast<uint16_t>(0xD800 + msh);
  1276. result[1] = static_cast<uint16_t>(0xDC00 + lsh);
  1277. return result + 2;
  1278. }
  1279. static value_type any(value_type result, uint32_t ch)
  1280. {
  1281. return (ch < 0x10000) ? low(result, ch) : high(result, ch);
  1282. }
  1283. };
  1284. struct utf32_counter
  1285. {
  1286. typedef size_t value_type;
  1287. static value_type low(value_type result, uint32_t)
  1288. {
  1289. return result + 1;
  1290. }
  1291. static value_type high(value_type result, uint32_t)
  1292. {
  1293. return result + 1;
  1294. }
  1295. };
  1296. struct utf32_writer
  1297. {
  1298. typedef uint32_t* value_type;
  1299. static value_type low(value_type result, uint32_t ch)
  1300. {
  1301. *result = ch;
  1302. return result + 1;
  1303. }
  1304. static value_type high(value_type result, uint32_t ch)
  1305. {
  1306. *result = ch;
  1307. return result + 1;
  1308. }
  1309. static value_type any(value_type result, uint32_t ch)
  1310. {
  1311. *result = ch;
  1312. return result + 1;
  1313. }
  1314. };
  1315. struct latin1_writer
  1316. {
  1317. typedef uint8_t* value_type;
  1318. static value_type low(value_type result, uint32_t ch)
  1319. {
  1320. *result = static_cast<uint8_t>(ch > 255 ? '?' : ch);
  1321. return result + 1;
  1322. }
  1323. static value_type high(value_type result, uint32_t ch)
  1324. {
  1325. (void)ch;
  1326. *result = '?';
  1327. return result + 1;
  1328. }
  1329. };
  1330. struct utf8_decoder
  1331. {
  1332. typedef uint8_t type;
  1333. template <typename Traits> static inline typename Traits::value_type process(const uint8_t* data, size_t size, typename Traits::value_type result, Traits)
  1334. {
  1335. const uint8_t utf8_byte_mask = 0x3f;
  1336. while (size)
  1337. {
  1338. uint8_t lead = *data;
  1339. // 0xxxxxxx -> U+0000..U+007F
  1340. if (lead < 0x80)
  1341. {
  1342. result = Traits::low(result, lead);
  1343. data += 1;
  1344. size -= 1;
  1345. // process aligned single-byte (ascii) blocks
  1346. if ((reinterpret_cast<uintptr_t>(data) & 3) == 0)
  1347. {
  1348. // round-trip through void* to silence 'cast increases required alignment of target type' warnings
  1349. while (size >= 4 && (*static_cast<const uint32_t*>(static_cast<const void*>(data)) & 0x80808080) == 0)
  1350. {
  1351. result = Traits::low(result, data[0]);
  1352. result = Traits::low(result, data[1]);
  1353. result = Traits::low(result, data[2]);
  1354. result = Traits::low(result, data[3]);
  1355. data += 4;
  1356. size -= 4;
  1357. }
  1358. }
  1359. }
  1360. // 110xxxxx -> U+0080..U+07FF
  1361. else if (static_cast<unsigned int>(lead - 0xC0) < 0x20 && size >= 2 && (data[1] & 0xc0) == 0x80)
  1362. {
  1363. result = Traits::low(result, ((lead & ~0xC0) << 6) | (data[1] & utf8_byte_mask));
  1364. data += 2;
  1365. size -= 2;
  1366. }
  1367. // 1110xxxx -> U+0800-U+FFFF
  1368. else if (static_cast<unsigned int>(lead - 0xE0) < 0x10 && size >= 3 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80)
  1369. {
  1370. result = Traits::low(result, ((lead & ~0xE0) << 12) | ((data[1] & utf8_byte_mask) << 6) | (data[2] & utf8_byte_mask));
  1371. data += 3;
  1372. size -= 3;
  1373. }
  1374. // 11110xxx -> U+10000..U+10FFFF
  1375. else if (static_cast<unsigned int>(lead - 0xF0) < 0x08 && size >= 4 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80 && (data[3] & 0xc0) == 0x80)
  1376. {
  1377. result = Traits::high(result, ((lead & ~0xF0) << 18) | ((data[1] & utf8_byte_mask) << 12) | ((data[2] & utf8_byte_mask) << 6) | (data[3] & utf8_byte_mask));
  1378. data += 4;
  1379. size -= 4;
  1380. }
  1381. // 10xxxxxx or 11111xxx -> invalid
  1382. else
  1383. {
  1384. data += 1;
  1385. size -= 1;
  1386. }
  1387. }
  1388. return result;
  1389. }
  1390. };
  1391. template <typename opt_swap> struct utf16_decoder
  1392. {
  1393. typedef uint16_t type;
  1394. template <typename Traits> static inline typename Traits::value_type process(const uint16_t* data, size_t size, typename Traits::value_type result, Traits)
  1395. {
  1396. while (size)
  1397. {
  1398. uint16_t lead = opt_swap::value ? endian_swap(*data) : *data;
  1399. // U+0000..U+D7FF
  1400. if (lead < 0xD800)
  1401. {
  1402. result = Traits::low(result, lead);
  1403. data += 1;
  1404. size -= 1;
  1405. }
  1406. // U+E000..U+FFFF
  1407. else if (static_cast<unsigned int>(lead - 0xE000) < 0x2000)
  1408. {
  1409. result = Traits::low(result, lead);
  1410. data += 1;
  1411. size -= 1;
  1412. }
  1413. // surrogate pair lead
  1414. else if (static_cast<unsigned int>(lead - 0xD800) < 0x400 && size >= 2)
  1415. {
  1416. uint16_t next = opt_swap::value ? endian_swap(data[1]) : data[1];
  1417. if (static_cast<unsigned int>(next - 0xDC00) < 0x400)
  1418. {
  1419. result = Traits::high(result, 0x10000 + ((lead & 0x3ff) << 10) + (next & 0x3ff));
  1420. data += 2;
  1421. size -= 2;
  1422. }
  1423. else
  1424. {
  1425. data += 1;
  1426. size -= 1;
  1427. }
  1428. }
  1429. else
  1430. {
  1431. data += 1;
  1432. size -= 1;
  1433. }
  1434. }
  1435. return result;
  1436. }
  1437. };
  1438. template <typename opt_swap> struct utf32_decoder
  1439. {
  1440. typedef uint32_t type;
  1441. template <typename Traits> static inline typename Traits::value_type process(const uint32_t* data, size_t size, typename Traits::value_type result, Traits)
  1442. {
  1443. while (size)
  1444. {
  1445. uint32_t lead = opt_swap::value ? endian_swap(*data) : *data;
  1446. // U+0000..U+FFFF
  1447. if (lead < 0x10000)
  1448. {
  1449. result = Traits::low(result, lead);
  1450. data += 1;
  1451. size -= 1;
  1452. }
  1453. // U+10000..U+10FFFF
  1454. else
  1455. {
  1456. result = Traits::high(result, lead);
  1457. data += 1;
  1458. size -= 1;
  1459. }
  1460. }
  1461. return result;
  1462. }
  1463. };
  1464. struct latin1_decoder
  1465. {
  1466. typedef uint8_t type;
  1467. template <typename Traits> static inline typename Traits::value_type process(const uint8_t* data, size_t size, typename Traits::value_type result, Traits)
  1468. {
  1469. while (size)
  1470. {
  1471. result = Traits::low(result, *data);
  1472. data += 1;
  1473. size -= 1;
  1474. }
  1475. return result;
  1476. }
  1477. };
  1478. template <size_t size> struct wchar_selector;
  1479. template <> struct wchar_selector<2>
  1480. {
  1481. typedef uint16_t type;
  1482. typedef utf16_counter counter;
  1483. typedef utf16_writer writer;
  1484. typedef utf16_decoder<opt_false> decoder;
  1485. };
  1486. template <> struct wchar_selector<4>
  1487. {
  1488. typedef uint32_t type;
  1489. typedef utf32_counter counter;
  1490. typedef utf32_writer writer;
  1491. typedef utf32_decoder<opt_false> decoder;
  1492. };
  1493. typedef wchar_selector<sizeof(wchar_t)>::counter wchar_counter;
  1494. typedef wchar_selector<sizeof(wchar_t)>::writer wchar_writer;
  1495. struct wchar_decoder
  1496. {
  1497. typedef wchar_t type;
  1498. template <typename Traits> static inline typename Traits::value_type process(const wchar_t* data, size_t size, typename Traits::value_type result, Traits traits)
  1499. {
  1500. typedef wchar_selector<sizeof(wchar_t)>::decoder decoder;
  1501. return decoder::process(reinterpret_cast<const typename decoder::type*>(data), size, result, traits);
  1502. }
  1503. };
  1504. #ifdef PUGIXML_WCHAR_MODE
  1505. PUGI_IMPL_FN void convert_wchar_endian_swap(wchar_t* result, const wchar_t* data, size_t length)
  1506. {
  1507. for (size_t i = 0; i < length; ++i)
  1508. result[i] = static_cast<wchar_t>(endian_swap(static_cast<wchar_selector<sizeof(wchar_t)>::type>(data[i])));
  1509. }
  1510. #endif
  1511. PUGI_IMPL_NS_END
  1512. PUGI_IMPL_NS_BEGIN
  1513. enum chartype_t
  1514. {
  1515. ct_parse_pcdata = 1, // \0, &, \r, <
  1516. ct_parse_attr = 2, // \0, &, \r, ', "
  1517. ct_parse_attr_ws = 4, // \0, &, \r, ', ", \n, tab
  1518. ct_space = 8, // \r, \n, space, tab
  1519. ct_parse_cdata = 16, // \0, ], >, \r
  1520. ct_parse_comment = 32, // \0, -, >, \r
  1521. ct_symbol = 64, // Any symbol > 127, a-z, A-Z, 0-9, _, :, -, .
  1522. ct_start_symbol = 128 // Any symbol > 127, a-z, A-Z, _, :
  1523. };
  1524. static const unsigned char chartype_table[256] =
  1525. {
  1526. 55, 0, 0, 0, 0, 0, 0, 0, 0, 12, 12, 0, 0, 63, 0, 0, // 0-15
  1527. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 16-31
  1528. 8, 0, 6, 0, 0, 0, 7, 6, 0, 0, 0, 0, 0, 96, 64, 0, // 32-47
  1529. 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 192, 0, 1, 0, 48, 0, // 48-63
  1530. 0, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 64-79
  1531. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 0, 0, 16, 0, 192, // 80-95
  1532. 0, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 96-111
  1533. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 0, 0, 0, 0, 0, // 112-127
  1534. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 128+
  1535. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192,
  1536. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192,
  1537. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192,
  1538. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192,
  1539. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192,
  1540. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192,
  1541. 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192
  1542. };
  1543. enum chartypex_t
  1544. {
  1545. ctx_special_pcdata = 1, // Any symbol >= 0 and < 32 (except \t, \r, \n), &, <, >
  1546. ctx_special_attr = 2, // Any symbol >= 0 and < 32, &, <, ", '
  1547. ctx_start_symbol = 4, // Any symbol > 127, a-z, A-Z, _
  1548. ctx_digit = 8, // 0-9
  1549. ctx_symbol = 16 // Any symbol > 127, a-z, A-Z, 0-9, _, -, .
  1550. };
  1551. static const unsigned char chartypex_table[256] =
  1552. {
  1553. 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3, // 0-15
  1554. 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, // 16-31
  1555. 0, 0, 2, 0, 0, 0, 3, 2, 0, 0, 0, 0, 0, 16, 16, 0, // 32-47
  1556. 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 0, 0, 3, 0, 1, 0, // 48-63
  1557. 0, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 64-79
  1558. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 0, 0, 0, 0, 20, // 80-95
  1559. 0, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 96-111
  1560. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 0, 0, 0, 0, 0, // 112-127
  1561. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 128+
  1562. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
  1563. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
  1564. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
  1565. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
  1566. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
  1567. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
  1568. 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20
  1569. };
  1570. #ifdef PUGIXML_WCHAR_MODE
  1571. #define PUGI_IMPL_IS_CHARTYPE_IMPL(c, ct, table) ((static_cast<unsigned int>(c) < 128 ? table[static_cast<unsigned int>(c)] : table[128]) & (ct))
  1572. #else
  1573. #define PUGI_IMPL_IS_CHARTYPE_IMPL(c, ct, table) (table[static_cast<unsigned char>(c)] & (ct))
  1574. #endif
  1575. #define PUGI_IMPL_IS_CHARTYPE(c, ct) PUGI_IMPL_IS_CHARTYPE_IMPL(c, ct, chartype_table)
  1576. #define PUGI_IMPL_IS_CHARTYPEX(c, ct) PUGI_IMPL_IS_CHARTYPE_IMPL(c, ct, chartypex_table)
  1577. PUGI_IMPL_FN bool is_little_endian()
  1578. {
  1579. unsigned int ui = 1;
  1580. return *reinterpret_cast<unsigned char*>(&ui) == 1;
  1581. }
  1582. PUGI_IMPL_FN xml_encoding get_wchar_encoding()
  1583. {
  1584. PUGI_IMPL_STATIC_ASSERT(sizeof(wchar_t) == 2 || sizeof(wchar_t) == 4);
  1585. if (sizeof(wchar_t) == 2)
  1586. return is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
  1587. else
  1588. return is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
  1589. }
  1590. PUGI_IMPL_FN bool parse_declaration_encoding(const uint8_t* data, size_t size, const uint8_t*& out_encoding, size_t& out_length)
  1591. {
  1592. #define PUGI_IMPL_SCANCHAR(ch) { if (offset >= size || data[offset] != ch) return false; offset++; }
  1593. #define PUGI_IMPL_SCANCHARTYPE(ct) { while (offset < size && PUGI_IMPL_IS_CHARTYPE(data[offset], ct)) offset++; }
  1594. // check if we have a non-empty XML declaration
  1595. if (size < 6 || !((data[0] == '<') & (data[1] == '?') & (data[2] == 'x') & (data[3] == 'm') & (data[4] == 'l') && PUGI_IMPL_IS_CHARTYPE(data[5], ct_space)))
  1596. return false;
  1597. // scan XML declaration until the encoding field
  1598. for (size_t i = 6; i + 1 < size; ++i)
  1599. {
  1600. // declaration can not contain ? in quoted values
  1601. if (data[i] == '?')
  1602. return false;
  1603. if (data[i] == 'e' && data[i + 1] == 'n')
  1604. {
  1605. size_t offset = i;
  1606. // encoding follows the version field which can't contain 'en' so this has to be the encoding if XML is well formed
  1607. PUGI_IMPL_SCANCHAR('e'); PUGI_IMPL_SCANCHAR('n'); PUGI_IMPL_SCANCHAR('c'); PUGI_IMPL_SCANCHAR('o');
  1608. PUGI_IMPL_SCANCHAR('d'); PUGI_IMPL_SCANCHAR('i'); PUGI_IMPL_SCANCHAR('n'); PUGI_IMPL_SCANCHAR('g');
  1609. // S? = S?
  1610. PUGI_IMPL_SCANCHARTYPE(ct_space);
  1611. PUGI_IMPL_SCANCHAR('=');
  1612. PUGI_IMPL_SCANCHARTYPE(ct_space);
  1613. // the only two valid delimiters are ' and "
  1614. uint8_t delimiter = (offset < size && data[offset] == '"') ? '"' : '\'';
  1615. PUGI_IMPL_SCANCHAR(delimiter);
  1616. size_t start = offset;
  1617. out_encoding = data + offset;
  1618. PUGI_IMPL_SCANCHARTYPE(ct_symbol);
  1619. out_length = offset - start;
  1620. PUGI_IMPL_SCANCHAR(delimiter);
  1621. return true;
  1622. }
  1623. }
  1624. return false;
  1625. #undef PUGI_IMPL_SCANCHAR
  1626. #undef PUGI_IMPL_SCANCHARTYPE
  1627. }
  1628. PUGI_IMPL_FN xml_encoding guess_buffer_encoding(const uint8_t* data, size_t size)
  1629. {
  1630. // skip encoding autodetection if input buffer is too small
  1631. if (size < 4) return encoding_utf8;
  1632. uint8_t d0 = data[0], d1 = data[1], d2 = data[2], d3 = data[3];
  1633. // look for BOM in first few bytes
  1634. if (d0 == 0 && d1 == 0 && d2 == 0xfe && d3 == 0xff) return encoding_utf32_be;
  1635. if (d0 == 0xff && d1 == 0xfe && d2 == 0 && d3 == 0) return encoding_utf32_le;
  1636. if (d0 == 0xfe && d1 == 0xff) return encoding_utf16_be;
  1637. if (d0 == 0xff && d1 == 0xfe) return encoding_utf16_le;
  1638. if (d0 == 0xef && d1 == 0xbb && d2 == 0xbf) return encoding_utf8;
  1639. // look for <, <? or <?xm in various encodings
  1640. if (d0 == 0 && d1 == 0 && d2 == 0 && d3 == 0x3c) return encoding_utf32_be;
  1641. if (d0 == 0x3c && d1 == 0 && d2 == 0 && d3 == 0) return encoding_utf32_le;
  1642. if (d0 == 0 && d1 == 0x3c && d2 == 0 && d3 == 0x3f) return encoding_utf16_be;
  1643. if (d0 == 0x3c && d1 == 0 && d2 == 0x3f && d3 == 0) return encoding_utf16_le;
  1644. // look for utf16 < followed by node name (this may fail, but is better than utf8 since it's zero terminated so early)
  1645. if (d0 == 0 && d1 == 0x3c) return encoding_utf16_be;
  1646. if (d0 == 0x3c && d1 == 0) return encoding_utf16_le;
  1647. // no known BOM detected; parse declaration
  1648. const uint8_t* enc = NULL;
  1649. size_t enc_length = 0;
  1650. if (d0 == 0x3c && d1 == 0x3f && d2 == 0x78 && d3 == 0x6d && parse_declaration_encoding(data, size, enc, enc_length))
  1651. {
  1652. // iso-8859-1 (case-insensitive)
  1653. if (enc_length == 10
  1654. && (enc[0] | ' ') == 'i' && (enc[1] | ' ') == 's' && (enc[2] | ' ') == 'o'
  1655. && enc[3] == '-' && enc[4] == '8' && enc[5] == '8' && enc[6] == '5' && enc[7] == '9'
  1656. && enc[8] == '-' && enc[9] == '1')
  1657. return encoding_latin1;
  1658. // latin1 (case-insensitive)
  1659. if (enc_length == 6
  1660. && (enc[0] | ' ') == 'l' && (enc[1] | ' ') == 'a' && (enc[2] | ' ') == 't'
  1661. && (enc[3] | ' ') == 'i' && (enc[4] | ' ') == 'n'
  1662. && enc[5] == '1')
  1663. return encoding_latin1;
  1664. }
  1665. return encoding_utf8;
  1666. }
  1667. PUGI_IMPL_FN xml_encoding get_buffer_encoding(xml_encoding encoding, const void* contents, size_t size)
  1668. {
  1669. // replace wchar encoding with utf implementation
  1670. if (encoding == encoding_wchar) return get_wchar_encoding();
  1671. // replace utf16 encoding with utf16 with specific endianness
  1672. if (encoding == encoding_utf16) return is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
  1673. // replace utf32 encoding with utf32 with specific endianness
  1674. if (encoding == encoding_utf32) return is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
  1675. // only do autodetection if no explicit encoding is requested
  1676. if (encoding != encoding_auto) return encoding;
  1677. // try to guess encoding (based on XML specification, Appendix F.1)
  1678. const uint8_t* data = static_cast<const uint8_t*>(contents);
  1679. return guess_buffer_encoding(data, size);
  1680. }
  1681. PUGI_IMPL_FN bool get_mutable_buffer(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable)
  1682. {
  1683. size_t length = size / sizeof(char_t);
  1684. if (is_mutable)
  1685. {
  1686. out_buffer = static_cast<char_t*>(const_cast<void*>(contents));
  1687. out_length = length;
  1688. }
  1689. else
  1690. {
  1691. char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
  1692. if (!buffer) return false;
  1693. if (contents)
  1694. memcpy(buffer, contents, length * sizeof(char_t));
  1695. else
  1696. assert(length == 0);
  1697. buffer[length] = 0;
  1698. out_buffer = buffer;
  1699. out_length = length + 1;
  1700. }
  1701. return true;
  1702. }
  1703. #ifdef PUGIXML_WCHAR_MODE
  1704. PUGI_IMPL_FN bool need_endian_swap_utf(xml_encoding le, xml_encoding re)
  1705. {
  1706. return (le == encoding_utf16_be && re == encoding_utf16_le) || (le == encoding_utf16_le && re == encoding_utf16_be) ||
  1707. (le == encoding_utf32_be && re == encoding_utf32_le) || (le == encoding_utf32_le && re == encoding_utf32_be);
  1708. }
  1709. PUGI_IMPL_FN bool convert_buffer_endian_swap(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable)
  1710. {
  1711. const char_t* data = static_cast<const char_t*>(contents);
  1712. size_t length = size / sizeof(char_t);
  1713. if (is_mutable)
  1714. {
  1715. char_t* buffer = const_cast<char_t*>(data);
  1716. convert_wchar_endian_swap(buffer, data, length);
  1717. out_buffer = buffer;
  1718. out_length = length;
  1719. }
  1720. else
  1721. {
  1722. char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
  1723. if (!buffer) return false;
  1724. convert_wchar_endian_swap(buffer, data, length);
  1725. buffer[length] = 0;
  1726. out_buffer = buffer;
  1727. out_length = length + 1;
  1728. }
  1729. return true;
  1730. }
  1731. template <typename D> PUGI_IMPL_FN bool convert_buffer_generic(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, D)
  1732. {
  1733. const typename D::type* data = static_cast<const typename D::type*>(contents);
  1734. size_t data_length = size / sizeof(typename D::type);
  1735. // first pass: get length in wchar_t units
  1736. size_t length = D::process(data, data_length, 0, wchar_counter());
  1737. // allocate buffer of suitable length
  1738. char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
  1739. if (!buffer) return false;
  1740. // second pass: convert utf16 input to wchar_t
  1741. wchar_writer::value_type obegin = reinterpret_cast<wchar_writer::value_type>(buffer);
  1742. wchar_writer::value_type oend = D::process(data, data_length, obegin, wchar_writer());
  1743. assert(oend == obegin + length);
  1744. *oend = 0;
  1745. out_buffer = buffer;
  1746. out_length = length + 1;
  1747. return true;
  1748. }
  1749. PUGI_IMPL_FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable)
  1750. {
  1751. // get native encoding
  1752. xml_encoding wchar_encoding = get_wchar_encoding();
  1753. // fast path: no conversion required
  1754. if (encoding == wchar_encoding)
  1755. return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable);
  1756. // only endian-swapping is required
  1757. if (need_endian_swap_utf(encoding, wchar_encoding))
  1758. return convert_buffer_endian_swap(out_buffer, out_length, contents, size, is_mutable);
  1759. // source encoding is utf8
  1760. if (encoding == encoding_utf8)
  1761. return convert_buffer_generic(out_buffer, out_length, contents, size, utf8_decoder());
  1762. // source encoding is utf16
  1763. if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
  1764. {
  1765. xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
  1766. return (native_encoding == encoding) ?
  1767. convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder<opt_false>()) :
  1768. convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder<opt_true>());
  1769. }
  1770. // source encoding is utf32
  1771. if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
  1772. {
  1773. xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
  1774. return (native_encoding == encoding) ?
  1775. convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder<opt_false>()) :
  1776. convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder<opt_true>());
  1777. }
  1778. // source encoding is latin1
  1779. if (encoding == encoding_latin1)
  1780. return convert_buffer_generic(out_buffer, out_length, contents, size, latin1_decoder());
  1781. assert(false && "Invalid encoding"); // unreachable
  1782. return false;
  1783. }
  1784. #else
  1785. template <typename D> PUGI_IMPL_FN bool convert_buffer_generic(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, D)
  1786. {
  1787. const typename D::type* data = static_cast<const typename D::type*>(contents);
  1788. size_t data_length = size / sizeof(typename D::type);
  1789. // first pass: get length in utf8 units
  1790. size_t length = D::process(data, data_length, 0, utf8_counter());
  1791. // allocate buffer of suitable length
  1792. char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
  1793. if (!buffer) return false;
  1794. // second pass: convert utf16 input to utf8
  1795. uint8_t* obegin = reinterpret_cast<uint8_t*>(buffer);
  1796. uint8_t* oend = D::process(data, data_length, obegin, utf8_writer());
  1797. assert(oend == obegin + length);
  1798. *oend = 0;
  1799. out_buffer = buffer;
  1800. out_length = length + 1;
  1801. return true;
  1802. }
  1803. PUGI_IMPL_FN size_t get_latin1_7bit_prefix_length(const uint8_t* data, size_t size)
  1804. {
  1805. for (size_t i = 0; i < size; ++i)
  1806. if (data[i] > 127)
  1807. return i;
  1808. return size;
  1809. }
  1810. PUGI_IMPL_FN bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable)
  1811. {
  1812. const uint8_t* data = static_cast<const uint8_t*>(contents);
  1813. size_t data_length = size;
  1814. // get size of prefix that does not need utf8 conversion
  1815. size_t prefix_length = get_latin1_7bit_prefix_length(data, data_length);
  1816. assert(prefix_length <= data_length);
  1817. const uint8_t* postfix = data + prefix_length;
  1818. size_t postfix_length = data_length - prefix_length;
  1819. // if no conversion is needed, just return the original buffer
  1820. if (postfix_length == 0) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable);
  1821. // first pass: get length in utf8 units
  1822. size_t length = prefix_length + latin1_decoder::process(postfix, postfix_length, 0, utf8_counter());
  1823. // allocate buffer of suitable length
  1824. char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
  1825. if (!buffer) return false;
  1826. // second pass: convert latin1 input to utf8
  1827. memcpy(buffer, data, prefix_length);
  1828. uint8_t* obegin = reinterpret_cast<uint8_t*>(buffer);
  1829. uint8_t* oend = latin1_decoder::process(postfix, postfix_length, obegin + prefix_length, utf8_writer());
  1830. assert(oend == obegin + length);
  1831. *oend = 0;
  1832. out_buffer = buffer;
  1833. out_length = length + 1;
  1834. return true;
  1835. }
  1836. PUGI_IMPL_FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable)
  1837. {
  1838. // fast path: no conversion required
  1839. if (encoding == encoding_utf8)
  1840. return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable);
  1841. // source encoding is utf16
  1842. if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
  1843. {
  1844. xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
  1845. return (native_encoding == encoding) ?
  1846. convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder<opt_false>()) :
  1847. convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder<opt_true>());
  1848. }
  1849. // source encoding is utf32
  1850. if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
  1851. {
  1852. xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
  1853. return (native_encoding == encoding) ?
  1854. convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder<opt_false>()) :
  1855. convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder<opt_true>());
  1856. }
  1857. // source encoding is latin1
  1858. if (encoding == encoding_latin1)
  1859. return convert_buffer_latin1(out_buffer, out_length, contents, size, is_mutable);
  1860. assert(false && "Invalid encoding"); // unreachable
  1861. return false;
  1862. }
  1863. #endif
  1864. PUGI_IMPL_FN size_t as_utf8_begin(const wchar_t* str, size_t length)
  1865. {
  1866. // get length in utf8 characters
  1867. return wchar_decoder::process(str, length, 0, utf8_counter());
  1868. }
  1869. PUGI_IMPL_FN void as_utf8_end(char* buffer, size_t size, const wchar_t* str, size_t length)
  1870. {
  1871. // convert to utf8
  1872. uint8_t* begin = reinterpret_cast<uint8_t*>(buffer);
  1873. uint8_t* end = wchar_decoder::process(str, length, begin, utf8_writer());
  1874. assert(begin + size == end);
  1875. (void)!end;
  1876. (void)!size;
  1877. }
  1878. #ifndef PUGIXML_NO_STL
  1879. PUGI_IMPL_FN std::string as_utf8_impl(const wchar_t* str, size_t length)
  1880. {
  1881. // first pass: get length in utf8 characters
  1882. size_t size = as_utf8_begin(str, length);
  1883. // allocate resulting string
  1884. std::string result;
  1885. result.resize(size);
  1886. // second pass: convert to utf8
  1887. if (size > 0) as_utf8_end(&result[0], size, str, length);
  1888. return result;
  1889. }
  1890. PUGI_IMPL_FN std::basic_string<wchar_t> as_wide_impl(const char* str, size_t size)
  1891. {
  1892. const uint8_t* data = reinterpret_cast<const uint8_t*>(str);
  1893. // first pass: get length in wchar_t units
  1894. size_t length = utf8_decoder::process(data, size, 0, wchar_counter());
  1895. // allocate resulting string
  1896. std::basic_string<wchar_t> result;
  1897. result.resize(length);
  1898. // second pass: convert to wchar_t
  1899. if (length > 0)
  1900. {
  1901. wchar_writer::value_type begin = reinterpret_cast<wchar_writer::value_type>(&result[0]);
  1902. wchar_writer::value_type end = utf8_decoder::process(data, size, begin, wchar_writer());
  1903. assert(begin + length == end);
  1904. (void)!end;
  1905. }
  1906. return result;
  1907. }
  1908. #endif
  1909. template <typename Header>
  1910. inline bool strcpy_insitu_allow(size_t length, const Header& header, uintptr_t header_mask, char_t* target)
  1911. {
  1912. // never reuse shared memory
  1913. if (header & xml_memory_page_contents_shared_mask) return false;
  1914. size_t target_length = strlength(target);
  1915. // always reuse document buffer memory if possible
  1916. if ((header & header_mask) == 0) return target_length >= length;
  1917. // reuse heap memory if waste is not too great
  1918. const size_t reuse_threshold = 32;
  1919. return target_length >= length && (target_length < reuse_threshold || target_length - length < target_length / 2);
  1920. }
  1921. template <typename String, typename Header>
  1922. PUGI_IMPL_FN bool strcpy_insitu(String& dest, Header& header, uintptr_t header_mask, const char_t* source, size_t source_length)
  1923. {
  1924. assert((header & header_mask) == 0 || dest); // header bit indicates whether dest was previously allocated
  1925. if (source_length == 0)
  1926. {
  1927. // empty string and null pointer are equivalent, so just deallocate old memory
  1928. xml_allocator* alloc = PUGI_IMPL_GETPAGE_IMPL(header)->allocator;
  1929. if (header & header_mask) alloc->deallocate_string(dest);
  1930. // mark the string as not allocated
  1931. dest = NULL;
  1932. header &= ~header_mask;
  1933. return true;
  1934. }
  1935. else if (dest && strcpy_insitu_allow(source_length, header, header_mask, dest))
  1936. {
  1937. // we can reuse old buffer, so just copy the new data (including zero terminator)
  1938. memcpy(dest, source, source_length * sizeof(char_t));
  1939. dest[source_length] = 0;
  1940. return true;
  1941. }
  1942. else
  1943. {
  1944. xml_allocator* alloc = PUGI_IMPL_GETPAGE_IMPL(header)->allocator;
  1945. if (!alloc->reserve()) return false;
  1946. // allocate new buffer
  1947. char_t* buf = alloc->allocate_string(source_length + 1);
  1948. if (!buf) return false;
  1949. // copy the string (including zero terminator)
  1950. memcpy(buf, source, source_length * sizeof(char_t));
  1951. buf[source_length] = 0;
  1952. // deallocate old buffer (*after* the above to protect against overlapping memory and/or allocation failures)
  1953. if (header & header_mask) alloc->deallocate_string(dest);
  1954. // the string is now allocated, so set the flag
  1955. dest = buf;
  1956. header |= header_mask;
  1957. return true;
  1958. }
  1959. }
  1960. struct gap
  1961. {
  1962. char_t* end;
  1963. size_t size;
  1964. gap(): end(NULL), size(0)
  1965. {
  1966. }
  1967. // Push new gap, move s count bytes further (skipping the gap).
  1968. // Collapse previous gap.
  1969. void push(char_t*& s, size_t count)
  1970. {
  1971. if (end) // there was a gap already; collapse it
  1972. {
  1973. // Move [old_gap_end, new_gap_start) to [old_gap_start, ...)
  1974. assert(s >= end);
  1975. memmove(end - size, end, (s - end) * sizeof(char_t));
  1976. }
  1977. s += count; // end of current gap
  1978. // "merge" two gaps
  1979. end = s;
  1980. size += count;
  1981. }
  1982. // Collapse all gaps, return past-the-end pointer
  1983. char_t* flush(char_t* s)
  1984. {
  1985. if (end)
  1986. {
  1987. // Move [old_gap_end, current_pos) to [old_gap_start, ...)
  1988. assert(s >= end);
  1989. memmove(end - size, end, (s - end) * sizeof(char_t));
  1990. return s - size;
  1991. }
  1992. else return s;
  1993. }
  1994. };
  1995. PUGI_IMPL_FN char_t* strconv_escape(char_t* s, gap& g)
  1996. {
  1997. char_t* stre = s + 1;
  1998. switch (*stre)
  1999. {
  2000. case '#': // &#...
  2001. {
  2002. unsigned int ucsc = 0;
  2003. if (stre[1] == 'x') // &#x... (hex code)
  2004. {
  2005. stre += 2;
  2006. char_t ch = *stre;
  2007. if (ch == ';') return stre;
  2008. for (;;)
  2009. {
  2010. if (static_cast<unsigned int>(ch - '0') <= 9)
  2011. ucsc = 16 * ucsc + (ch - '0');
  2012. else if (static_cast<unsigned int>((ch | ' ') - 'a') <= 5)
  2013. ucsc = 16 * ucsc + ((ch | ' ') - 'a' + 10);
  2014. else if (ch == ';')
  2015. break;
  2016. else // cancel
  2017. return stre;
  2018. ch = *++stre;
  2019. }
  2020. ++stre;
  2021. }
  2022. else // &#... (dec code)
  2023. {
  2024. char_t ch = *++stre;
  2025. if (ch == ';') return stre;
  2026. for (;;)
  2027. {
  2028. if (static_cast<unsigned int>(ch - '0') <= 9)
  2029. ucsc = 10 * ucsc + (ch - '0');
  2030. else if (ch == ';')
  2031. break;
  2032. else // cancel
  2033. return stre;
  2034. ch = *++stre;
  2035. }
  2036. ++stre;
  2037. }
  2038. #ifdef PUGIXML_WCHAR_MODE
  2039. s = reinterpret_cast<char_t*>(wchar_writer::any(reinterpret_cast<wchar_writer::value_type>(s), ucsc));
  2040. #else
  2041. s = reinterpret_cast<char_t*>(utf8_writer::any(reinterpret_cast<uint8_t*>(s), ucsc));
  2042. #endif
  2043. g.push(s, stre - s);
  2044. return stre;
  2045. }
  2046. case 'a': // &a
  2047. {
  2048. ++stre;
  2049. if (*stre == 'm') // &am
  2050. {
  2051. if (*++stre == 'p' && *++stre == ';') // &amp;
  2052. {
  2053. *s++ = '&';
  2054. ++stre;
  2055. g.push(s, stre - s);
  2056. return stre;
  2057. }
  2058. }
  2059. else if (*stre == 'p') // &ap
  2060. {
  2061. if (*++stre == 'o' && *++stre == 's' && *++stre == ';') // &apos;
  2062. {
  2063. *s++ = '\'';
  2064. ++stre;
  2065. g.push(s, stre - s);
  2066. return stre;
  2067. }
  2068. }
  2069. break;
  2070. }
  2071. case 'g': // &g
  2072. {
  2073. if (*++stre == 't' && *++stre == ';') // &gt;
  2074. {
  2075. *s++ = '>';
  2076. ++stre;
  2077. g.push(s, stre - s);
  2078. return stre;
  2079. }
  2080. break;
  2081. }
  2082. case 'l': // &l
  2083. {
  2084. if (*++stre == 't' && *++stre == ';') // &lt;
  2085. {
  2086. *s++ = '<';
  2087. ++stre;
  2088. g.push(s, stre - s);
  2089. return stre;
  2090. }
  2091. break;
  2092. }
  2093. case 'q': // &q
  2094. {
  2095. if (*++stre == 'u' && *++stre == 'o' && *++stre == 't' && *++stre == ';') // &quot;
  2096. {
  2097. *s++ = '"';
  2098. ++stre;
  2099. g.push(s, stre - s);
  2100. return stre;
  2101. }
  2102. break;
  2103. }
  2104. default:
  2105. break;
  2106. }
  2107. return stre;
  2108. }
  2109. // Parser utilities
  2110. #define PUGI_IMPL_ENDSWITH(c, e) ((c) == (e) || ((c) == 0 && endch == (e)))
  2111. #define PUGI_IMPL_SKIPWS() { while (PUGI_IMPL_IS_CHARTYPE(*s, ct_space)) ++s; }
  2112. #define PUGI_IMPL_OPTSET(OPT) ( optmsk & (OPT) )
  2113. #define PUGI_IMPL_PUSHNODE(TYPE) { cursor = append_new_node(cursor, *alloc, TYPE); if (!cursor) PUGI_IMPL_THROW_ERROR(status_out_of_memory, s); }
  2114. #define PUGI_IMPL_POPNODE() { cursor = cursor->parent; }
  2115. #define PUGI_IMPL_SCANFOR(X) { while (*s != 0 && !(X)) ++s; }
  2116. #define PUGI_IMPL_SCANWHILE(X) { while (X) ++s; }
  2117. #define PUGI_IMPL_SCANWHILE_UNROLL(X) { for (;;) { char_t ss = s[0]; if (PUGI_IMPL_UNLIKELY(!(X))) { break; } ss = s[1]; if (PUGI_IMPL_UNLIKELY(!(X))) { s += 1; break; } ss = s[2]; if (PUGI_IMPL_UNLIKELY(!(X))) { s += 2; break; } ss = s[3]; if (PUGI_IMPL_UNLIKELY(!(X))) { s += 3; break; } s += 4; } }
  2118. #define PUGI_IMPL_ENDSEG() { ch = *s; *s = 0; ++s; }
  2119. #define PUGI_IMPL_THROW_ERROR(err, m) return error_offset = m, error_status = err, static_cast<char_t*>(NULL)
  2120. #define PUGI_IMPL_CHECK_ERROR(err, m) { if (*s == 0) PUGI_IMPL_THROW_ERROR(err, m); }
  2121. PUGI_IMPL_FN char_t* strconv_comment(char_t* s, char_t endch)
  2122. {
  2123. gap g;
  2124. while (true)
  2125. {
  2126. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPE(ss, ct_parse_comment));
  2127. if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair
  2128. {
  2129. *s++ = '\n'; // replace first one with 0x0a
  2130. if (*s == '\n') g.push(s, 1);
  2131. }
  2132. else if (s[0] == '-' && s[1] == '-' && PUGI_IMPL_ENDSWITH(s[2], '>')) // comment ends here
  2133. {
  2134. *g.flush(s) = 0;
  2135. return s + (s[2] == '>' ? 3 : 2);
  2136. }
  2137. else if (*s == 0)
  2138. {
  2139. return NULL;
  2140. }
  2141. else ++s;
  2142. }
  2143. }
  2144. PUGI_IMPL_FN char_t* strconv_cdata(char_t* s, char_t endch)
  2145. {
  2146. gap g;
  2147. while (true)
  2148. {
  2149. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPE(ss, ct_parse_cdata));
  2150. if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair
  2151. {
  2152. *s++ = '\n'; // replace first one with 0x0a
  2153. if (*s == '\n') g.push(s, 1);
  2154. }
  2155. else if (s[0] == ']' && s[1] == ']' && PUGI_IMPL_ENDSWITH(s[2], '>')) // CDATA ends here
  2156. {
  2157. *g.flush(s) = 0;
  2158. return s + 1;
  2159. }
  2160. else if (*s == 0)
  2161. {
  2162. return NULL;
  2163. }
  2164. else ++s;
  2165. }
  2166. }
  2167. typedef char_t* (*strconv_pcdata_t)(char_t*);
  2168. template <typename opt_trim, typename opt_eol, typename opt_escape> struct strconv_pcdata_impl
  2169. {
  2170. static char_t* parse(char_t* s)
  2171. {
  2172. gap g;
  2173. char_t* begin = s;
  2174. while (true)
  2175. {
  2176. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPE(ss, ct_parse_pcdata));
  2177. if (*s == '<') // PCDATA ends here
  2178. {
  2179. char_t* end = g.flush(s);
  2180. if (opt_trim::value)
  2181. while (end > begin && PUGI_IMPL_IS_CHARTYPE(end[-1], ct_space))
  2182. --end;
  2183. *end = 0;
  2184. return s + 1;
  2185. }
  2186. else if (opt_eol::value && *s == '\r') // Either a single 0x0d or 0x0d 0x0a pair
  2187. {
  2188. *s++ = '\n'; // replace first one with 0x0a
  2189. if (*s == '\n') g.push(s, 1);
  2190. }
  2191. else if (opt_escape::value && *s == '&')
  2192. {
  2193. s = strconv_escape(s, g);
  2194. }
  2195. else if (*s == 0)
  2196. {
  2197. char_t* end = g.flush(s);
  2198. if (opt_trim::value)
  2199. while (end > begin && PUGI_IMPL_IS_CHARTYPE(end[-1], ct_space))
  2200. --end;
  2201. *end = 0;
  2202. return s;
  2203. }
  2204. else ++s;
  2205. }
  2206. }
  2207. };
  2208. PUGI_IMPL_FN strconv_pcdata_t get_strconv_pcdata(unsigned int optmask)
  2209. {
  2210. PUGI_IMPL_STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20 && parse_trim_pcdata == 0x0800);
  2211. switch (((optmask >> 4) & 3) | ((optmask >> 9) & 4)) // get bitmask for flags (trim eol escapes); this simultaneously checks 3 options from assertion above
  2212. {
  2213. case 0: return strconv_pcdata_impl<opt_false, opt_false, opt_false>::parse;
  2214. case 1: return strconv_pcdata_impl<opt_false, opt_false, opt_true>::parse;
  2215. case 2: return strconv_pcdata_impl<opt_false, opt_true, opt_false>::parse;
  2216. case 3: return strconv_pcdata_impl<opt_false, opt_true, opt_true>::parse;
  2217. case 4: return strconv_pcdata_impl<opt_true, opt_false, opt_false>::parse;
  2218. case 5: return strconv_pcdata_impl<opt_true, opt_false, opt_true>::parse;
  2219. case 6: return strconv_pcdata_impl<opt_true, opt_true, opt_false>::parse;
  2220. case 7: return strconv_pcdata_impl<opt_true, opt_true, opt_true>::parse;
  2221. default: assert(false); return NULL; // unreachable
  2222. }
  2223. }
  2224. typedef char_t* (*strconv_attribute_t)(char_t*, char_t);
  2225. template <typename opt_escape> struct strconv_attribute_impl
  2226. {
  2227. static char_t* parse_wnorm(char_t* s, char_t end_quote)
  2228. {
  2229. gap g;
  2230. // trim leading whitespaces
  2231. if (PUGI_IMPL_IS_CHARTYPE(*s, ct_space))
  2232. {
  2233. char_t* str = s;
  2234. do ++str;
  2235. while (PUGI_IMPL_IS_CHARTYPE(*str, ct_space));
  2236. g.push(s, str - s);
  2237. }
  2238. while (true)
  2239. {
  2240. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPE(ss, ct_parse_attr_ws | ct_space));
  2241. if (*s == end_quote)
  2242. {
  2243. char_t* str = g.flush(s);
  2244. do *str-- = 0;
  2245. while (PUGI_IMPL_IS_CHARTYPE(*str, ct_space));
  2246. return s + 1;
  2247. }
  2248. else if (PUGI_IMPL_IS_CHARTYPE(*s, ct_space))
  2249. {
  2250. *s++ = ' ';
  2251. if (PUGI_IMPL_IS_CHARTYPE(*s, ct_space))
  2252. {
  2253. char_t* str = s + 1;
  2254. while (PUGI_IMPL_IS_CHARTYPE(*str, ct_space)) ++str;
  2255. g.push(s, str - s);
  2256. }
  2257. }
  2258. else if (opt_escape::value && *s == '&')
  2259. {
  2260. s = strconv_escape(s, g);
  2261. }
  2262. else if (!*s)
  2263. {
  2264. return NULL;
  2265. }
  2266. else ++s;
  2267. }
  2268. }
  2269. static char_t* parse_wconv(char_t* s, char_t end_quote)
  2270. {
  2271. gap g;
  2272. while (true)
  2273. {
  2274. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPE(ss, ct_parse_attr_ws));
  2275. if (*s == end_quote)
  2276. {
  2277. *g.flush(s) = 0;
  2278. return s + 1;
  2279. }
  2280. else if (PUGI_IMPL_IS_CHARTYPE(*s, ct_space))
  2281. {
  2282. if (*s == '\r')
  2283. {
  2284. *s++ = ' ';
  2285. if (*s == '\n') g.push(s, 1);
  2286. }
  2287. else *s++ = ' ';
  2288. }
  2289. else if (opt_escape::value && *s == '&')
  2290. {
  2291. s = strconv_escape(s, g);
  2292. }
  2293. else if (!*s)
  2294. {
  2295. return NULL;
  2296. }
  2297. else ++s;
  2298. }
  2299. }
  2300. static char_t* parse_eol(char_t* s, char_t end_quote)
  2301. {
  2302. gap g;
  2303. while (true)
  2304. {
  2305. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPE(ss, ct_parse_attr));
  2306. if (*s == end_quote)
  2307. {
  2308. *g.flush(s) = 0;
  2309. return s + 1;
  2310. }
  2311. else if (*s == '\r')
  2312. {
  2313. *s++ = '\n';
  2314. if (*s == '\n') g.push(s, 1);
  2315. }
  2316. else if (opt_escape::value && *s == '&')
  2317. {
  2318. s = strconv_escape(s, g);
  2319. }
  2320. else if (!*s)
  2321. {
  2322. return NULL;
  2323. }
  2324. else ++s;
  2325. }
  2326. }
  2327. static char_t* parse_simple(char_t* s, char_t end_quote)
  2328. {
  2329. gap g;
  2330. while (true)
  2331. {
  2332. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPE(ss, ct_parse_attr));
  2333. if (*s == end_quote)
  2334. {
  2335. *g.flush(s) = 0;
  2336. return s + 1;
  2337. }
  2338. else if (opt_escape::value && *s == '&')
  2339. {
  2340. s = strconv_escape(s, g);
  2341. }
  2342. else if (!*s)
  2343. {
  2344. return NULL;
  2345. }
  2346. else ++s;
  2347. }
  2348. }
  2349. };
  2350. PUGI_IMPL_FN strconv_attribute_t get_strconv_attribute(unsigned int optmask)
  2351. {
  2352. PUGI_IMPL_STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20 && parse_wconv_attribute == 0x40 && parse_wnorm_attribute == 0x80);
  2353. switch ((optmask >> 4) & 15) // get bitmask for flags (wnorm wconv eol escapes); this simultaneously checks 4 options from assertion above
  2354. {
  2355. case 0: return strconv_attribute_impl<opt_false>::parse_simple;
  2356. case 1: return strconv_attribute_impl<opt_true>::parse_simple;
  2357. case 2: return strconv_attribute_impl<opt_false>::parse_eol;
  2358. case 3: return strconv_attribute_impl<opt_true>::parse_eol;
  2359. case 4: return strconv_attribute_impl<opt_false>::parse_wconv;
  2360. case 5: return strconv_attribute_impl<opt_true>::parse_wconv;
  2361. case 6: return strconv_attribute_impl<opt_false>::parse_wconv;
  2362. case 7: return strconv_attribute_impl<opt_true>::parse_wconv;
  2363. case 8: return strconv_attribute_impl<opt_false>::parse_wnorm;
  2364. case 9: return strconv_attribute_impl<opt_true>::parse_wnorm;
  2365. case 10: return strconv_attribute_impl<opt_false>::parse_wnorm;
  2366. case 11: return strconv_attribute_impl<opt_true>::parse_wnorm;
  2367. case 12: return strconv_attribute_impl<opt_false>::parse_wnorm;
  2368. case 13: return strconv_attribute_impl<opt_true>::parse_wnorm;
  2369. case 14: return strconv_attribute_impl<opt_false>::parse_wnorm;
  2370. case 15: return strconv_attribute_impl<opt_true>::parse_wnorm;
  2371. default: assert(false); return NULL; // unreachable
  2372. }
  2373. }
  2374. inline xml_parse_result make_parse_result(xml_parse_status status, ptrdiff_t offset = 0)
  2375. {
  2376. xml_parse_result result;
  2377. result.status = status;
  2378. result.offset = offset;
  2379. return result;
  2380. }
  2381. struct xml_parser
  2382. {
  2383. xml_allocator* alloc;
  2384. char_t* error_offset;
  2385. xml_parse_status error_status;
  2386. xml_parser(xml_allocator* alloc_): alloc(alloc_), error_offset(NULL), error_status(status_ok)
  2387. {
  2388. }
  2389. // DOCTYPE consists of nested sections of the following possible types:
  2390. // <!-- ... -->, <? ... ?>, "...", '...'
  2391. // <![...]]>
  2392. // <!...>
  2393. // First group can not contain nested groups
  2394. // Second group can contain nested groups of the same type
  2395. // Third group can contain all other groups
  2396. char_t* parse_doctype_primitive(char_t* s)
  2397. {
  2398. if (*s == '"' || *s == '\'')
  2399. {
  2400. // quoted string
  2401. char_t ch = *s++;
  2402. PUGI_IMPL_SCANFOR(*s == ch);
  2403. if (!*s) PUGI_IMPL_THROW_ERROR(status_bad_doctype, s);
  2404. s++;
  2405. }
  2406. else if (s[0] == '<' && s[1] == '?')
  2407. {
  2408. // <? ... ?>
  2409. s += 2;
  2410. PUGI_IMPL_SCANFOR(s[0] == '?' && s[1] == '>'); // no need for ENDSWITH because ?> can't terminate proper doctype
  2411. if (!*s) PUGI_IMPL_THROW_ERROR(status_bad_doctype, s);
  2412. s += 2;
  2413. }
  2414. else if (s[0] == '<' && s[1] == '!' && s[2] == '-' && s[3] == '-')
  2415. {
  2416. s += 4;
  2417. PUGI_IMPL_SCANFOR(s[0] == '-' && s[1] == '-' && s[2] == '>'); // no need for ENDSWITH because --> can't terminate proper doctype
  2418. if (!*s) PUGI_IMPL_THROW_ERROR(status_bad_doctype, s);
  2419. s += 3;
  2420. }
  2421. else PUGI_IMPL_THROW_ERROR(status_bad_doctype, s);
  2422. return s;
  2423. }
  2424. char_t* parse_doctype_ignore(char_t* s)
  2425. {
  2426. size_t depth = 0;
  2427. assert(s[0] == '<' && s[1] == '!' && s[2] == '[');
  2428. s += 3;
  2429. while (*s)
  2430. {
  2431. if (s[0] == '<' && s[1] == '!' && s[2] == '[')
  2432. {
  2433. // nested ignore section
  2434. s += 3;
  2435. depth++;
  2436. }
  2437. else if (s[0] == ']' && s[1] == ']' && s[2] == '>')
  2438. {
  2439. // ignore section end
  2440. s += 3;
  2441. if (depth == 0)
  2442. return s;
  2443. depth--;
  2444. }
  2445. else s++;
  2446. }
  2447. PUGI_IMPL_THROW_ERROR(status_bad_doctype, s);
  2448. }
  2449. char_t* parse_doctype_group(char_t* s, char_t endch)
  2450. {
  2451. size_t depth = 0;
  2452. assert((s[0] == '<' || s[0] == 0) && s[1] == '!');
  2453. s += 2;
  2454. while (*s)
  2455. {
  2456. if (s[0] == '<' && s[1] == '!' && s[2] != '-')
  2457. {
  2458. if (s[2] == '[')
  2459. {
  2460. // ignore
  2461. s = parse_doctype_ignore(s);
  2462. if (!s) return s;
  2463. }
  2464. else
  2465. {
  2466. // some control group
  2467. s += 2;
  2468. depth++;
  2469. }
  2470. }
  2471. else if (s[0] == '<' || s[0] == '"' || s[0] == '\'')
  2472. {
  2473. // unknown tag (forbidden), or some primitive group
  2474. s = parse_doctype_primitive(s);
  2475. if (!s) return s;
  2476. }
  2477. else if (*s == '>')
  2478. {
  2479. if (depth == 0)
  2480. return s;
  2481. depth--;
  2482. s++;
  2483. }
  2484. else s++;
  2485. }
  2486. if (depth != 0 || endch != '>') PUGI_IMPL_THROW_ERROR(status_bad_doctype, s);
  2487. return s;
  2488. }
  2489. char_t* parse_exclamation(char_t* s, xml_node_struct* cursor, unsigned int optmsk, char_t endch)
  2490. {
  2491. // parse node contents, starting with exclamation mark
  2492. ++s;
  2493. if (*s == '-') // '<!-...'
  2494. {
  2495. ++s;
  2496. if (*s == '-') // '<!--...'
  2497. {
  2498. ++s;
  2499. if (PUGI_IMPL_OPTSET(parse_comments))
  2500. {
  2501. PUGI_IMPL_PUSHNODE(node_comment); // Append a new node on the tree.
  2502. cursor->value = s; // Save the offset.
  2503. }
  2504. if (PUGI_IMPL_OPTSET(parse_eol) && PUGI_IMPL_OPTSET(parse_comments))
  2505. {
  2506. s = strconv_comment(s, endch);
  2507. if (!s) PUGI_IMPL_THROW_ERROR(status_bad_comment, cursor->value);
  2508. }
  2509. else
  2510. {
  2511. // Scan for terminating '-->'.
  2512. PUGI_IMPL_SCANFOR(s[0] == '-' && s[1] == '-' && PUGI_IMPL_ENDSWITH(s[2], '>'));
  2513. PUGI_IMPL_CHECK_ERROR(status_bad_comment, s);
  2514. if (PUGI_IMPL_OPTSET(parse_comments))
  2515. *s = 0; // Zero-terminate this segment at the first terminating '-'.
  2516. s += (s[2] == '>' ? 3 : 2); // Step over the '\0->'.
  2517. }
  2518. }
  2519. else PUGI_IMPL_THROW_ERROR(status_bad_comment, s);
  2520. }
  2521. else if (*s == '[')
  2522. {
  2523. // '<![CDATA[...'
  2524. if (*++s=='C' && *++s=='D' && *++s=='A' && *++s=='T' && *++s=='A' && *++s == '[')
  2525. {
  2526. ++s;
  2527. if (PUGI_IMPL_OPTSET(parse_cdata))
  2528. {
  2529. PUGI_IMPL_PUSHNODE(node_cdata); // Append a new node on the tree.
  2530. cursor->value = s; // Save the offset.
  2531. if (PUGI_IMPL_OPTSET(parse_eol))
  2532. {
  2533. s = strconv_cdata(s, endch);
  2534. if (!s) PUGI_IMPL_THROW_ERROR(status_bad_cdata, cursor->value);
  2535. }
  2536. else
  2537. {
  2538. // Scan for terminating ']]>'.
  2539. PUGI_IMPL_SCANFOR(s[0] == ']' && s[1] == ']' && PUGI_IMPL_ENDSWITH(s[2], '>'));
  2540. PUGI_IMPL_CHECK_ERROR(status_bad_cdata, s);
  2541. *s++ = 0; // Zero-terminate this segment.
  2542. }
  2543. }
  2544. else // Flagged for discard, but we still have to scan for the terminator.
  2545. {
  2546. // Scan for terminating ']]>'.
  2547. PUGI_IMPL_SCANFOR(s[0] == ']' && s[1] == ']' && PUGI_IMPL_ENDSWITH(s[2], '>'));
  2548. PUGI_IMPL_CHECK_ERROR(status_bad_cdata, s);
  2549. ++s;
  2550. }
  2551. s += (s[1] == '>' ? 2 : 1); // Step over the last ']>'.
  2552. }
  2553. else PUGI_IMPL_THROW_ERROR(status_bad_cdata, s);
  2554. }
  2555. else if (s[0] == 'D' && s[1] == 'O' && s[2] == 'C' && s[3] == 'T' && s[4] == 'Y' && s[5] == 'P' && PUGI_IMPL_ENDSWITH(s[6], 'E'))
  2556. {
  2557. s -= 2;
  2558. if (cursor->parent) PUGI_IMPL_THROW_ERROR(status_bad_doctype, s);
  2559. char_t* mark = s + 9;
  2560. s = parse_doctype_group(s, endch);
  2561. if (!s) return s;
  2562. assert((*s == 0 && endch == '>') || *s == '>');
  2563. if (*s) *s++ = 0;
  2564. if (PUGI_IMPL_OPTSET(parse_doctype))
  2565. {
  2566. while (PUGI_IMPL_IS_CHARTYPE(*mark, ct_space)) ++mark;
  2567. PUGI_IMPL_PUSHNODE(node_doctype);
  2568. cursor->value = mark;
  2569. }
  2570. }
  2571. else if (*s == 0 && endch == '-') PUGI_IMPL_THROW_ERROR(status_bad_comment, s);
  2572. else if (*s == 0 && endch == '[') PUGI_IMPL_THROW_ERROR(status_bad_cdata, s);
  2573. else PUGI_IMPL_THROW_ERROR(status_unrecognized_tag, s);
  2574. return s;
  2575. }
  2576. char_t* parse_question(char_t* s, xml_node_struct*& ref_cursor, unsigned int optmsk, char_t endch)
  2577. {
  2578. // load into registers
  2579. xml_node_struct* cursor = ref_cursor;
  2580. char_t ch = 0;
  2581. // parse node contents, starting with question mark
  2582. ++s;
  2583. // read PI target
  2584. char_t* target = s;
  2585. if (!PUGI_IMPL_IS_CHARTYPE(*s, ct_start_symbol)) PUGI_IMPL_THROW_ERROR(status_bad_pi, s);
  2586. PUGI_IMPL_SCANWHILE(PUGI_IMPL_IS_CHARTYPE(*s, ct_symbol));
  2587. PUGI_IMPL_CHECK_ERROR(status_bad_pi, s);
  2588. // determine node type; stricmp / strcasecmp is not portable
  2589. bool declaration = (target[0] | ' ') == 'x' && (target[1] | ' ') == 'm' && (target[2] | ' ') == 'l' && target + 3 == s;
  2590. if (declaration ? PUGI_IMPL_OPTSET(parse_declaration) : PUGI_IMPL_OPTSET(parse_pi))
  2591. {
  2592. if (declaration)
  2593. {
  2594. // disallow non top-level declarations
  2595. if (cursor->parent) PUGI_IMPL_THROW_ERROR(status_bad_pi, s);
  2596. PUGI_IMPL_PUSHNODE(node_declaration);
  2597. }
  2598. else
  2599. {
  2600. PUGI_IMPL_PUSHNODE(node_pi);
  2601. }
  2602. cursor->name = target;
  2603. PUGI_IMPL_ENDSEG();
  2604. // parse value/attributes
  2605. if (ch == '?')
  2606. {
  2607. // empty node
  2608. if (!PUGI_IMPL_ENDSWITH(*s, '>')) PUGI_IMPL_THROW_ERROR(status_bad_pi, s);
  2609. s += (*s == '>');
  2610. PUGI_IMPL_POPNODE();
  2611. }
  2612. else if (PUGI_IMPL_IS_CHARTYPE(ch, ct_space))
  2613. {
  2614. PUGI_IMPL_SKIPWS();
  2615. // scan for tag end
  2616. char_t* value = s;
  2617. PUGI_IMPL_SCANFOR(s[0] == '?' && PUGI_IMPL_ENDSWITH(s[1], '>'));
  2618. PUGI_IMPL_CHECK_ERROR(status_bad_pi, s);
  2619. if (declaration)
  2620. {
  2621. // replace ending ? with / so that 'element' terminates properly
  2622. *s = '/';
  2623. // we exit from this function with cursor at node_declaration, which is a signal to parse() to go to LOC_ATTRIBUTES
  2624. s = value;
  2625. }
  2626. else
  2627. {
  2628. // store value and step over >
  2629. cursor->value = value;
  2630. PUGI_IMPL_POPNODE();
  2631. PUGI_IMPL_ENDSEG();
  2632. s += (*s == '>');
  2633. }
  2634. }
  2635. else PUGI_IMPL_THROW_ERROR(status_bad_pi, s);
  2636. }
  2637. else
  2638. {
  2639. // scan for tag end
  2640. PUGI_IMPL_SCANFOR(s[0] == '?' && PUGI_IMPL_ENDSWITH(s[1], '>'));
  2641. PUGI_IMPL_CHECK_ERROR(status_bad_pi, s);
  2642. s += (s[1] == '>' ? 2 : 1);
  2643. }
  2644. // store from registers
  2645. ref_cursor = cursor;
  2646. return s;
  2647. }
  2648. char_t* parse_tree(char_t* s, xml_node_struct* root, unsigned int optmsk, char_t endch)
  2649. {
  2650. strconv_attribute_t strconv_attribute = get_strconv_attribute(optmsk);
  2651. strconv_pcdata_t strconv_pcdata = get_strconv_pcdata(optmsk);
  2652. char_t ch = 0;
  2653. xml_node_struct* cursor = root;
  2654. char_t* mark = s;
  2655. char_t* merged_pcdata = s;
  2656. while (*s != 0)
  2657. {
  2658. if (*s == '<')
  2659. {
  2660. ++s;
  2661. LOC_TAG:
  2662. if (PUGI_IMPL_IS_CHARTYPE(*s, ct_start_symbol)) // '<#...'
  2663. {
  2664. PUGI_IMPL_PUSHNODE(node_element); // Append a new node to the tree.
  2665. cursor->name = s;
  2666. PUGI_IMPL_SCANWHILE_UNROLL(PUGI_IMPL_IS_CHARTYPE(ss, ct_symbol)); // Scan for a terminator.
  2667. PUGI_IMPL_ENDSEG(); // Save char in 'ch', terminate & step over.
  2668. if (ch == '>')
  2669. {
  2670. // end of tag
  2671. }
  2672. else if (PUGI_IMPL_IS_CHARTYPE(ch, ct_space))
  2673. {
  2674. LOC_ATTRIBUTES:
  2675. while (true)
  2676. {
  2677. PUGI_IMPL_SKIPWS(); // Eat any whitespace.
  2678. if (PUGI_IMPL_IS_CHARTYPE(*s, ct_start_symbol)) // <... #...
  2679. {
  2680. xml_attribute_struct* a = append_new_attribute(cursor, *alloc); // Make space for this attribute.
  2681. if (!a) PUGI_IMPL_THROW_ERROR(status_out_of_memory, s);
  2682. a->name = s; // Save the offset.
  2683. PUGI_IMPL_SCANWHILE_UNROLL(PUGI_IMPL_IS_CHARTYPE(ss, ct_symbol)); // Scan for a terminator.
  2684. PUGI_IMPL_ENDSEG(); // Save char in 'ch', terminate & step over.
  2685. if (PUGI_IMPL_IS_CHARTYPE(ch, ct_space))
  2686. {
  2687. PUGI_IMPL_SKIPWS(); // Eat any whitespace.
  2688. ch = *s;
  2689. ++s;
  2690. }
  2691. if (ch == '=') // '<... #=...'
  2692. {
  2693. PUGI_IMPL_SKIPWS(); // Eat any whitespace.
  2694. if (*s == '"' || *s == '\'') // '<... #="...'
  2695. {
  2696. ch = *s; // Save quote char to avoid breaking on "''" -or- '""'.
  2697. ++s; // Step over the quote.
  2698. a->value = s; // Save the offset.
  2699. s = strconv_attribute(s, ch);
  2700. if (!s) PUGI_IMPL_THROW_ERROR(status_bad_attribute, a->value);
  2701. // After this line the loop continues from the start;
  2702. // Whitespaces, / and > are ok, symbols and EOF are wrong,
  2703. // everything else will be detected
  2704. if (PUGI_IMPL_IS_CHARTYPE(*s, ct_start_symbol)) PUGI_IMPL_THROW_ERROR(status_bad_attribute, s);
  2705. }
  2706. else PUGI_IMPL_THROW_ERROR(status_bad_attribute, s);
  2707. }
  2708. else PUGI_IMPL_THROW_ERROR(status_bad_attribute, s);
  2709. }
  2710. else if (*s == '/')
  2711. {
  2712. ++s;
  2713. if (*s == '>')
  2714. {
  2715. PUGI_IMPL_POPNODE();
  2716. s++;
  2717. break;
  2718. }
  2719. else if (*s == 0 && endch == '>')
  2720. {
  2721. PUGI_IMPL_POPNODE();
  2722. break;
  2723. }
  2724. else PUGI_IMPL_THROW_ERROR(status_bad_start_element, s);
  2725. }
  2726. else if (*s == '>')
  2727. {
  2728. ++s;
  2729. break;
  2730. }
  2731. else if (*s == 0 && endch == '>')
  2732. {
  2733. break;
  2734. }
  2735. else PUGI_IMPL_THROW_ERROR(status_bad_start_element, s);
  2736. }
  2737. // !!!
  2738. }
  2739. else if (ch == '/') // '<#.../'
  2740. {
  2741. if (!PUGI_IMPL_ENDSWITH(*s, '>')) PUGI_IMPL_THROW_ERROR(status_bad_start_element, s);
  2742. PUGI_IMPL_POPNODE(); // Pop.
  2743. s += (*s == '>');
  2744. }
  2745. else if (ch == 0)
  2746. {
  2747. // we stepped over null terminator, backtrack & handle closing tag
  2748. --s;
  2749. if (endch != '>') PUGI_IMPL_THROW_ERROR(status_bad_start_element, s);
  2750. }
  2751. else PUGI_IMPL_THROW_ERROR(status_bad_start_element, s);
  2752. }
  2753. else if (*s == '/')
  2754. {
  2755. ++s;
  2756. mark = s;
  2757. char_t* name = cursor->name;
  2758. if (!name) PUGI_IMPL_THROW_ERROR(status_end_element_mismatch, mark);
  2759. while (PUGI_IMPL_IS_CHARTYPE(*s, ct_symbol))
  2760. {
  2761. if (*s++ != *name++) PUGI_IMPL_THROW_ERROR(status_end_element_mismatch, mark);
  2762. }
  2763. if (*name)
  2764. {
  2765. if (*s == 0 && name[0] == endch && name[1] == 0) PUGI_IMPL_THROW_ERROR(status_bad_end_element, s);
  2766. else PUGI_IMPL_THROW_ERROR(status_end_element_mismatch, mark);
  2767. }
  2768. PUGI_IMPL_POPNODE(); // Pop.
  2769. PUGI_IMPL_SKIPWS();
  2770. if (*s == 0)
  2771. {
  2772. if (endch != '>') PUGI_IMPL_THROW_ERROR(status_bad_end_element, s);
  2773. }
  2774. else
  2775. {
  2776. if (*s != '>') PUGI_IMPL_THROW_ERROR(status_bad_end_element, s);
  2777. ++s;
  2778. }
  2779. }
  2780. else if (*s == '?') // '<?...'
  2781. {
  2782. s = parse_question(s, cursor, optmsk, endch);
  2783. if (!s) return s;
  2784. assert(cursor);
  2785. if (PUGI_IMPL_NODETYPE(cursor) == node_declaration) goto LOC_ATTRIBUTES;
  2786. }
  2787. else if (*s == '!') // '<!...'
  2788. {
  2789. s = parse_exclamation(s, cursor, optmsk, endch);
  2790. if (!s) return s;
  2791. }
  2792. else if (*s == 0 && endch == '?') PUGI_IMPL_THROW_ERROR(status_bad_pi, s);
  2793. else PUGI_IMPL_THROW_ERROR(status_unrecognized_tag, s);
  2794. }
  2795. else
  2796. {
  2797. mark = s; // Save this offset while searching for a terminator.
  2798. PUGI_IMPL_SKIPWS(); // Eat whitespace if no genuine PCDATA here.
  2799. if (*s == '<' || !*s)
  2800. {
  2801. // We skipped some whitespace characters because otherwise we would take the tag branch instead of PCDATA one
  2802. assert(mark != s);
  2803. if (!PUGI_IMPL_OPTSET(parse_ws_pcdata | parse_ws_pcdata_single) || PUGI_IMPL_OPTSET(parse_trim_pcdata))
  2804. {
  2805. continue;
  2806. }
  2807. else if (PUGI_IMPL_OPTSET(parse_ws_pcdata_single))
  2808. {
  2809. if (s[0] != '<' || s[1] != '/' || cursor->first_child) continue;
  2810. }
  2811. }
  2812. if (!PUGI_IMPL_OPTSET(parse_trim_pcdata))
  2813. s = mark;
  2814. if (cursor->parent || PUGI_IMPL_OPTSET(parse_fragment))
  2815. {
  2816. char_t* parsed_pcdata = s;
  2817. s = strconv_pcdata(s);
  2818. if (PUGI_IMPL_OPTSET(parse_embed_pcdata) && cursor->parent && !cursor->first_child && !cursor->value)
  2819. {
  2820. cursor->value = parsed_pcdata; // Save the offset.
  2821. }
  2822. else if (PUGI_IMPL_OPTSET(parse_merge_pcdata) && cursor->first_child && PUGI_IMPL_NODETYPE(cursor->first_child->prev_sibling_c) == node_pcdata)
  2823. {
  2824. assert(merged_pcdata >= cursor->first_child->prev_sibling_c->value);
  2825. // Catch up to the end of last parsed value; only needed for the first fragment.
  2826. merged_pcdata += strlength(merged_pcdata);
  2827. size_t length = strlength(parsed_pcdata);
  2828. // Must use memmove instead of memcpy as this move may overlap
  2829. memmove(merged_pcdata, parsed_pcdata, (length + 1) * sizeof(char_t));
  2830. merged_pcdata += length;
  2831. }
  2832. else
  2833. {
  2834. xml_node_struct* prev_cursor = cursor;
  2835. PUGI_IMPL_PUSHNODE(node_pcdata); // Append a new node on the tree.
  2836. cursor->value = parsed_pcdata; // Save the offset.
  2837. merged_pcdata = parsed_pcdata; // Used for parse_merge_pcdata above, cheaper to save unconditionally
  2838. cursor = prev_cursor; // Pop since this is a standalone.
  2839. }
  2840. if (!*s) break;
  2841. }
  2842. else
  2843. {
  2844. PUGI_IMPL_SCANFOR(*s == '<'); // '...<'
  2845. if (!*s) break;
  2846. ++s;
  2847. }
  2848. // We're after '<'
  2849. goto LOC_TAG;
  2850. }
  2851. }
  2852. // check that last tag is closed
  2853. if (cursor != root) PUGI_IMPL_THROW_ERROR(status_end_element_mismatch, s);
  2854. return s;
  2855. }
  2856. #ifdef PUGIXML_WCHAR_MODE
  2857. static char_t* parse_skip_bom(char_t* s)
  2858. {
  2859. unsigned int bom = 0xfeff;
  2860. return (s[0] == static_cast<wchar_t>(bom)) ? s + 1 : s;
  2861. }
  2862. #else
  2863. static char_t* parse_skip_bom(char_t* s)
  2864. {
  2865. return (s[0] == '\xef' && s[1] == '\xbb' && s[2] == '\xbf') ? s + 3 : s;
  2866. }
  2867. #endif
  2868. static bool has_element_node_siblings(xml_node_struct* node)
  2869. {
  2870. while (node)
  2871. {
  2872. if (PUGI_IMPL_NODETYPE(node) == node_element) return true;
  2873. node = node->next_sibling;
  2874. }
  2875. return false;
  2876. }
  2877. static xml_parse_result parse(char_t* buffer, size_t length, xml_document_struct* xmldoc, xml_node_struct* root, unsigned int optmsk)
  2878. {
  2879. // early-out for empty documents
  2880. if (length == 0)
  2881. return make_parse_result(PUGI_IMPL_OPTSET(parse_fragment) ? status_ok : status_no_document_element);
  2882. // get last child of the root before parsing
  2883. xml_node_struct* last_root_child = root->first_child ? root->first_child->prev_sibling_c + 0 : NULL;
  2884. // create parser on stack
  2885. xml_parser parser(static_cast<xml_allocator*>(xmldoc));
  2886. // save last character and make buffer zero-terminated (speeds up parsing)
  2887. char_t endch = buffer[length - 1];
  2888. buffer[length - 1] = 0;
  2889. // skip BOM to make sure it does not end up as part of parse output
  2890. char_t* buffer_data = parse_skip_bom(buffer);
  2891. // perform actual parsing
  2892. parser.parse_tree(buffer_data, root, optmsk, endch);
  2893. xml_parse_result result = make_parse_result(parser.error_status, parser.error_offset ? parser.error_offset - buffer : 0);
  2894. assert(result.offset >= 0 && static_cast<size_t>(result.offset) <= length);
  2895. if (result)
  2896. {
  2897. // since we removed last character, we have to handle the only possible false positive (stray <)
  2898. if (endch == '<')
  2899. return make_parse_result(status_unrecognized_tag, length - 1);
  2900. // check if there are any element nodes parsed
  2901. xml_node_struct* first_root_child_parsed = last_root_child ? last_root_child->next_sibling + 0 : root->first_child + 0;
  2902. if (!PUGI_IMPL_OPTSET(parse_fragment) && !has_element_node_siblings(first_root_child_parsed))
  2903. return make_parse_result(status_no_document_element, length - 1);
  2904. }
  2905. else
  2906. {
  2907. // roll back offset if it occurs on a null terminator in the source buffer
  2908. if (result.offset > 0 && static_cast<size_t>(result.offset) == length - 1 && endch == 0)
  2909. result.offset--;
  2910. }
  2911. return result;
  2912. }
  2913. };
  2914. // Output facilities
  2915. PUGI_IMPL_FN xml_encoding get_write_native_encoding()
  2916. {
  2917. #ifdef PUGIXML_WCHAR_MODE
  2918. return get_wchar_encoding();
  2919. #else
  2920. return encoding_utf8;
  2921. #endif
  2922. }
  2923. PUGI_IMPL_FN xml_encoding get_write_encoding(xml_encoding encoding)
  2924. {
  2925. // replace wchar encoding with utf implementation
  2926. if (encoding == encoding_wchar) return get_wchar_encoding();
  2927. // replace utf16 encoding with utf16 with specific endianness
  2928. if (encoding == encoding_utf16) return is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
  2929. // replace utf32 encoding with utf32 with specific endianness
  2930. if (encoding == encoding_utf32) return is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
  2931. // only do autodetection if no explicit encoding is requested
  2932. if (encoding != encoding_auto) return encoding;
  2933. // assume utf8 encoding
  2934. return encoding_utf8;
  2935. }
  2936. template <typename D, typename T> PUGI_IMPL_FN size_t convert_buffer_output_generic(typename T::value_type dest, const char_t* data, size_t length, D, T)
  2937. {
  2938. PUGI_IMPL_STATIC_ASSERT(sizeof(char_t) == sizeof(typename D::type));
  2939. typename T::value_type end = D::process(reinterpret_cast<const typename D::type*>(data), length, dest, T());
  2940. return static_cast<size_t>(end - dest) * sizeof(*dest);
  2941. }
  2942. template <typename D, typename T> PUGI_IMPL_FN size_t convert_buffer_output_generic(typename T::value_type dest, const char_t* data, size_t length, D, T, bool opt_swap)
  2943. {
  2944. PUGI_IMPL_STATIC_ASSERT(sizeof(char_t) == sizeof(typename D::type));
  2945. typename T::value_type end = D::process(reinterpret_cast<const typename D::type*>(data), length, dest, T());
  2946. if (opt_swap)
  2947. {
  2948. for (typename T::value_type i = dest; i != end; ++i)
  2949. *i = endian_swap(*i);
  2950. }
  2951. return static_cast<size_t>(end - dest) * sizeof(*dest);
  2952. }
  2953. #ifdef PUGIXML_WCHAR_MODE
  2954. PUGI_IMPL_FN size_t get_valid_length(const char_t* data, size_t length)
  2955. {
  2956. if (length < 1) return 0;
  2957. // discard last character if it's the lead of a surrogate pair
  2958. return (sizeof(wchar_t) == 2 && static_cast<unsigned int>(static_cast<uint16_t>(data[length - 1]) - 0xD800) < 0x400) ? length - 1 : length;
  2959. }
  2960. PUGI_IMPL_FN size_t convert_buffer_output(char_t* r_char, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding)
  2961. {
  2962. // only endian-swapping is required
  2963. if (need_endian_swap_utf(encoding, get_wchar_encoding()))
  2964. {
  2965. convert_wchar_endian_swap(r_char, data, length);
  2966. return length * sizeof(char_t);
  2967. }
  2968. // convert to utf8
  2969. if (encoding == encoding_utf8)
  2970. return convert_buffer_output_generic(r_u8, data, length, wchar_decoder(), utf8_writer());
  2971. // convert to utf16
  2972. if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
  2973. {
  2974. xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
  2975. return convert_buffer_output_generic(r_u16, data, length, wchar_decoder(), utf16_writer(), native_encoding != encoding);
  2976. }
  2977. // convert to utf32
  2978. if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
  2979. {
  2980. xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
  2981. return convert_buffer_output_generic(r_u32, data, length, wchar_decoder(), utf32_writer(), native_encoding != encoding);
  2982. }
  2983. // convert to latin1
  2984. if (encoding == encoding_latin1)
  2985. return convert_buffer_output_generic(r_u8, data, length, wchar_decoder(), latin1_writer());
  2986. assert(false && "Invalid encoding"); // unreachable
  2987. return 0;
  2988. }
  2989. #else
  2990. PUGI_IMPL_FN size_t get_valid_length(const char_t* data, size_t length)
  2991. {
  2992. if (length < 5) return 0;
  2993. for (size_t i = 1; i <= 4; ++i)
  2994. {
  2995. uint8_t ch = static_cast<uint8_t>(data[length - i]);
  2996. // either a standalone character or a leading one
  2997. if ((ch & 0xc0) != 0x80) return length - i;
  2998. }
  2999. // there are four non-leading characters at the end, sequence tail is broken so might as well process the whole chunk
  3000. return length;
  3001. }
  3002. PUGI_IMPL_FN size_t convert_buffer_output(char_t* /* r_char */, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding)
  3003. {
  3004. if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
  3005. {
  3006. xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
  3007. return convert_buffer_output_generic(r_u16, data, length, utf8_decoder(), utf16_writer(), native_encoding != encoding);
  3008. }
  3009. if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
  3010. {
  3011. xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
  3012. return convert_buffer_output_generic(r_u32, data, length, utf8_decoder(), utf32_writer(), native_encoding != encoding);
  3013. }
  3014. if (encoding == encoding_latin1)
  3015. return convert_buffer_output_generic(r_u8, data, length, utf8_decoder(), latin1_writer());
  3016. assert(false && "Invalid encoding"); // unreachable
  3017. return 0;
  3018. }
  3019. #endif
  3020. class xml_buffered_writer
  3021. {
  3022. xml_buffered_writer(const xml_buffered_writer&);
  3023. xml_buffered_writer& operator=(const xml_buffered_writer&);
  3024. public:
  3025. xml_buffered_writer(xml_writer& writer_, xml_encoding user_encoding): writer(writer_), bufsize(0), encoding(get_write_encoding(user_encoding))
  3026. {
  3027. PUGI_IMPL_STATIC_ASSERT(bufcapacity >= 8);
  3028. }
  3029. size_t flush()
  3030. {
  3031. flush(buffer, bufsize);
  3032. bufsize = 0;
  3033. return 0;
  3034. }
  3035. void flush(const char_t* data, size_t size)
  3036. {
  3037. if (size == 0) return;
  3038. // fast path, just write data
  3039. if (encoding == get_write_native_encoding())
  3040. writer.write(data, size * sizeof(char_t));
  3041. else
  3042. {
  3043. // convert chunk
  3044. size_t result = convert_buffer_output(scratch.data_char, scratch.data_u8, scratch.data_u16, scratch.data_u32, data, size, encoding);
  3045. assert(result <= sizeof(scratch));
  3046. // write data
  3047. writer.write(scratch.data_u8, result);
  3048. }
  3049. }
  3050. void write_direct(const char_t* data, size_t length)
  3051. {
  3052. // flush the remaining buffer contents
  3053. flush();
  3054. // handle large chunks
  3055. if (length > bufcapacity)
  3056. {
  3057. if (encoding == get_write_native_encoding())
  3058. {
  3059. // fast path, can just write data chunk
  3060. writer.write(data, length * sizeof(char_t));
  3061. return;
  3062. }
  3063. // need to convert in suitable chunks
  3064. while (length > bufcapacity)
  3065. {
  3066. // get chunk size by selecting such number of characters that are guaranteed to fit into scratch buffer
  3067. // and form a complete codepoint sequence (i.e. discard start of last codepoint if necessary)
  3068. size_t chunk_size = get_valid_length(data, bufcapacity);
  3069. assert(chunk_size);
  3070. // convert chunk and write
  3071. flush(data, chunk_size);
  3072. // iterate
  3073. data += chunk_size;
  3074. length -= chunk_size;
  3075. }
  3076. // small tail is copied below
  3077. bufsize = 0;
  3078. }
  3079. memcpy(buffer + bufsize, data, length * sizeof(char_t));
  3080. bufsize += length;
  3081. }
  3082. void write_buffer(const char_t* data, size_t length)
  3083. {
  3084. size_t offset = bufsize;
  3085. if (offset + length <= bufcapacity)
  3086. {
  3087. memcpy(buffer + offset, data, length * sizeof(char_t));
  3088. bufsize = offset + length;
  3089. }
  3090. else
  3091. {
  3092. write_direct(data, length);
  3093. }
  3094. }
  3095. void write_string(const char_t* data)
  3096. {
  3097. // write the part of the string that fits in the buffer
  3098. size_t offset = bufsize;
  3099. while (*data && offset < bufcapacity)
  3100. buffer[offset++] = *data++;
  3101. // write the rest
  3102. if (offset < bufcapacity)
  3103. {
  3104. bufsize = offset;
  3105. }
  3106. else
  3107. {
  3108. // backtrack a bit if we have split the codepoint
  3109. size_t length = offset - bufsize;
  3110. size_t extra = length - get_valid_length(data - length, length);
  3111. bufsize = offset - extra;
  3112. write_direct(data - extra, strlength(data) + extra);
  3113. }
  3114. }
  3115. void write(char_t d0)
  3116. {
  3117. size_t offset = bufsize;
  3118. if (offset > bufcapacity - 1) offset = flush();
  3119. buffer[offset + 0] = d0;
  3120. bufsize = offset + 1;
  3121. }
  3122. void write(char_t d0, char_t d1)
  3123. {
  3124. size_t offset = bufsize;
  3125. if (offset > bufcapacity - 2) offset = flush();
  3126. buffer[offset + 0] = d0;
  3127. buffer[offset + 1] = d1;
  3128. bufsize = offset + 2;
  3129. }
  3130. void write(char_t d0, char_t d1, char_t d2)
  3131. {
  3132. size_t offset = bufsize;
  3133. if (offset > bufcapacity - 3) offset = flush();
  3134. buffer[offset + 0] = d0;
  3135. buffer[offset + 1] = d1;
  3136. buffer[offset + 2] = d2;
  3137. bufsize = offset + 3;
  3138. }
  3139. void write(char_t d0, char_t d1, char_t d2, char_t d3)
  3140. {
  3141. size_t offset = bufsize;
  3142. if (offset > bufcapacity - 4) offset = flush();
  3143. buffer[offset + 0] = d0;
  3144. buffer[offset + 1] = d1;
  3145. buffer[offset + 2] = d2;
  3146. buffer[offset + 3] = d3;
  3147. bufsize = offset + 4;
  3148. }
  3149. void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4)
  3150. {
  3151. size_t offset = bufsize;
  3152. if (offset > bufcapacity - 5) offset = flush();
  3153. buffer[offset + 0] = d0;
  3154. buffer[offset + 1] = d1;
  3155. buffer[offset + 2] = d2;
  3156. buffer[offset + 3] = d3;
  3157. buffer[offset + 4] = d4;
  3158. bufsize = offset + 5;
  3159. }
  3160. void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4, char_t d5)
  3161. {
  3162. size_t offset = bufsize;
  3163. if (offset > bufcapacity - 6) offset = flush();
  3164. buffer[offset + 0] = d0;
  3165. buffer[offset + 1] = d1;
  3166. buffer[offset + 2] = d2;
  3167. buffer[offset + 3] = d3;
  3168. buffer[offset + 4] = d4;
  3169. buffer[offset + 5] = d5;
  3170. bufsize = offset + 6;
  3171. }
  3172. // utf8 maximum expansion: x4 (-> utf32)
  3173. // utf16 maximum expansion: x2 (-> utf32)
  3174. // utf32 maximum expansion: x1
  3175. enum
  3176. {
  3177. bufcapacitybytes =
  3178. #ifdef PUGIXML_MEMORY_OUTPUT_STACK
  3179. PUGIXML_MEMORY_OUTPUT_STACK
  3180. #else
  3181. 10240
  3182. #endif
  3183. ,
  3184. bufcapacity = bufcapacitybytes / (sizeof(char_t) + 4)
  3185. };
  3186. char_t buffer[bufcapacity];
  3187. union
  3188. {
  3189. uint8_t data_u8[4 * bufcapacity];
  3190. uint16_t data_u16[2 * bufcapacity];
  3191. uint32_t data_u32[bufcapacity];
  3192. char_t data_char[bufcapacity];
  3193. } scratch;
  3194. xml_writer& writer;
  3195. size_t bufsize;
  3196. xml_encoding encoding;
  3197. };
  3198. PUGI_IMPL_FN void text_output_escaped(xml_buffered_writer& writer, const char_t* s, chartypex_t type, unsigned int flags)
  3199. {
  3200. while (*s)
  3201. {
  3202. const char_t* prev = s;
  3203. // While *s is a usual symbol
  3204. PUGI_IMPL_SCANWHILE_UNROLL(!PUGI_IMPL_IS_CHARTYPEX(ss, type));
  3205. writer.write_buffer(prev, static_cast<size_t>(s - prev));
  3206. switch (*s)
  3207. {
  3208. case 0: break;
  3209. case '&':
  3210. writer.write('&', 'a', 'm', 'p', ';');
  3211. ++s;
  3212. break;
  3213. case '<':
  3214. writer.write('&', 'l', 't', ';');
  3215. ++s;
  3216. break;
  3217. case '>':
  3218. writer.write('&', 'g', 't', ';');
  3219. ++s;
  3220. break;
  3221. case '"':
  3222. if (flags & format_attribute_single_quote)
  3223. writer.write('"');
  3224. else
  3225. writer.write('&', 'q', 'u', 'o', 't', ';');
  3226. ++s;
  3227. break;
  3228. case '\'':
  3229. if (flags & format_attribute_single_quote)
  3230. writer.write('&', 'a', 'p', 'o', 's', ';');
  3231. else
  3232. writer.write('\'');
  3233. ++s;
  3234. break;
  3235. default: // s is not a usual symbol
  3236. {
  3237. unsigned int ch = static_cast<unsigned int>(*s++);
  3238. assert(ch < 32);
  3239. if (!(flags & format_skip_control_chars))
  3240. writer.write('&', '#', static_cast<char_t>((ch / 10) + '0'), static_cast<char_t>((ch % 10) + '0'), ';');
  3241. }
  3242. }
  3243. }
  3244. }
  3245. PUGI_IMPL_FN void text_output(xml_buffered_writer& writer, const char_t* s, chartypex_t type, unsigned int flags)
  3246. {
  3247. if (flags & format_no_escapes)
  3248. writer.write_string(s);
  3249. else
  3250. text_output_escaped(writer, s, type, flags);
  3251. }
  3252. PUGI_IMPL_FN void text_output_cdata(xml_buffered_writer& writer, const char_t* s)
  3253. {
  3254. do
  3255. {
  3256. writer.write('<', '!', '[', 'C', 'D');
  3257. writer.write('A', 'T', 'A', '[');
  3258. const char_t* prev = s;
  3259. // look for ]]> sequence - we can't output it as is since it terminates CDATA
  3260. while (*s && !(s[0] == ']' && s[1] == ']' && s[2] == '>')) ++s;
  3261. // skip ]] if we stopped at ]]>, > will go to the next CDATA section
  3262. if (*s) s += 2;
  3263. writer.write_buffer(prev, static_cast<size_t>(s - prev));
  3264. writer.write(']', ']', '>');
  3265. }
  3266. while (*s);
  3267. }
  3268. PUGI_IMPL_FN void text_output_indent(xml_buffered_writer& writer, const char_t* indent, size_t indent_length, unsigned int depth)
  3269. {
  3270. switch (indent_length)
  3271. {
  3272. case 1:
  3273. {
  3274. for (unsigned int i = 0; i < depth; ++i)
  3275. writer.write(indent[0]);
  3276. break;
  3277. }
  3278. case 2:
  3279. {
  3280. for (unsigned int i = 0; i < depth; ++i)
  3281. writer.write(indent[0], indent[1]);
  3282. break;
  3283. }
  3284. case 3:
  3285. {
  3286. for (unsigned int i = 0; i < depth; ++i)
  3287. writer.write(indent[0], indent[1], indent[2]);
  3288. break;
  3289. }
  3290. case 4:
  3291. {
  3292. for (unsigned int i = 0; i < depth; ++i)
  3293. writer.write(indent[0], indent[1], indent[2], indent[3]);
  3294. break;
  3295. }
  3296. default:
  3297. {
  3298. for (unsigned int i = 0; i < depth; ++i)
  3299. writer.write_buffer(indent, indent_length);
  3300. }
  3301. }
  3302. }
  3303. PUGI_IMPL_FN void node_output_comment(xml_buffered_writer& writer, const char_t* s)
  3304. {
  3305. writer.write('<', '!', '-', '-');
  3306. while (*s)
  3307. {
  3308. const char_t* prev = s;
  3309. // look for -\0 or -- sequence - we can't output it since -- is illegal in comment body
  3310. while (*s && !(s[0] == '-' && (s[1] == '-' || s[1] == 0))) ++s;
  3311. writer.write_buffer(prev, static_cast<size_t>(s - prev));
  3312. if (*s)
  3313. {
  3314. assert(*s == '-');
  3315. writer.write('-', ' ');
  3316. ++s;
  3317. }
  3318. }
  3319. writer.write('-', '-', '>');
  3320. }
  3321. PUGI_IMPL_FN void node_output_pi_value(xml_buffered_writer& writer, const char_t* s)
  3322. {
  3323. while (*s)
  3324. {
  3325. const char_t* prev = s;
  3326. // look for ?> sequence - we can't output it since ?> terminates PI
  3327. while (*s && !(s[0] == '?' && s[1] == '>')) ++s;
  3328. writer.write_buffer(prev, static_cast<size_t>(s - prev));
  3329. if (*s)
  3330. {
  3331. assert(s[0] == '?' && s[1] == '>');
  3332. writer.write('?', ' ', '>');
  3333. s += 2;
  3334. }
  3335. }
  3336. }
  3337. PUGI_IMPL_FN void node_output_attributes(xml_buffered_writer& writer, xml_node_struct* node, const char_t* indent, size_t indent_length, unsigned int flags, unsigned int depth)
  3338. {
  3339. const char_t* default_name = PUGIXML_TEXT(":anonymous");
  3340. const char_t enquotation_char = (flags & format_attribute_single_quote) ? '\'' : '"';
  3341. for (xml_attribute_struct* a = node->first_attribute; a; a = a->next_attribute)
  3342. {
  3343. if ((flags & (format_indent_attributes | format_raw)) == format_indent_attributes)
  3344. {
  3345. writer.write('\n');
  3346. text_output_indent(writer, indent, indent_length, depth + 1);
  3347. }
  3348. else
  3349. {
  3350. writer.write(' ');
  3351. }
  3352. writer.write_string(a->name ? a->name + 0 : default_name);
  3353. writer.write('=', enquotation_char);
  3354. if (a->value)
  3355. text_output(writer, a->value, ctx_special_attr, flags);
  3356. writer.write(enquotation_char);
  3357. }
  3358. }
  3359. PUGI_IMPL_FN bool node_output_start(xml_buffered_writer& writer, xml_node_struct* node, const char_t* indent, size_t indent_length, unsigned int flags, unsigned int depth)
  3360. {
  3361. const char_t* default_name = PUGIXML_TEXT(":anonymous");
  3362. const char_t* name = node->name ? node->name + 0 : default_name;
  3363. writer.write('<');
  3364. writer.write_string(name);
  3365. if (node->first_attribute)
  3366. node_output_attributes(writer, node, indent, indent_length, flags, depth);
  3367. // element nodes can have value if parse_embed_pcdata was used
  3368. if (!node->value)
  3369. {
  3370. if (!node->first_child)
  3371. {
  3372. if (flags & format_no_empty_element_tags)
  3373. {
  3374. writer.write('>', '<', '/');
  3375. writer.write_string(name);
  3376. writer.write('>');
  3377. return false;
  3378. }
  3379. else
  3380. {
  3381. if ((flags & format_raw) == 0)
  3382. writer.write(' ');
  3383. writer.write('/', '>');
  3384. return false;
  3385. }
  3386. }
  3387. else
  3388. {
  3389. writer.write('>');
  3390. return true;
  3391. }
  3392. }
  3393. else
  3394. {
  3395. writer.write('>');
  3396. text_output(writer, node->value, ctx_special_pcdata, flags);
  3397. if (!node->first_child)
  3398. {
  3399. writer.write('<', '/');
  3400. writer.write_string(name);
  3401. writer.write('>');
  3402. return false;
  3403. }
  3404. else
  3405. {
  3406. return true;
  3407. }
  3408. }
  3409. }
  3410. PUGI_IMPL_FN void node_output_end(xml_buffered_writer& writer, xml_node_struct* node)
  3411. {
  3412. const char_t* default_name = PUGIXML_TEXT(":anonymous");
  3413. const char_t* name = node->name ? node->name + 0 : default_name;
  3414. writer.write('<', '/');
  3415. writer.write_string(name);
  3416. writer.write('>');
  3417. }
  3418. PUGI_IMPL_FN void node_output_simple(xml_buffered_writer& writer, xml_node_struct* node, unsigned int flags)
  3419. {
  3420. const char_t* default_name = PUGIXML_TEXT(":anonymous");
  3421. switch (PUGI_IMPL_NODETYPE(node))
  3422. {
  3423. case node_pcdata:
  3424. text_output(writer, node->value ? node->value + 0 : PUGIXML_TEXT(""), ctx_special_pcdata, flags);
  3425. break;
  3426. case node_cdata:
  3427. text_output_cdata(writer, node->value ? node->value + 0 : PUGIXML_TEXT(""));
  3428. break;
  3429. case node_comment:
  3430. node_output_comment(writer, node->value ? node->value + 0 : PUGIXML_TEXT(""));
  3431. break;
  3432. case node_pi:
  3433. writer.write('<', '?');
  3434. writer.write_string(node->name ? node->name + 0 : default_name);
  3435. if (node->value)
  3436. {
  3437. writer.write(' ');
  3438. node_output_pi_value(writer, node->value);
  3439. }
  3440. writer.write('?', '>');
  3441. break;
  3442. case node_declaration:
  3443. writer.write('<', '?');
  3444. writer.write_string(node->name ? node->name + 0 : default_name);
  3445. node_output_attributes(writer, node, PUGIXML_TEXT(""), 0, flags | format_raw, 0);
  3446. writer.write('?', '>');
  3447. break;
  3448. case node_doctype:
  3449. writer.write('<', '!', 'D', 'O', 'C');
  3450. writer.write('T', 'Y', 'P', 'E');
  3451. if (node->value)
  3452. {
  3453. writer.write(' ');
  3454. writer.write_string(node->value);
  3455. }
  3456. writer.write('>');
  3457. break;
  3458. default:
  3459. assert(false && "Invalid node type"); // unreachable
  3460. }
  3461. }
  3462. enum indent_flags_t
  3463. {
  3464. indent_newline = 1,
  3465. indent_indent = 2
  3466. };
  3467. PUGI_IMPL_FN void node_output(xml_buffered_writer& writer, xml_node_struct* root, const char_t* indent, unsigned int flags, unsigned int depth)
  3468. {
  3469. size_t indent_length = ((flags & (format_indent | format_indent_attributes)) && (flags & format_raw) == 0) ? strlength(indent) : 0;
  3470. unsigned int indent_flags = indent_indent;
  3471. xml_node_struct* node = root;
  3472. do
  3473. {
  3474. assert(node);
  3475. // begin writing current node
  3476. if (PUGI_IMPL_NODETYPE(node) == node_pcdata || PUGI_IMPL_NODETYPE(node) == node_cdata)
  3477. {
  3478. node_output_simple(writer, node, flags);
  3479. indent_flags = 0;
  3480. }
  3481. else
  3482. {
  3483. if ((indent_flags & indent_newline) && (flags & format_raw) == 0)
  3484. writer.write('\n');
  3485. if ((indent_flags & indent_indent) && indent_length)
  3486. text_output_indent(writer, indent, indent_length, depth);
  3487. if (PUGI_IMPL_NODETYPE(node) == node_element)
  3488. {
  3489. indent_flags = indent_newline | indent_indent;
  3490. if (node_output_start(writer, node, indent, indent_length, flags, depth))
  3491. {
  3492. // element nodes can have value if parse_embed_pcdata was used
  3493. if (node->value)
  3494. indent_flags = 0;
  3495. node = node->first_child;
  3496. depth++;
  3497. continue;
  3498. }
  3499. }
  3500. else if (PUGI_IMPL_NODETYPE(node) == node_document)
  3501. {
  3502. indent_flags = indent_indent;
  3503. if (node->first_child)
  3504. {
  3505. node = node->first_child;
  3506. continue;
  3507. }
  3508. }
  3509. else
  3510. {
  3511. node_output_simple(writer, node, flags);
  3512. indent_flags = indent_newline | indent_indent;
  3513. }
  3514. }
  3515. // continue to the next node
  3516. while (node != root)
  3517. {
  3518. if (node->next_sibling)
  3519. {
  3520. node = node->next_sibling;
  3521. break;
  3522. }
  3523. node = node->parent;
  3524. // write closing node
  3525. if (PUGI_IMPL_NODETYPE(node) == node_element)
  3526. {
  3527. depth--;
  3528. if ((indent_flags & indent_newline) && (flags & format_raw) == 0)
  3529. writer.write('\n');
  3530. if ((indent_flags & indent_indent) && indent_length)
  3531. text_output_indent(writer, indent, indent_length, depth);
  3532. node_output_end(writer, node);
  3533. indent_flags = indent_newline | indent_indent;
  3534. }
  3535. }
  3536. }
  3537. while (node != root);
  3538. if ((indent_flags & indent_newline) && (flags & format_raw) == 0)
  3539. writer.write('\n');
  3540. }
  3541. PUGI_IMPL_FN bool has_declaration(xml_node_struct* node)
  3542. {
  3543. for (xml_node_struct* child = node->first_child; child; child = child->next_sibling)
  3544. {
  3545. xml_node_type type = PUGI_IMPL_NODETYPE(child);
  3546. if (type == node_declaration) return true;
  3547. if (type == node_element) return false;
  3548. }
  3549. return false;
  3550. }
  3551. PUGI_IMPL_FN bool is_attribute_of(xml_attribute_struct* attr, xml_node_struct* node)
  3552. {
  3553. for (xml_attribute_struct* a = node->first_attribute; a; a = a->next_attribute)
  3554. if (a == attr)
  3555. return true;
  3556. return false;
  3557. }
  3558. PUGI_IMPL_FN bool allow_insert_attribute(xml_node_type parent)
  3559. {
  3560. return parent == node_element || parent == node_declaration;
  3561. }
  3562. PUGI_IMPL_FN bool allow_insert_child(xml_node_type parent, xml_node_type child)
  3563. {
  3564. if (parent != node_document && parent != node_element) return false;
  3565. if (child == node_document || child == node_null) return false;
  3566. if (parent != node_document && (child == node_declaration || child == node_doctype)) return false;
  3567. return true;
  3568. }
  3569. PUGI_IMPL_FN bool allow_move(xml_node parent, xml_node child)
  3570. {
  3571. // check that child can be a child of parent
  3572. if (!allow_insert_child(parent.type(), child.type()))
  3573. return false;
  3574. // check that node is not moved between documents
  3575. if (parent.root() != child.root())
  3576. return false;
  3577. // check that new parent is not in the child subtree
  3578. xml_node cur = parent;
  3579. while (cur)
  3580. {
  3581. if (cur == child)
  3582. return false;
  3583. cur = cur.parent();
  3584. }
  3585. return true;
  3586. }
  3587. template <typename String, typename Header>
  3588. PUGI_IMPL_FN void node_copy_string(String& dest, Header& header, uintptr_t header_mask, char_t* source, Header& source_header, xml_allocator* alloc)
  3589. {
  3590. assert(!dest && (header & header_mask) == 0); // copies are performed into fresh nodes
  3591. if (source)
  3592. {
  3593. if (alloc && (source_header & header_mask) == 0)
  3594. {
  3595. dest = source;
  3596. // since strcpy_insitu can reuse document buffer memory we need to mark both source and dest as shared
  3597. header |= xml_memory_page_contents_shared_mask;
  3598. source_header |= xml_memory_page_contents_shared_mask;
  3599. }
  3600. else
  3601. {
  3602. // if strcpy_insitu fails (out of memory) we just leave the destination name/value empty
  3603. (void)strcpy_insitu(dest, header, header_mask, source, strlength(source));
  3604. }
  3605. }
  3606. }
  3607. PUGI_IMPL_FN void node_copy_contents(xml_node_struct* dn, xml_node_struct* sn, xml_allocator* shared_alloc)
  3608. {
  3609. node_copy_string(dn->name, dn->header, xml_memory_page_name_allocated_mask, sn->name, sn->header, shared_alloc);
  3610. node_copy_string(dn->value, dn->header, xml_memory_page_value_allocated_mask, sn->value, sn->header, shared_alloc);
  3611. for (xml_attribute_struct* sa = sn->first_attribute; sa; sa = sa->next_attribute)
  3612. {
  3613. xml_attribute_struct* da = append_new_attribute(dn, get_allocator(dn));
  3614. if (da)
  3615. {
  3616. node_copy_string(da->name, da->header, xml_memory_page_name_allocated_mask, sa->name, sa->header, shared_alloc);
  3617. node_copy_string(da->value, da->header, xml_memory_page_value_allocated_mask, sa->value, sa->header, shared_alloc);
  3618. }
  3619. }
  3620. }
  3621. PUGI_IMPL_FN void node_copy_tree(xml_node_struct* dn, xml_node_struct* sn)
  3622. {
  3623. xml_allocator& alloc = get_allocator(dn);
  3624. xml_allocator* shared_alloc = (&alloc == &get_allocator(sn)) ? &alloc : NULL;
  3625. node_copy_contents(dn, sn, shared_alloc);
  3626. xml_node_struct* dit = dn;
  3627. xml_node_struct* sit = sn->first_child;
  3628. while (sit && sit != sn)
  3629. {
  3630. // loop invariant: dit is inside the subtree rooted at dn
  3631. assert(dit);
  3632. // when a tree is copied into one of the descendants, we need to skip that subtree to avoid an infinite loop
  3633. if (sit != dn)
  3634. {
  3635. xml_node_struct* copy = append_new_node(dit, alloc, PUGI_IMPL_NODETYPE(sit));
  3636. if (copy)
  3637. {
  3638. node_copy_contents(copy, sit, shared_alloc);
  3639. if (sit->first_child)
  3640. {
  3641. dit = copy;
  3642. sit = sit->first_child;
  3643. continue;
  3644. }
  3645. }
  3646. }
  3647. // continue to the next node
  3648. do
  3649. {
  3650. if (sit->next_sibling)
  3651. {
  3652. sit = sit->next_sibling;
  3653. break;
  3654. }
  3655. sit = sit->parent;
  3656. dit = dit->parent;
  3657. // loop invariant: dit is inside the subtree rooted at dn while sit is inside sn
  3658. assert(sit == sn || dit);
  3659. }
  3660. while (sit != sn);
  3661. }
  3662. assert(!sit || dit == dn->parent);
  3663. }
  3664. PUGI_IMPL_FN void node_copy_attribute(xml_attribute_struct* da, xml_attribute_struct* sa)
  3665. {
  3666. xml_allocator& alloc = get_allocator(da);
  3667. xml_allocator* shared_alloc = (&alloc == &get_allocator(sa)) ? &alloc : NULL;
  3668. node_copy_string(da->name, da->header, xml_memory_page_name_allocated_mask, sa->name, sa->header, shared_alloc);
  3669. node_copy_string(da->value, da->header, xml_memory_page_value_allocated_mask, sa->value, sa->header, shared_alloc);
  3670. }
  3671. inline bool is_text_node(xml_node_struct* node)
  3672. {
  3673. xml_node_type type = PUGI_IMPL_NODETYPE(node);
  3674. return type == node_pcdata || type == node_cdata;
  3675. }
  3676. // get value with conversion functions
  3677. template <typename U> PUGI_IMPL_FN PUGI_IMPL_UNSIGNED_OVERFLOW U string_to_integer(const char_t* value, U minv, U maxv)
  3678. {
  3679. U result = 0;
  3680. const char_t* s = value;
  3681. while (PUGI_IMPL_IS_CHARTYPE(*s, ct_space))
  3682. s++;
  3683. bool negative = (*s == '-');
  3684. s += (*s == '+' || *s == '-');
  3685. bool overflow = false;
  3686. if (s[0] == '0' && (s[1] | ' ') == 'x')
  3687. {
  3688. s += 2;
  3689. // since overflow detection relies on length of the sequence skip leading zeros
  3690. while (*s == '0')
  3691. s++;
  3692. const char_t* start = s;
  3693. for (;;)
  3694. {
  3695. if (static_cast<unsigned>(*s - '0') < 10)
  3696. result = result * 16 + (*s - '0');
  3697. else if (static_cast<unsigned>((*s | ' ') - 'a') < 6)
  3698. result = result * 16 + ((*s | ' ') - 'a' + 10);
  3699. else
  3700. break;
  3701. s++;
  3702. }
  3703. size_t digits = static_cast<size_t>(s - start);
  3704. overflow = digits > sizeof(U) * 2;
  3705. }
  3706. else
  3707. {
  3708. // since overflow detection relies on length of the sequence skip leading zeros
  3709. while (*s == '0')
  3710. s++;
  3711. const char_t* start = s;
  3712. for (;;)
  3713. {
  3714. if (static_cast<unsigned>(*s - '0') < 10)
  3715. result = result * 10 + (*s - '0');
  3716. else
  3717. break;
  3718. s++;
  3719. }
  3720. size_t digits = static_cast<size_t>(s - start);
  3721. PUGI_IMPL_STATIC_ASSERT(sizeof(U) == 8 || sizeof(U) == 4 || sizeof(U) == 2);
  3722. const size_t max_digits10 = sizeof(U) == 8 ? 20 : sizeof(U) == 4 ? 10 : 5;
  3723. const char_t max_lead = sizeof(U) == 8 ? '1' : sizeof(U) == 4 ? '4' : '6';
  3724. const size_t high_bit = sizeof(U) * 8 - 1;
  3725. overflow = digits >= max_digits10 && !(digits == max_digits10 && (*start < max_lead || (*start == max_lead && result >> high_bit)));
  3726. }
  3727. if (negative)
  3728. {
  3729. // Workaround for crayc++ CC-3059: Expected no overflow in routine.
  3730. #ifdef _CRAYC
  3731. return (overflow || result > ~minv + 1) ? minv : ~result + 1;
  3732. #else
  3733. return (overflow || result > 0 - minv) ? minv : 0 - result;
  3734. #endif
  3735. }
  3736. else
  3737. return (overflow || result > maxv) ? maxv : result;
  3738. }
  3739. PUGI_IMPL_FN int get_value_int(const char_t* value)
  3740. {
  3741. return string_to_integer<unsigned int>(value, static_cast<unsigned int>(INT_MIN), INT_MAX);
  3742. }
  3743. PUGI_IMPL_FN unsigned int get_value_uint(const char_t* value)
  3744. {
  3745. return string_to_integer<unsigned int>(value, 0, UINT_MAX);
  3746. }
  3747. PUGI_IMPL_FN double get_value_double(const char_t* value)
  3748. {
  3749. #ifdef PUGIXML_WCHAR_MODE
  3750. return wcstod(value, NULL);
  3751. #else
  3752. return strtod(value, NULL);
  3753. #endif
  3754. }
  3755. PUGI_IMPL_FN float get_value_float(const char_t* value)
  3756. {
  3757. #ifdef PUGIXML_WCHAR_MODE
  3758. return static_cast<float>(wcstod(value, NULL));
  3759. #else
  3760. return static_cast<float>(strtod(value, NULL));
  3761. #endif
  3762. }
  3763. PUGI_IMPL_FN bool get_value_bool(const char_t* value)
  3764. {
  3765. // only look at first char
  3766. char_t first = *value;
  3767. // 1*, t* (true), T* (True), y* (yes), Y* (YES)
  3768. return (first == '1' || first == 't' || first == 'T' || first == 'y' || first == 'Y');
  3769. }
  3770. #ifdef PUGIXML_HAS_LONG_LONG
  3771. PUGI_IMPL_FN long long get_value_llong(const char_t* value)
  3772. {
  3773. return string_to_integer<unsigned long long>(value, static_cast<unsigned long long>(LLONG_MIN), LLONG_MAX);
  3774. }
  3775. PUGI_IMPL_FN unsigned long long get_value_ullong(const char_t* value)
  3776. {
  3777. return string_to_integer<unsigned long long>(value, 0, ULLONG_MAX);
  3778. }
  3779. #endif
  3780. template <typename U> PUGI_IMPL_FN PUGI_IMPL_UNSIGNED_OVERFLOW char_t* integer_to_string(char_t* begin, char_t* end, U value, bool negative)
  3781. {
  3782. char_t* result = end - 1;
  3783. U rest = negative ? 0 - value : value;
  3784. do
  3785. {
  3786. *result-- = static_cast<char_t>('0' + (rest % 10));
  3787. rest /= 10;
  3788. }
  3789. while (rest);
  3790. assert(result >= begin);
  3791. (void)begin;
  3792. *result = '-';
  3793. return result + !negative;
  3794. }
  3795. // set value with conversion functions
  3796. template <typename String, typename Header>
  3797. PUGI_IMPL_FN bool set_value_ascii(String& dest, Header& header, uintptr_t header_mask, char* buf)
  3798. {
  3799. #ifdef PUGIXML_WCHAR_MODE
  3800. char_t wbuf[128];
  3801. assert(strlen(buf) < sizeof(wbuf) / sizeof(wbuf[0]));
  3802. size_t offset = 0;
  3803. for (; buf[offset]; ++offset) wbuf[offset] = buf[offset];
  3804. return strcpy_insitu(dest, header, header_mask, wbuf, offset);
  3805. #else
  3806. return strcpy_insitu(dest, header, header_mask, buf, strlen(buf));
  3807. #endif
  3808. }
  3809. template <typename U, typename String, typename Header>
  3810. PUGI_IMPL_FN bool set_value_integer(String& dest, Header& header, uintptr_t header_mask, U value, bool negative)
  3811. {
  3812. char_t buf[64];
  3813. char_t* end = buf + sizeof(buf) / sizeof(buf[0]);
  3814. char_t* begin = integer_to_string(buf, end, value, negative);
  3815. return strcpy_insitu(dest, header, header_mask, begin, end - begin);
  3816. }
  3817. template <typename String, typename Header>
  3818. PUGI_IMPL_FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, float value, int precision)
  3819. {
  3820. char buf[128];
  3821. PUGI_IMPL_SNPRINTF(buf, "%.*g", precision, double(value));
  3822. return set_value_ascii(dest, header, header_mask, buf);
  3823. }
  3824. template <typename String, typename Header>
  3825. PUGI_IMPL_FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, double value, int precision)
  3826. {
  3827. char buf[128];
  3828. PUGI_IMPL_SNPRINTF(buf, "%.*g", precision, value);
  3829. return set_value_ascii(dest, header, header_mask, buf);
  3830. }
  3831. template <typename String, typename Header>
  3832. PUGI_IMPL_FN bool set_value_bool(String& dest, Header& header, uintptr_t header_mask, bool value)
  3833. {
  3834. return strcpy_insitu(dest, header, header_mask, value ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false"), value ? 4 : 5);
  3835. }
  3836. PUGI_IMPL_FN xml_parse_result load_buffer_impl(xml_document_struct* doc, xml_node_struct* root, void* contents, size_t size, unsigned int options, xml_encoding encoding, bool is_mutable, bool own, char_t** out_buffer)
  3837. {
  3838. // check input buffer
  3839. if (!contents && size) return make_parse_result(status_io_error);
  3840. // get actual encoding
  3841. xml_encoding buffer_encoding = impl::get_buffer_encoding(encoding, contents, size);
  3842. // if convert_buffer below throws bad_alloc, we still need to deallocate contents if we own it
  3843. auto_deleter<void> contents_guard(own ? contents : NULL, xml_memory::deallocate);
  3844. // early-out for empty documents to avoid buffer allocation overhead
  3845. if (size == 0) return make_parse_result((options & parse_fragment) ? status_ok : status_no_document_element);
  3846. // get private buffer
  3847. char_t* buffer = NULL;
  3848. size_t length = 0;
  3849. if (!impl::convert_buffer(buffer, length, buffer_encoding, contents, size, is_mutable)) return impl::make_parse_result(status_out_of_memory);
  3850. // after this we either deallocate contents (below) or hold on to it via doc->buffer, so we don't need to guard it
  3851. contents_guard.release();
  3852. // delete original buffer if we performed a conversion
  3853. if (own && buffer != contents && contents) impl::xml_memory::deallocate(contents);
  3854. // grab onto buffer if it's our buffer, user is responsible for deallocating contents himself
  3855. if (own || buffer != contents) *out_buffer = buffer;
  3856. // store buffer for offset_debug
  3857. doc->buffer = buffer;
  3858. // parse
  3859. xml_parse_result res = impl::xml_parser::parse(buffer, length, doc, root, options);
  3860. // remember encoding
  3861. res.encoding = buffer_encoding;
  3862. return res;
  3863. }
  3864. template <typename T> PUGI_IMPL_FN xml_parse_status convert_file_size(T length, size_t& out_result)
  3865. {
  3866. // check for I/O errors
  3867. if (length < 0) return status_io_error;
  3868. // check for overflow
  3869. size_t result = static_cast<size_t>(length);
  3870. if (static_cast<T>(result) != length) return status_out_of_memory;
  3871. out_result = result;
  3872. return status_ok;
  3873. }
  3874. // we need to get length of entire file to load it in memory; the only (relatively) sane way to do it is via seek/tell trick
  3875. PUGI_IMPL_FN xml_parse_status get_file_size(FILE* file, size_t& out_result)
  3876. {
  3877. #if defined(__linux__) || defined(__APPLE__)
  3878. // this simultaneously retrieves the file size and file mode (to guard against loading non-files)
  3879. struct stat st;
  3880. if (fstat(fileno(file), &st) != 0) return status_io_error;
  3881. // anything that's not a regular file doesn't have a coherent length
  3882. if (!S_ISREG(st.st_mode)) return status_io_error;
  3883. xml_parse_status status = convert_file_size(st.st_size, out_result);
  3884. #elif defined(PUGI_IMPL_MSVC_CRT_VERSION) && PUGI_IMPL_MSVC_CRT_VERSION >= 1400
  3885. // there are 64-bit versions of fseek/ftell, let's use them
  3886. _fseeki64(file, 0, SEEK_END);
  3887. __int64 length = _ftelli64(file);
  3888. _fseeki64(file, 0, SEEK_SET);
  3889. xml_parse_status status = convert_file_size(length, out_result);
  3890. #elif defined(__MINGW32__) && !defined(__NO_MINGW_LFS) && (!defined(__STRICT_ANSI__) || defined(__MINGW64_VERSION_MAJOR))
  3891. // there are 64-bit versions of fseek/ftell, let's use them
  3892. fseeko64(file, 0, SEEK_END);
  3893. off64_t length = ftello64(file);
  3894. fseeko64(file, 0, SEEK_SET);
  3895. xml_parse_status status = convert_file_size(length, out_result);
  3896. #else
  3897. // if this is a 32-bit OS, long is enough; if this is a unix system, long is 64-bit, which is enough; otherwise we can't do anything anyway.
  3898. fseek(file, 0, SEEK_END);
  3899. long length = ftell(file);
  3900. fseek(file, 0, SEEK_SET);
  3901. xml_parse_status status = convert_file_size(length, out_result);
  3902. #endif
  3903. return status;
  3904. }
  3905. // This function assumes that buffer has extra sizeof(char_t) writable bytes after size
  3906. PUGI_IMPL_FN size_t zero_terminate_buffer(void* buffer, size_t size, xml_encoding encoding)
  3907. {
  3908. // We only need to zero-terminate if encoding conversion does not do it for us
  3909. #ifdef PUGIXML_WCHAR_MODE
  3910. xml_encoding wchar_encoding = get_wchar_encoding();
  3911. if (encoding == wchar_encoding || need_endian_swap_utf(encoding, wchar_encoding))
  3912. {
  3913. size_t length = size / sizeof(char_t);
  3914. static_cast<char_t*>(buffer)[length] = 0;
  3915. return (length + 1) * sizeof(char_t);
  3916. }
  3917. #else
  3918. if (encoding == encoding_utf8)
  3919. {
  3920. static_cast<char*>(buffer)[size] = 0;
  3921. return size + 1;
  3922. }
  3923. #endif
  3924. return size;
  3925. }
  3926. PUGI_IMPL_FN xml_parse_result load_file_impl(xml_document_struct* doc, FILE* file, unsigned int options, xml_encoding encoding, char_t** out_buffer)
  3927. {
  3928. if (!file) return make_parse_result(status_file_not_found);
  3929. // get file size (can result in I/O errors)
  3930. size_t size = 0;
  3931. xml_parse_status size_status = get_file_size(file, size);
  3932. if (size_status != status_ok) return make_parse_result(size_status);
  3933. size_t max_suffix_size = sizeof(char_t);
  3934. // allocate buffer for the whole file
  3935. char* contents = static_cast<char*>(xml_memory::allocate(size + max_suffix_size));
  3936. if (!contents) return make_parse_result(status_out_of_memory);
  3937. // read file in memory
  3938. size_t read_size = fread(contents, 1, size, file);
  3939. if (read_size != size)
  3940. {
  3941. xml_memory::deallocate(contents);
  3942. return make_parse_result(status_io_error);
  3943. }
  3944. xml_encoding real_encoding = get_buffer_encoding(encoding, contents, size);
  3945. return load_buffer_impl(doc, doc, contents, zero_terminate_buffer(contents, size, real_encoding), options, real_encoding, true, true, out_buffer);
  3946. }
  3947. PUGI_IMPL_FN void close_file(FILE* file)
  3948. {
  3949. fclose(file);
  3950. }
  3951. #ifndef PUGIXML_NO_STL
  3952. template <typename T> struct xml_stream_chunk
  3953. {
  3954. static xml_stream_chunk* create()
  3955. {
  3956. void* memory = xml_memory::allocate(sizeof(xml_stream_chunk));
  3957. if (!memory) return NULL;
  3958. return new (memory) xml_stream_chunk();
  3959. }
  3960. static void destroy(xml_stream_chunk* chunk)
  3961. {
  3962. // free chunk chain
  3963. while (chunk)
  3964. {
  3965. xml_stream_chunk* next_ = chunk->next;
  3966. xml_memory::deallocate(chunk);
  3967. chunk = next_;
  3968. }
  3969. }
  3970. xml_stream_chunk(): next(NULL), size(0)
  3971. {
  3972. }
  3973. xml_stream_chunk* next;
  3974. size_t size;
  3975. T data[xml_memory_page_size / sizeof(T)];
  3976. };
  3977. template <typename T> PUGI_IMPL_FN xml_parse_status load_stream_data_noseek(std::basic_istream<T>& stream, void** out_buffer, size_t* out_size)
  3978. {
  3979. auto_deleter<xml_stream_chunk<T> > chunks(NULL, xml_stream_chunk<T>::destroy);
  3980. // read file to a chunk list
  3981. size_t total = 0;
  3982. xml_stream_chunk<T>* last = NULL;
  3983. while (!stream.eof())
  3984. {
  3985. // allocate new chunk
  3986. xml_stream_chunk<T>* chunk = xml_stream_chunk<T>::create();
  3987. if (!chunk) return status_out_of_memory;
  3988. // append chunk to list
  3989. if (last) last = last->next = chunk;
  3990. else chunks.data = last = chunk;
  3991. // read data to chunk
  3992. stream.read(chunk->data, static_cast<std::streamsize>(sizeof(chunk->data) / sizeof(T)));
  3993. chunk->size = static_cast<size_t>(stream.gcount()) * sizeof(T);
  3994. // read may set failbit | eofbit in case gcount() is less than read length, so check for other I/O errors
  3995. if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error;
  3996. // guard against huge files (chunk size is small enough to make this overflow check work)
  3997. if (total + chunk->size < total) return status_out_of_memory;
  3998. total += chunk->size;
  3999. }
  4000. size_t max_suffix_size = sizeof(char_t);
  4001. // copy chunk list to a contiguous buffer
  4002. char* buffer = static_cast<char*>(xml_memory::allocate(total + max_suffix_size));
  4003. if (!buffer) return status_out_of_memory;
  4004. char* write = buffer;
  4005. for (xml_stream_chunk<T>* chunk = chunks.data; chunk; chunk = chunk->next)
  4006. {
  4007. assert(write + chunk->size <= buffer + total);
  4008. memcpy(write, chunk->data, chunk->size);
  4009. write += chunk->size;
  4010. }
  4011. assert(write == buffer + total);
  4012. // return buffer
  4013. *out_buffer = buffer;
  4014. *out_size = total;
  4015. return status_ok;
  4016. }
  4017. template <typename T> PUGI_IMPL_FN xml_parse_status load_stream_data_seek(std::basic_istream<T>& stream, void** out_buffer, size_t* out_size)
  4018. {
  4019. // get length of remaining data in stream
  4020. typename std::basic_istream<T>::pos_type pos = stream.tellg();
  4021. stream.seekg(0, std::ios::end);
  4022. std::streamoff length = stream.tellg() - pos;
  4023. stream.seekg(pos);
  4024. if (stream.fail() || pos < 0) return status_io_error;
  4025. // guard against huge files
  4026. size_t read_length = static_cast<size_t>(length);
  4027. if (static_cast<std::streamsize>(read_length) != length || length < 0) return status_out_of_memory;
  4028. size_t max_suffix_size = sizeof(char_t);
  4029. // read stream data into memory (guard against stream exceptions with buffer holder)
  4030. auto_deleter<void> buffer(xml_memory::allocate(read_length * sizeof(T) + max_suffix_size), xml_memory::deallocate);
  4031. if (!buffer.data) return status_out_of_memory;
  4032. stream.read(static_cast<T*>(buffer.data), static_cast<std::streamsize>(read_length));
  4033. // read may set failbit | eofbit in case gcount() is less than read_length (i.e. line ending conversion), so check for other I/O errors
  4034. if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error;
  4035. // return buffer
  4036. size_t actual_length = static_cast<size_t>(stream.gcount());
  4037. assert(actual_length <= read_length);
  4038. *out_buffer = buffer.release();
  4039. *out_size = actual_length * sizeof(T);
  4040. return status_ok;
  4041. }
  4042. template <typename T> PUGI_IMPL_FN xml_parse_result load_stream_impl(xml_document_struct* doc, std::basic_istream<T>& stream, unsigned int options, xml_encoding encoding, char_t** out_buffer)
  4043. {
  4044. void* buffer = NULL;
  4045. size_t size = 0;
  4046. xml_parse_status status = status_ok;
  4047. // if stream has an error bit set, bail out (otherwise tellg() can fail and we'll clear error bits)
  4048. if (stream.fail()) return make_parse_result(status_io_error);
  4049. // load stream to memory (using seek-based implementation if possible, since it's faster and takes less memory)
  4050. if (stream.tellg() < 0)
  4051. {
  4052. stream.clear(); // clear error flags that could be set by a failing tellg
  4053. status = load_stream_data_noseek(stream, &buffer, &size);
  4054. }
  4055. else
  4056. status = load_stream_data_seek(stream, &buffer, &size);
  4057. if (status != status_ok) return make_parse_result(status);
  4058. xml_encoding real_encoding = get_buffer_encoding(encoding, buffer, size);
  4059. return load_buffer_impl(doc, doc, buffer, zero_terminate_buffer(buffer, size, real_encoding), options, real_encoding, true, true, out_buffer);
  4060. }
  4061. #endif
  4062. #if defined(PUGI_IMPL_MSVC_CRT_VERSION) || defined(__BORLANDC__) || (defined(__MINGW32__) && (!defined(__STRICT_ANSI__) || defined(__MINGW64_VERSION_MAJOR)))
  4063. PUGI_IMPL_FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode)
  4064. {
  4065. #ifdef PUGIXML_NO_STL
  4066. // ensure these symbols are consistently referenced to avoid 'unreferenced function' warnings
  4067. // note that generally these functions are used in STL builds, but PUGIXML_NO_STL leaves the only usage in convert_path_heap
  4068. (void)&as_utf8_begin;
  4069. (void)&as_utf8_end;
  4070. (void)&strlength_wide;
  4071. #endif
  4072. #if defined(PUGI_IMPL_MSVC_CRT_VERSION) && PUGI_IMPL_MSVC_CRT_VERSION >= 1400
  4073. FILE* file = NULL;
  4074. return _wfopen_s(&file, path, mode) == 0 ? file : NULL;
  4075. #else
  4076. return _wfopen(path, mode);
  4077. #endif
  4078. }
  4079. #else
  4080. PUGI_IMPL_FN char* convert_path_heap(const wchar_t* str)
  4081. {
  4082. assert(str);
  4083. // first pass: get length in utf8 characters
  4084. size_t length = strlength_wide(str);
  4085. size_t size = as_utf8_begin(str, length);
  4086. // allocate resulting string
  4087. char* result = static_cast<char*>(xml_memory::allocate(size + 1));
  4088. if (!result) return NULL;
  4089. // second pass: convert to utf8
  4090. as_utf8_end(result, size, str, length);
  4091. // zero-terminate
  4092. result[size] = 0;
  4093. return result;
  4094. }
  4095. PUGI_IMPL_FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode)
  4096. {
  4097. // there is no standard function to open wide paths, so our best bet is to try utf8 path
  4098. char* path_utf8 = convert_path_heap(path);
  4099. if (!path_utf8) return NULL;
  4100. // convert mode to ASCII (we mirror _wfopen interface)
  4101. char mode_ascii[4] = {0};
  4102. for (size_t i = 0; mode[i]; ++i) mode_ascii[i] = static_cast<char>(mode[i]);
  4103. // try to open the utf8 path
  4104. FILE* result = fopen(path_utf8, mode_ascii);
  4105. // free dummy buffer
  4106. xml_memory::deallocate(path_utf8);
  4107. return result;
  4108. }
  4109. #endif
  4110. PUGI_IMPL_FN FILE* open_file(const char* path, const char* mode)
  4111. {
  4112. #if defined(PUGI_IMPL_MSVC_CRT_VERSION) && PUGI_IMPL_MSVC_CRT_VERSION >= 1400
  4113. FILE* file = NULL;
  4114. return fopen_s(&file, path, mode) == 0 ? file : NULL;
  4115. #else
  4116. return fopen(path, mode);
  4117. #endif
  4118. }
  4119. PUGI_IMPL_FN bool save_file_impl(const xml_document& doc, FILE* file, const char_t* indent, unsigned int flags, xml_encoding encoding)
  4120. {
  4121. if (!file) return false;
  4122. xml_writer_file writer(file);
  4123. doc.save(writer, indent, flags, encoding);
  4124. return fflush(file) == 0 && ferror(file) == 0;
  4125. }
  4126. struct name_null_sentry
  4127. {
  4128. xml_node_struct* node;
  4129. char_t* name;
  4130. name_null_sentry(xml_node_struct* node_): node(node_), name(node_->name)
  4131. {
  4132. node->name = NULL;
  4133. }
  4134. ~name_null_sentry()
  4135. {
  4136. node->name = name;
  4137. }
  4138. };
  4139. PUGI_IMPL_NS_END
  4140. namespace pugi
  4141. {
  4142. PUGI_IMPL_FN xml_writer::~xml_writer()
  4143. {
  4144. }
  4145. PUGI_IMPL_FN xml_writer_file::xml_writer_file(void* file_): file(file_)
  4146. {
  4147. }
  4148. PUGI_IMPL_FN void xml_writer_file::write(const void* data, size_t size)
  4149. {
  4150. size_t result = fwrite(data, 1, size, static_cast<FILE*>(file));
  4151. (void)!result; // unfortunately we can't do proper error handling here
  4152. }
  4153. #ifndef PUGIXML_NO_STL
  4154. PUGI_IMPL_FN xml_writer_stream::xml_writer_stream(std::basic_ostream<char>& stream): narrow_stream(&stream), wide_stream(NULL)
  4155. {
  4156. }
  4157. PUGI_IMPL_FN xml_writer_stream::xml_writer_stream(std::basic_ostream<wchar_t>& stream): narrow_stream(NULL), wide_stream(&stream)
  4158. {
  4159. }
  4160. PUGI_IMPL_FN void xml_writer_stream::write(const void* data, size_t size)
  4161. {
  4162. if (narrow_stream)
  4163. {
  4164. assert(!wide_stream);
  4165. narrow_stream->write(reinterpret_cast<const char*>(data), static_cast<std::streamsize>(size));
  4166. }
  4167. else
  4168. {
  4169. assert(wide_stream);
  4170. assert(size % sizeof(wchar_t) == 0);
  4171. wide_stream->write(reinterpret_cast<const wchar_t*>(data), static_cast<std::streamsize>(size / sizeof(wchar_t)));
  4172. }
  4173. }
  4174. #endif
  4175. PUGI_IMPL_FN xml_tree_walker::xml_tree_walker(): _depth(0)
  4176. {
  4177. }
  4178. PUGI_IMPL_FN xml_tree_walker::~xml_tree_walker()
  4179. {
  4180. }
  4181. PUGI_IMPL_FN int xml_tree_walker::depth() const
  4182. {
  4183. return _depth;
  4184. }
  4185. PUGI_IMPL_FN bool xml_tree_walker::begin(xml_node&)
  4186. {
  4187. return true;
  4188. }
  4189. PUGI_IMPL_FN bool xml_tree_walker::end(xml_node&)
  4190. {
  4191. return true;
  4192. }
  4193. PUGI_IMPL_FN xml_attribute::xml_attribute(): _attr(NULL)
  4194. {
  4195. }
  4196. PUGI_IMPL_FN xml_attribute::xml_attribute(xml_attribute_struct* attr): _attr(attr)
  4197. {
  4198. }
  4199. PUGI_IMPL_FN static void unspecified_bool_xml_attribute(xml_attribute***)
  4200. {
  4201. }
  4202. PUGI_IMPL_FN xml_attribute::operator xml_attribute::unspecified_bool_type() const
  4203. {
  4204. return _attr ? unspecified_bool_xml_attribute : NULL;
  4205. }
  4206. PUGI_IMPL_FN bool xml_attribute::operator!() const
  4207. {
  4208. return !_attr;
  4209. }
  4210. PUGI_IMPL_FN bool xml_attribute::operator==(const xml_attribute& r) const
  4211. {
  4212. return (_attr == r._attr);
  4213. }
  4214. PUGI_IMPL_FN bool xml_attribute::operator!=(const xml_attribute& r) const
  4215. {
  4216. return (_attr != r._attr);
  4217. }
  4218. PUGI_IMPL_FN bool xml_attribute::operator<(const xml_attribute& r) const
  4219. {
  4220. return (_attr < r._attr);
  4221. }
  4222. PUGI_IMPL_FN bool xml_attribute::operator>(const xml_attribute& r) const
  4223. {
  4224. return (_attr > r._attr);
  4225. }
  4226. PUGI_IMPL_FN bool xml_attribute::operator<=(const xml_attribute& r) const
  4227. {
  4228. return (_attr <= r._attr);
  4229. }
  4230. PUGI_IMPL_FN bool xml_attribute::operator>=(const xml_attribute& r) const
  4231. {
  4232. return (_attr >= r._attr);
  4233. }
  4234. PUGI_IMPL_FN xml_attribute xml_attribute::next_attribute() const
  4235. {
  4236. if (!_attr) return xml_attribute();
  4237. return xml_attribute(_attr->next_attribute);
  4238. }
  4239. PUGI_IMPL_FN xml_attribute xml_attribute::previous_attribute() const
  4240. {
  4241. if (!_attr) return xml_attribute();
  4242. xml_attribute_struct* prev = _attr->prev_attribute_c;
  4243. return prev->next_attribute ? xml_attribute(prev) : xml_attribute();
  4244. }
  4245. PUGI_IMPL_FN const char_t* xml_attribute::as_string(const char_t* def) const
  4246. {
  4247. if (!_attr) return def;
  4248. const char_t* value = _attr->value;
  4249. return value ? value : def;
  4250. }
  4251. PUGI_IMPL_FN int xml_attribute::as_int(int def) const
  4252. {
  4253. if (!_attr) return def;
  4254. const char_t* value = _attr->value;
  4255. return value ? impl::get_value_int(value) : def;
  4256. }
  4257. PUGI_IMPL_FN unsigned int xml_attribute::as_uint(unsigned int def) const
  4258. {
  4259. if (!_attr) return def;
  4260. const char_t* value = _attr->value;
  4261. return value ? impl::get_value_uint(value) : def;
  4262. }
  4263. PUGI_IMPL_FN double xml_attribute::as_double(double def) const
  4264. {
  4265. if (!_attr) return def;
  4266. const char_t* value = _attr->value;
  4267. return value ? impl::get_value_double(value) : def;
  4268. }
  4269. PUGI_IMPL_FN float xml_attribute::as_float(float def) const
  4270. {
  4271. if (!_attr) return def;
  4272. const char_t* value = _attr->value;
  4273. return value ? impl::get_value_float(value) : def;
  4274. }
  4275. PUGI_IMPL_FN bool xml_attribute::as_bool(bool def) const
  4276. {
  4277. if (!_attr) return def;
  4278. const char_t* value = _attr->value;
  4279. return value ? impl::get_value_bool(value) : def;
  4280. }
  4281. #ifdef PUGIXML_HAS_LONG_LONG
  4282. PUGI_IMPL_FN long long xml_attribute::as_llong(long long def) const
  4283. {
  4284. if (!_attr) return def;
  4285. const char_t* value = _attr->value;
  4286. return value ? impl::get_value_llong(value) : def;
  4287. }
  4288. PUGI_IMPL_FN unsigned long long xml_attribute::as_ullong(unsigned long long def) const
  4289. {
  4290. if (!_attr) return def;
  4291. const char_t* value = _attr->value;
  4292. return value ? impl::get_value_ullong(value) : def;
  4293. }
  4294. #endif
  4295. PUGI_IMPL_FN bool xml_attribute::empty() const
  4296. {
  4297. return !_attr;
  4298. }
  4299. PUGI_IMPL_FN const char_t* xml_attribute::name() const
  4300. {
  4301. if (!_attr) return PUGIXML_TEXT("");
  4302. const char_t* name = _attr->name;
  4303. return name ? name : PUGIXML_TEXT("");
  4304. }
  4305. PUGI_IMPL_FN const char_t* xml_attribute::value() const
  4306. {
  4307. if (!_attr) return PUGIXML_TEXT("");
  4308. const char_t* value = _attr->value;
  4309. return value ? value : PUGIXML_TEXT("");
  4310. }
  4311. PUGI_IMPL_FN size_t xml_attribute::hash_value() const
  4312. {
  4313. return reinterpret_cast<uintptr_t>(_attr) / sizeof(xml_attribute_struct);
  4314. }
  4315. PUGI_IMPL_FN xml_attribute_struct* xml_attribute::internal_object() const
  4316. {
  4317. return _attr;
  4318. }
  4319. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(const char_t* rhs)
  4320. {
  4321. set_value(rhs);
  4322. return *this;
  4323. }
  4324. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(int rhs)
  4325. {
  4326. set_value(rhs);
  4327. return *this;
  4328. }
  4329. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(unsigned int rhs)
  4330. {
  4331. set_value(rhs);
  4332. return *this;
  4333. }
  4334. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(long rhs)
  4335. {
  4336. set_value(rhs);
  4337. return *this;
  4338. }
  4339. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(unsigned long rhs)
  4340. {
  4341. set_value(rhs);
  4342. return *this;
  4343. }
  4344. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(double rhs)
  4345. {
  4346. set_value(rhs);
  4347. return *this;
  4348. }
  4349. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(float rhs)
  4350. {
  4351. set_value(rhs);
  4352. return *this;
  4353. }
  4354. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(bool rhs)
  4355. {
  4356. set_value(rhs);
  4357. return *this;
  4358. }
  4359. #ifdef PUGIXML_HAS_STRING_VIEW
  4360. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(string_view_t rhs)
  4361. {
  4362. set_value(rhs);
  4363. return *this;
  4364. }
  4365. #endif
  4366. #ifdef PUGIXML_HAS_LONG_LONG
  4367. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(long long rhs)
  4368. {
  4369. set_value(rhs);
  4370. return *this;
  4371. }
  4372. PUGI_IMPL_FN xml_attribute& xml_attribute::operator=(unsigned long long rhs)
  4373. {
  4374. set_value(rhs);
  4375. return *this;
  4376. }
  4377. #endif
  4378. PUGI_IMPL_FN bool xml_attribute::set_name(const char_t* rhs)
  4379. {
  4380. if (!_attr) return false;
  4381. return impl::strcpy_insitu(_attr->name, _attr->header, impl::xml_memory_page_name_allocated_mask, rhs, impl::strlength(rhs));
  4382. }
  4383. PUGI_IMPL_FN bool xml_attribute::set_name(const char_t* rhs, size_t size)
  4384. {
  4385. if (!_attr) return false;
  4386. return impl::strcpy_insitu(_attr->name, _attr->header, impl::xml_memory_page_name_allocated_mask, rhs, size);
  4387. }
  4388. #ifdef PUGIXML_HAS_STRING_VIEW
  4389. PUGI_IMPL_FN bool xml_attribute::set_name(string_view_t rhs)
  4390. {
  4391. if (!_attr) return false;
  4392. return impl::strcpy_insitu(_attr->name, _attr->header, impl::xml_memory_page_name_allocated_mask, rhs.data(), rhs.size());
  4393. }
  4394. #endif
  4395. PUGI_IMPL_FN bool xml_attribute::set_value(const char_t* rhs)
  4396. {
  4397. if (!_attr) return false;
  4398. return impl::strcpy_insitu(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, impl::strlength(rhs));
  4399. }
  4400. PUGI_IMPL_FN bool xml_attribute::set_value(const char_t* rhs, size_t size)
  4401. {
  4402. if (!_attr) return false;
  4403. return impl::strcpy_insitu(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, size);
  4404. }
  4405. #ifdef PUGIXML_HAS_STRING_VIEW
  4406. PUGI_IMPL_FN bool xml_attribute::set_value(string_view_t rhs)
  4407. {
  4408. if (!_attr) return false;
  4409. return impl::strcpy_insitu(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs.data(), rhs.size());
  4410. }
  4411. #endif
  4412. PUGI_IMPL_FN bool xml_attribute::set_value(int rhs)
  4413. {
  4414. if (!_attr) return false;
  4415. return impl::set_value_integer<unsigned int>(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0);
  4416. }
  4417. PUGI_IMPL_FN bool xml_attribute::set_value(unsigned int rhs)
  4418. {
  4419. if (!_attr) return false;
  4420. return impl::set_value_integer<unsigned int>(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, false);
  4421. }
  4422. PUGI_IMPL_FN bool xml_attribute::set_value(long rhs)
  4423. {
  4424. if (!_attr) return false;
  4425. return impl::set_value_integer<unsigned long>(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0);
  4426. }
  4427. PUGI_IMPL_FN bool xml_attribute::set_value(unsigned long rhs)
  4428. {
  4429. if (!_attr) return false;
  4430. return impl::set_value_integer<unsigned long>(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, false);
  4431. }
  4432. PUGI_IMPL_FN bool xml_attribute::set_value(double rhs)
  4433. {
  4434. if (!_attr) return false;
  4435. return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, default_double_precision);
  4436. }
  4437. PUGI_IMPL_FN bool xml_attribute::set_value(double rhs, int precision)
  4438. {
  4439. if (!_attr) return false;
  4440. return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, precision);
  4441. }
  4442. PUGI_IMPL_FN bool xml_attribute::set_value(float rhs)
  4443. {
  4444. if (!_attr) return false;
  4445. return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, default_float_precision);
  4446. }
  4447. PUGI_IMPL_FN bool xml_attribute::set_value(float rhs, int precision)
  4448. {
  4449. if (!_attr) return false;
  4450. return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, precision);
  4451. }
  4452. PUGI_IMPL_FN bool xml_attribute::set_value(bool rhs)
  4453. {
  4454. if (!_attr) return false;
  4455. return impl::set_value_bool(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
  4456. }
  4457. #ifdef PUGIXML_HAS_LONG_LONG
  4458. PUGI_IMPL_FN bool xml_attribute::set_value(long long rhs)
  4459. {
  4460. if (!_attr) return false;
  4461. return impl::set_value_integer<unsigned long long>(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0);
  4462. }
  4463. PUGI_IMPL_FN bool xml_attribute::set_value(unsigned long long rhs)
  4464. {
  4465. if (!_attr) return false;
  4466. return impl::set_value_integer<unsigned long long>(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, false);
  4467. }
  4468. #endif
  4469. #ifdef __BORLANDC__
  4470. PUGI_IMPL_FN bool operator&&(const xml_attribute& lhs, bool rhs)
  4471. {
  4472. return (bool)lhs && rhs;
  4473. }
  4474. PUGI_IMPL_FN bool operator||(const xml_attribute& lhs, bool rhs)
  4475. {
  4476. return (bool)lhs || rhs;
  4477. }
  4478. #endif
  4479. PUGI_IMPL_FN xml_node::xml_node(): _root(NULL)
  4480. {
  4481. }
  4482. PUGI_IMPL_FN xml_node::xml_node(xml_node_struct* p): _root(p)
  4483. {
  4484. }
  4485. PUGI_IMPL_FN static void unspecified_bool_xml_node(xml_node***)
  4486. {
  4487. }
  4488. PUGI_IMPL_FN xml_node::operator xml_node::unspecified_bool_type() const
  4489. {
  4490. return _root ? unspecified_bool_xml_node : NULL;
  4491. }
  4492. PUGI_IMPL_FN bool xml_node::operator!() const
  4493. {
  4494. return !_root;
  4495. }
  4496. PUGI_IMPL_FN xml_node::iterator xml_node::begin() const
  4497. {
  4498. return iterator(_root ? _root->first_child + 0 : NULL, _root);
  4499. }
  4500. PUGI_IMPL_FN xml_node::iterator xml_node::end() const
  4501. {
  4502. return iterator(NULL, _root);
  4503. }
  4504. PUGI_IMPL_FN xml_node::attribute_iterator xml_node::attributes_begin() const
  4505. {
  4506. return attribute_iterator(_root ? _root->first_attribute + 0 : NULL, _root);
  4507. }
  4508. PUGI_IMPL_FN xml_node::attribute_iterator xml_node::attributes_end() const
  4509. {
  4510. return attribute_iterator(NULL, _root);
  4511. }
  4512. PUGI_IMPL_FN xml_object_range<xml_node_iterator> xml_node::children() const
  4513. {
  4514. return xml_object_range<xml_node_iterator>(begin(), end());
  4515. }
  4516. PUGI_IMPL_FN xml_object_range<xml_named_node_iterator> xml_node::children(const char_t* name_) const
  4517. {
  4518. return xml_object_range<xml_named_node_iterator>(xml_named_node_iterator(child(name_)._root, _root, name_), xml_named_node_iterator(NULL, _root, name_));
  4519. }
  4520. PUGI_IMPL_FN xml_object_range<xml_attribute_iterator> xml_node::attributes() const
  4521. {
  4522. return xml_object_range<xml_attribute_iterator>(attributes_begin(), attributes_end());
  4523. }
  4524. PUGI_IMPL_FN bool xml_node::operator==(const xml_node& r) const
  4525. {
  4526. return (_root == r._root);
  4527. }
  4528. PUGI_IMPL_FN bool xml_node::operator!=(const xml_node& r) const
  4529. {
  4530. return (_root != r._root);
  4531. }
  4532. PUGI_IMPL_FN bool xml_node::operator<(const xml_node& r) const
  4533. {
  4534. return (_root < r._root);
  4535. }
  4536. PUGI_IMPL_FN bool xml_node::operator>(const xml_node& r) const
  4537. {
  4538. return (_root > r._root);
  4539. }
  4540. PUGI_IMPL_FN bool xml_node::operator<=(const xml_node& r) const
  4541. {
  4542. return (_root <= r._root);
  4543. }
  4544. PUGI_IMPL_FN bool xml_node::operator>=(const xml_node& r) const
  4545. {
  4546. return (_root >= r._root);
  4547. }
  4548. PUGI_IMPL_FN bool xml_node::empty() const
  4549. {
  4550. return !_root;
  4551. }
  4552. PUGI_IMPL_FN const char_t* xml_node::name() const
  4553. {
  4554. if (!_root) return PUGIXML_TEXT("");
  4555. const char_t* name = _root->name;
  4556. return name ? name : PUGIXML_TEXT("");
  4557. }
  4558. PUGI_IMPL_FN xml_node_type xml_node::type() const
  4559. {
  4560. return _root ? PUGI_IMPL_NODETYPE(_root) : node_null;
  4561. }
  4562. PUGI_IMPL_FN const char_t* xml_node::value() const
  4563. {
  4564. if (!_root) return PUGIXML_TEXT("");
  4565. const char_t* value = _root->value;
  4566. return value ? value : PUGIXML_TEXT("");
  4567. }
  4568. PUGI_IMPL_FN xml_node xml_node::child(const char_t* name_) const
  4569. {
  4570. if (!_root) return xml_node();
  4571. for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
  4572. {
  4573. const char_t* iname = i->name;
  4574. if (iname && impl::strequal(name_, iname))
  4575. return xml_node(i);
  4576. }
  4577. return xml_node();
  4578. }
  4579. PUGI_IMPL_FN xml_attribute xml_node::attribute(const char_t* name_) const
  4580. {
  4581. if (!_root) return xml_attribute();
  4582. for (xml_attribute_struct* i = _root->first_attribute; i; i = i->next_attribute)
  4583. {
  4584. const char_t* iname = i->name;
  4585. if (iname && impl::strequal(name_, iname))
  4586. return xml_attribute(i);
  4587. }
  4588. return xml_attribute();
  4589. }
  4590. PUGI_IMPL_FN xml_node xml_node::next_sibling(const char_t* name_) const
  4591. {
  4592. if (!_root) return xml_node();
  4593. for (xml_node_struct* i = _root->next_sibling; i; i = i->next_sibling)
  4594. {
  4595. const char_t* iname = i->name;
  4596. if (iname && impl::strequal(name_, iname))
  4597. return xml_node(i);
  4598. }
  4599. return xml_node();
  4600. }
  4601. PUGI_IMPL_FN xml_node xml_node::next_sibling() const
  4602. {
  4603. return _root ? xml_node(_root->next_sibling) : xml_node();
  4604. }
  4605. PUGI_IMPL_FN xml_node xml_node::previous_sibling(const char_t* name_) const
  4606. {
  4607. if (!_root) return xml_node();
  4608. for (xml_node_struct* i = _root->prev_sibling_c; i->next_sibling; i = i->prev_sibling_c)
  4609. {
  4610. const char_t* iname = i->name;
  4611. if (iname && impl::strequal(name_, iname))
  4612. return xml_node(i);
  4613. }
  4614. return xml_node();
  4615. }
  4616. #ifdef PUGIXML_HAS_STRING_VIEW
  4617. PUGI_IMPL_FN xml_node xml_node::child(string_view_t name_) const
  4618. {
  4619. if (!_root) return xml_node();
  4620. for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
  4621. {
  4622. const char_t* iname = i->name;
  4623. if (iname && impl::stringview_equal(name_, iname))
  4624. return xml_node(i);
  4625. }
  4626. return xml_node();
  4627. }
  4628. PUGI_IMPL_FN xml_attribute xml_node::attribute(string_view_t name_) const
  4629. {
  4630. if (!_root) return xml_attribute();
  4631. for (xml_attribute_struct* i = _root->first_attribute; i; i = i->next_attribute)
  4632. {
  4633. const char_t* iname = i->name;
  4634. if (iname && impl::stringview_equal(name_, iname))
  4635. return xml_attribute(i);
  4636. }
  4637. return xml_attribute();
  4638. }
  4639. PUGI_IMPL_FN xml_node xml_node::next_sibling(string_view_t name_) const
  4640. {
  4641. if (!_root) return xml_node();
  4642. for (xml_node_struct* i = _root->next_sibling; i; i = i->next_sibling)
  4643. {
  4644. const char_t* iname = i->name;
  4645. if (iname && impl::stringview_equal(name_, iname))
  4646. return xml_node(i);
  4647. }
  4648. return xml_node();
  4649. }
  4650. PUGI_IMPL_FN xml_node xml_node::previous_sibling(string_view_t name_) const
  4651. {
  4652. if (!_root) return xml_node();
  4653. for (xml_node_struct* i = _root->prev_sibling_c; i->next_sibling; i = i->prev_sibling_c)
  4654. {
  4655. const char_t* iname = i->name;
  4656. if (iname && impl::stringview_equal(name_, iname))
  4657. return xml_node(i);
  4658. }
  4659. return xml_node();
  4660. }
  4661. #endif
  4662. PUGI_IMPL_FN xml_attribute xml_node::attribute(const char_t* name_, xml_attribute& hint_) const
  4663. {
  4664. xml_attribute_struct* hint = hint_._attr;
  4665. // if hint is not an attribute of node, behavior is not defined
  4666. assert(!hint || (_root && impl::is_attribute_of(hint, _root)));
  4667. if (!_root) return xml_attribute();
  4668. // optimistically search from hint up until the end
  4669. for (xml_attribute_struct* i = hint; i; i = i->next_attribute)
  4670. {
  4671. const char_t* iname = i->name;
  4672. if (iname && impl::strequal(name_, iname))
  4673. {
  4674. // update hint to maximize efficiency of searching for consecutive attributes
  4675. hint_._attr = i->next_attribute;
  4676. return xml_attribute(i);
  4677. }
  4678. }
  4679. // wrap around and search from the first attribute until the hint
  4680. // 'j' null pointer check is technically redundant, but it prevents a crash in case the assertion above fails
  4681. for (xml_attribute_struct* j = _root->first_attribute; j && j != hint; j = j->next_attribute)
  4682. {
  4683. const char_t* jname = j->name;
  4684. if (jname && impl::strequal(name_, jname))
  4685. {
  4686. // update hint to maximize efficiency of searching for consecutive attributes
  4687. hint_._attr = j->next_attribute;
  4688. return xml_attribute(j);
  4689. }
  4690. }
  4691. return xml_attribute();
  4692. }
  4693. #ifdef PUGIXML_HAS_STRING_VIEW
  4694. PUGI_IMPL_FN xml_attribute xml_node::attribute(string_view_t name_, xml_attribute& hint_) const
  4695. {
  4696. xml_attribute_struct* hint = hint_._attr;
  4697. // if hint is not an attribute of node, behavior is not defined
  4698. assert(!hint || (_root && impl::is_attribute_of(hint, _root)));
  4699. if (!_root) return xml_attribute();
  4700. // optimistically search from hint up until the end
  4701. for (xml_attribute_struct* i = hint; i; i = i->next_attribute)
  4702. {
  4703. const char_t* iname = i->name;
  4704. if (iname && impl::stringview_equal(name_, iname))
  4705. {
  4706. // update hint to maximize efficiency of searching for consecutive attributes
  4707. hint_._attr = i->next_attribute;
  4708. return xml_attribute(i);
  4709. }
  4710. }
  4711. // wrap around and search from the first attribute until the hint
  4712. // 'j' null pointer check is technically redundant, but it prevents a crash in case the assertion above fails
  4713. for (xml_attribute_struct* j = _root->first_attribute; j && j != hint; j = j->next_attribute)
  4714. {
  4715. const char_t* jname = j->name;
  4716. if (jname && impl::stringview_equal(name_, jname))
  4717. {
  4718. // update hint to maximize efficiency of searching for consecutive attributes
  4719. hint_._attr = j->next_attribute;
  4720. return xml_attribute(j);
  4721. }
  4722. }
  4723. return xml_attribute();
  4724. }
  4725. #endif
  4726. PUGI_IMPL_FN xml_node xml_node::previous_sibling() const
  4727. {
  4728. if (!_root) return xml_node();
  4729. xml_node_struct* prev = _root->prev_sibling_c;
  4730. return prev->next_sibling ? xml_node(prev) : xml_node();
  4731. }
  4732. PUGI_IMPL_FN xml_node xml_node::parent() const
  4733. {
  4734. return _root ? xml_node(_root->parent) : xml_node();
  4735. }
  4736. PUGI_IMPL_FN xml_node xml_node::root() const
  4737. {
  4738. return _root ? xml_node(&impl::get_document(_root)) : xml_node();
  4739. }
  4740. PUGI_IMPL_FN xml_text xml_node::text() const
  4741. {
  4742. return xml_text(_root);
  4743. }
  4744. PUGI_IMPL_FN const char_t* xml_node::child_value() const
  4745. {
  4746. if (!_root) return PUGIXML_TEXT("");
  4747. // element nodes can have value if parse_embed_pcdata was used
  4748. if (PUGI_IMPL_NODETYPE(_root) == node_element && _root->value)
  4749. return _root->value;
  4750. for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
  4751. {
  4752. const char_t* ivalue = i->value;
  4753. if (impl::is_text_node(i) && ivalue)
  4754. return ivalue;
  4755. }
  4756. return PUGIXML_TEXT("");
  4757. }
  4758. PUGI_IMPL_FN const char_t* xml_node::child_value(const char_t* name_) const
  4759. {
  4760. return child(name_).child_value();
  4761. }
  4762. PUGI_IMPL_FN xml_attribute xml_node::first_attribute() const
  4763. {
  4764. if (!_root) return xml_attribute();
  4765. return xml_attribute(_root->first_attribute);
  4766. }
  4767. PUGI_IMPL_FN xml_attribute xml_node::last_attribute() const
  4768. {
  4769. if (!_root) return xml_attribute();
  4770. xml_attribute_struct* first = _root->first_attribute;
  4771. return first ? xml_attribute(first->prev_attribute_c) : xml_attribute();
  4772. }
  4773. PUGI_IMPL_FN xml_node xml_node::first_child() const
  4774. {
  4775. if (!_root) return xml_node();
  4776. return xml_node(_root->first_child);
  4777. }
  4778. PUGI_IMPL_FN xml_node xml_node::last_child() const
  4779. {
  4780. if (!_root) return xml_node();
  4781. xml_node_struct* first = _root->first_child;
  4782. return first ? xml_node(first->prev_sibling_c) : xml_node();
  4783. }
  4784. PUGI_IMPL_FN bool xml_node::set_name(const char_t* rhs)
  4785. {
  4786. xml_node_type type_ = _root ? PUGI_IMPL_NODETYPE(_root) : node_null;
  4787. if (type_ != node_element && type_ != node_pi && type_ != node_declaration)
  4788. return false;
  4789. return impl::strcpy_insitu(_root->name, _root->header, impl::xml_memory_page_name_allocated_mask, rhs, impl::strlength(rhs));
  4790. }
  4791. PUGI_IMPL_FN bool xml_node::set_name(const char_t* rhs, size_t size)
  4792. {
  4793. xml_node_type type_ = _root ? PUGI_IMPL_NODETYPE(_root) : node_null;
  4794. if (type_ != node_element && type_ != node_pi && type_ != node_declaration)
  4795. return false;
  4796. return impl::strcpy_insitu(_root->name, _root->header, impl::xml_memory_page_name_allocated_mask, rhs, size);
  4797. }
  4798. #ifdef PUGIXML_HAS_STRING_VIEW
  4799. PUGI_IMPL_FN bool xml_node::set_name(string_view_t rhs)
  4800. {
  4801. xml_node_type type_ = _root ? PUGI_IMPL_NODETYPE(_root) : node_null;
  4802. if (type_ != node_element && type_ != node_pi && type_ != node_declaration)
  4803. return false;
  4804. return impl::strcpy_insitu(_root->name, _root->header, impl::xml_memory_page_name_allocated_mask, rhs.data(), rhs.size());
  4805. }
  4806. #endif
  4807. PUGI_IMPL_FN bool xml_node::set_value(const char_t* rhs)
  4808. {
  4809. xml_node_type type_ = _root ? PUGI_IMPL_NODETYPE(_root) : node_null;
  4810. if (type_ != node_pcdata && type_ != node_cdata && type_ != node_comment && type_ != node_pi && type_ != node_doctype)
  4811. return false;
  4812. return impl::strcpy_insitu(_root->value, _root->header, impl::xml_memory_page_value_allocated_mask, rhs, impl::strlength(rhs));
  4813. }
  4814. PUGI_IMPL_FN bool xml_node::set_value(const char_t* rhs, size_t size)
  4815. {
  4816. xml_node_type type_ = _root ? PUGI_IMPL_NODETYPE(_root) : node_null;
  4817. if (type_ != node_pcdata && type_ != node_cdata && type_ != node_comment && type_ != node_pi && type_ != node_doctype)
  4818. return false;
  4819. return impl::strcpy_insitu(_root->value, _root->header, impl::xml_memory_page_value_allocated_mask, rhs, size);
  4820. }
  4821. #ifdef PUGIXML_HAS_STRING_VIEW
  4822. PUGI_IMPL_FN bool xml_node::set_value(string_view_t rhs)
  4823. {
  4824. xml_node_type type_ = _root ? PUGI_IMPL_NODETYPE(_root) : node_null;
  4825. if (type_ != node_pcdata && type_ != node_cdata && type_ != node_comment && type_ != node_pi && type_ != node_doctype)
  4826. return false;
  4827. return impl::strcpy_insitu(_root->value, _root->header, impl::xml_memory_page_value_allocated_mask, rhs.data(), rhs.size());
  4828. }
  4829. #endif
  4830. PUGI_IMPL_FN xml_attribute xml_node::append_attribute(const char_t* name_)
  4831. {
  4832. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4833. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4834. if (!alloc.reserve()) return xml_attribute();
  4835. xml_attribute a(impl::allocate_attribute(alloc));
  4836. if (!a) return xml_attribute();
  4837. impl::append_attribute(a._attr, _root);
  4838. a.set_name(name_);
  4839. return a;
  4840. }
  4841. PUGI_IMPL_FN xml_attribute xml_node::prepend_attribute(const char_t* name_)
  4842. {
  4843. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4844. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4845. if (!alloc.reserve()) return xml_attribute();
  4846. xml_attribute a(impl::allocate_attribute(alloc));
  4847. if (!a) return xml_attribute();
  4848. impl::prepend_attribute(a._attr, _root);
  4849. a.set_name(name_);
  4850. return a;
  4851. }
  4852. PUGI_IMPL_FN xml_attribute xml_node::insert_attribute_after(const char_t* name_, const xml_attribute& attr)
  4853. {
  4854. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4855. if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
  4856. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4857. if (!alloc.reserve()) return xml_attribute();
  4858. xml_attribute a(impl::allocate_attribute(alloc));
  4859. if (!a) return xml_attribute();
  4860. impl::insert_attribute_after(a._attr, attr._attr, _root);
  4861. a.set_name(name_);
  4862. return a;
  4863. }
  4864. PUGI_IMPL_FN xml_attribute xml_node::insert_attribute_before(const char_t* name_, const xml_attribute& attr)
  4865. {
  4866. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4867. if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
  4868. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4869. if (!alloc.reserve()) return xml_attribute();
  4870. xml_attribute a(impl::allocate_attribute(alloc));
  4871. if (!a) return xml_attribute();
  4872. impl::insert_attribute_before(a._attr, attr._attr, _root);
  4873. a.set_name(name_);
  4874. return a;
  4875. }
  4876. #ifdef PUGIXML_HAS_STRING_VIEW
  4877. PUGI_IMPL_FN xml_attribute xml_node::append_attribute(string_view_t name_)
  4878. {
  4879. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4880. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4881. if (!alloc.reserve()) return xml_attribute();
  4882. xml_attribute a(impl::allocate_attribute(alloc));
  4883. if (!a) return xml_attribute();
  4884. impl::append_attribute(a._attr, _root);
  4885. a.set_name(name_);
  4886. return a;
  4887. }
  4888. PUGI_IMPL_FN xml_attribute xml_node::prepend_attribute(string_view_t name_)
  4889. {
  4890. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4891. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4892. if (!alloc.reserve()) return xml_attribute();
  4893. xml_attribute a(impl::allocate_attribute(alloc));
  4894. if (!a) return xml_attribute();
  4895. impl::prepend_attribute(a._attr, _root);
  4896. a.set_name(name_);
  4897. return a;
  4898. }
  4899. PUGI_IMPL_FN xml_attribute xml_node::insert_attribute_after(string_view_t name_, const xml_attribute& attr)
  4900. {
  4901. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4902. if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
  4903. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4904. if (!alloc.reserve()) return xml_attribute();
  4905. xml_attribute a(impl::allocate_attribute(alloc));
  4906. if (!a) return xml_attribute();
  4907. impl::insert_attribute_after(a._attr, attr._attr, _root);
  4908. a.set_name(name_);
  4909. return a;
  4910. }
  4911. PUGI_IMPL_FN xml_attribute xml_node::insert_attribute_before(string_view_t name_, const xml_attribute& attr)
  4912. {
  4913. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4914. if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
  4915. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4916. if (!alloc.reserve()) return xml_attribute();
  4917. xml_attribute a(impl::allocate_attribute(alloc));
  4918. if (!a) return xml_attribute();
  4919. impl::insert_attribute_before(a._attr, attr._attr, _root);
  4920. a.set_name(name_);
  4921. return a;
  4922. }
  4923. #endif
  4924. PUGI_IMPL_FN xml_attribute xml_node::append_copy(const xml_attribute& proto)
  4925. {
  4926. if (!proto) return xml_attribute();
  4927. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4928. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4929. if (!alloc.reserve()) return xml_attribute();
  4930. xml_attribute a(impl::allocate_attribute(alloc));
  4931. if (!a) return xml_attribute();
  4932. impl::append_attribute(a._attr, _root);
  4933. impl::node_copy_attribute(a._attr, proto._attr);
  4934. return a;
  4935. }
  4936. PUGI_IMPL_FN xml_attribute xml_node::prepend_copy(const xml_attribute& proto)
  4937. {
  4938. if (!proto) return xml_attribute();
  4939. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4940. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4941. if (!alloc.reserve()) return xml_attribute();
  4942. xml_attribute a(impl::allocate_attribute(alloc));
  4943. if (!a) return xml_attribute();
  4944. impl::prepend_attribute(a._attr, _root);
  4945. impl::node_copy_attribute(a._attr, proto._attr);
  4946. return a;
  4947. }
  4948. PUGI_IMPL_FN xml_attribute xml_node::insert_copy_after(const xml_attribute& proto, const xml_attribute& attr)
  4949. {
  4950. if (!proto) return xml_attribute();
  4951. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4952. if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
  4953. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4954. if (!alloc.reserve()) return xml_attribute();
  4955. xml_attribute a(impl::allocate_attribute(alloc));
  4956. if (!a) return xml_attribute();
  4957. impl::insert_attribute_after(a._attr, attr._attr, _root);
  4958. impl::node_copy_attribute(a._attr, proto._attr);
  4959. return a;
  4960. }
  4961. PUGI_IMPL_FN xml_attribute xml_node::insert_copy_before(const xml_attribute& proto, const xml_attribute& attr)
  4962. {
  4963. if (!proto) return xml_attribute();
  4964. if (!impl::allow_insert_attribute(type())) return xml_attribute();
  4965. if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
  4966. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4967. if (!alloc.reserve()) return xml_attribute();
  4968. xml_attribute a(impl::allocate_attribute(alloc));
  4969. if (!a) return xml_attribute();
  4970. impl::insert_attribute_before(a._attr, attr._attr, _root);
  4971. impl::node_copy_attribute(a._attr, proto._attr);
  4972. return a;
  4973. }
  4974. PUGI_IMPL_FN xml_node xml_node::append_child(xml_node_type type_)
  4975. {
  4976. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  4977. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4978. if (!alloc.reserve()) return xml_node();
  4979. xml_node n(impl::allocate_node(alloc, type_));
  4980. if (!n) return xml_node();
  4981. impl::append_node(n._root, _root);
  4982. if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
  4983. return n;
  4984. }
  4985. PUGI_IMPL_FN xml_node xml_node::prepend_child(xml_node_type type_)
  4986. {
  4987. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  4988. impl::xml_allocator& alloc = impl::get_allocator(_root);
  4989. if (!alloc.reserve()) return xml_node();
  4990. xml_node n(impl::allocate_node(alloc, type_));
  4991. if (!n) return xml_node();
  4992. impl::prepend_node(n._root, _root);
  4993. if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
  4994. return n;
  4995. }
  4996. PUGI_IMPL_FN xml_node xml_node::insert_child_before(xml_node_type type_, const xml_node& node)
  4997. {
  4998. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  4999. if (!node._root || node._root->parent != _root) return xml_node();
  5000. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5001. if (!alloc.reserve()) return xml_node();
  5002. xml_node n(impl::allocate_node(alloc, type_));
  5003. if (!n) return xml_node();
  5004. impl::insert_node_before(n._root, node._root);
  5005. if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
  5006. return n;
  5007. }
  5008. PUGI_IMPL_FN xml_node xml_node::insert_child_after(xml_node_type type_, const xml_node& node)
  5009. {
  5010. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  5011. if (!node._root || node._root->parent != _root) return xml_node();
  5012. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5013. if (!alloc.reserve()) return xml_node();
  5014. xml_node n(impl::allocate_node(alloc, type_));
  5015. if (!n) return xml_node();
  5016. impl::insert_node_after(n._root, node._root);
  5017. if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
  5018. return n;
  5019. }
  5020. PUGI_IMPL_FN xml_node xml_node::append_child(const char_t* name_)
  5021. {
  5022. xml_node result = append_child(node_element);
  5023. result.set_name(name_);
  5024. return result;
  5025. }
  5026. PUGI_IMPL_FN xml_node xml_node::prepend_child(const char_t* name_)
  5027. {
  5028. xml_node result = prepend_child(node_element);
  5029. result.set_name(name_);
  5030. return result;
  5031. }
  5032. PUGI_IMPL_FN xml_node xml_node::insert_child_after(const char_t* name_, const xml_node& node)
  5033. {
  5034. xml_node result = insert_child_after(node_element, node);
  5035. result.set_name(name_);
  5036. return result;
  5037. }
  5038. PUGI_IMPL_FN xml_node xml_node::insert_child_before(const char_t* name_, const xml_node& node)
  5039. {
  5040. xml_node result = insert_child_before(node_element, node);
  5041. result.set_name(name_);
  5042. return result;
  5043. }
  5044. #ifdef PUGIXML_HAS_STRING_VIEW
  5045. PUGI_IMPL_FN xml_node xml_node::append_child(string_view_t name_)
  5046. {
  5047. xml_node result = append_child(node_element);
  5048. result.set_name(name_);
  5049. return result;
  5050. }
  5051. PUGI_IMPL_FN xml_node xml_node::prepend_child(string_view_t name_)
  5052. {
  5053. xml_node result = prepend_child(node_element);
  5054. result.set_name(name_);
  5055. return result;
  5056. }
  5057. PUGI_IMPL_FN xml_node xml_node::insert_child_after(string_view_t name_, const xml_node& node)
  5058. {
  5059. xml_node result = insert_child_after(node_element, node);
  5060. result.set_name(name_);
  5061. return result;
  5062. }
  5063. PUGI_IMPL_FN xml_node xml_node::insert_child_before(string_view_t name_, const xml_node& node)
  5064. {
  5065. xml_node result = insert_child_before(node_element, node);
  5066. result.set_name(name_);
  5067. return result;
  5068. }
  5069. #endif
  5070. PUGI_IMPL_FN xml_node xml_node::append_copy(const xml_node& proto)
  5071. {
  5072. xml_node_type type_ = proto.type();
  5073. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  5074. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5075. if (!alloc.reserve()) return xml_node();
  5076. xml_node n(impl::allocate_node(alloc, type_));
  5077. if (!n) return xml_node();
  5078. impl::append_node(n._root, _root);
  5079. impl::node_copy_tree(n._root, proto._root);
  5080. return n;
  5081. }
  5082. PUGI_IMPL_FN xml_node xml_node::prepend_copy(const xml_node& proto)
  5083. {
  5084. xml_node_type type_ = proto.type();
  5085. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  5086. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5087. if (!alloc.reserve()) return xml_node();
  5088. xml_node n(impl::allocate_node(alloc, type_));
  5089. if (!n) return xml_node();
  5090. impl::prepend_node(n._root, _root);
  5091. impl::node_copy_tree(n._root, proto._root);
  5092. return n;
  5093. }
  5094. PUGI_IMPL_FN xml_node xml_node::insert_copy_after(const xml_node& proto, const xml_node& node)
  5095. {
  5096. xml_node_type type_ = proto.type();
  5097. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  5098. if (!node._root || node._root->parent != _root) return xml_node();
  5099. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5100. if (!alloc.reserve()) return xml_node();
  5101. xml_node n(impl::allocate_node(alloc, type_));
  5102. if (!n) return xml_node();
  5103. impl::insert_node_after(n._root, node._root);
  5104. impl::node_copy_tree(n._root, proto._root);
  5105. return n;
  5106. }
  5107. PUGI_IMPL_FN xml_node xml_node::insert_copy_before(const xml_node& proto, const xml_node& node)
  5108. {
  5109. xml_node_type type_ = proto.type();
  5110. if (!impl::allow_insert_child(type(), type_)) return xml_node();
  5111. if (!node._root || node._root->parent != _root) return xml_node();
  5112. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5113. if (!alloc.reserve()) return xml_node();
  5114. xml_node n(impl::allocate_node(alloc, type_));
  5115. if (!n) return xml_node();
  5116. impl::insert_node_before(n._root, node._root);
  5117. impl::node_copy_tree(n._root, proto._root);
  5118. return n;
  5119. }
  5120. PUGI_IMPL_FN xml_node xml_node::append_move(const xml_node& moved)
  5121. {
  5122. if (!impl::allow_move(*this, moved)) return xml_node();
  5123. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5124. if (!alloc.reserve()) return xml_node();
  5125. // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
  5126. impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
  5127. impl::remove_node(moved._root);
  5128. impl::append_node(moved._root, _root);
  5129. return moved;
  5130. }
  5131. PUGI_IMPL_FN xml_node xml_node::prepend_move(const xml_node& moved)
  5132. {
  5133. if (!impl::allow_move(*this, moved)) return xml_node();
  5134. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5135. if (!alloc.reserve()) return xml_node();
  5136. // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
  5137. impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
  5138. impl::remove_node(moved._root);
  5139. impl::prepend_node(moved._root, _root);
  5140. return moved;
  5141. }
  5142. PUGI_IMPL_FN xml_node xml_node::insert_move_after(const xml_node& moved, const xml_node& node)
  5143. {
  5144. if (!impl::allow_move(*this, moved)) return xml_node();
  5145. if (!node._root || node._root->parent != _root) return xml_node();
  5146. if (moved._root == node._root) return xml_node();
  5147. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5148. if (!alloc.reserve()) return xml_node();
  5149. // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
  5150. impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
  5151. impl::remove_node(moved._root);
  5152. impl::insert_node_after(moved._root, node._root);
  5153. return moved;
  5154. }
  5155. PUGI_IMPL_FN xml_node xml_node::insert_move_before(const xml_node& moved, const xml_node& node)
  5156. {
  5157. if (!impl::allow_move(*this, moved)) return xml_node();
  5158. if (!node._root || node._root->parent != _root) return xml_node();
  5159. if (moved._root == node._root) return xml_node();
  5160. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5161. if (!alloc.reserve()) return xml_node();
  5162. // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
  5163. impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
  5164. impl::remove_node(moved._root);
  5165. impl::insert_node_before(moved._root, node._root);
  5166. return moved;
  5167. }
  5168. PUGI_IMPL_FN bool xml_node::remove_attribute(const char_t* name_)
  5169. {
  5170. return remove_attribute(attribute(name_));
  5171. }
  5172. #ifdef PUGIXML_HAS_STRING_VIEW
  5173. PUGI_IMPL_FN bool xml_node::remove_attribute(string_view_t name_)
  5174. {
  5175. return remove_attribute(attribute(name_));
  5176. }
  5177. #endif
  5178. PUGI_IMPL_FN bool xml_node::remove_attribute(const xml_attribute& a)
  5179. {
  5180. if (!_root || !a._attr) return false;
  5181. if (!impl::is_attribute_of(a._attr, _root)) return false;
  5182. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5183. if (!alloc.reserve()) return false;
  5184. impl::remove_attribute(a._attr, _root);
  5185. impl::destroy_attribute(a._attr, alloc);
  5186. return true;
  5187. }
  5188. PUGI_IMPL_FN bool xml_node::remove_attributes()
  5189. {
  5190. if (!_root) return false;
  5191. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5192. if (!alloc.reserve()) return false;
  5193. for (xml_attribute_struct* attr = _root->first_attribute; attr; )
  5194. {
  5195. xml_attribute_struct* next = attr->next_attribute;
  5196. impl::destroy_attribute(attr, alloc);
  5197. attr = next;
  5198. }
  5199. _root->first_attribute = NULL;
  5200. return true;
  5201. }
  5202. PUGI_IMPL_FN bool xml_node::remove_child(const char_t* name_)
  5203. {
  5204. return remove_child(child(name_));
  5205. }
  5206. #ifdef PUGIXML_HAS_STRING_VIEW
  5207. PUGI_IMPL_FN bool xml_node::remove_child(string_view_t name_)
  5208. {
  5209. return remove_child(child(name_));
  5210. }
  5211. #endif
  5212. PUGI_IMPL_FN bool xml_node::remove_child(const xml_node& n)
  5213. {
  5214. if (!_root || !n._root || n._root->parent != _root) return false;
  5215. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5216. if (!alloc.reserve()) return false;
  5217. impl::remove_node(n._root);
  5218. impl::destroy_node(n._root, alloc);
  5219. return true;
  5220. }
  5221. PUGI_IMPL_FN bool xml_node::remove_children()
  5222. {
  5223. if (!_root) return false;
  5224. impl::xml_allocator& alloc = impl::get_allocator(_root);
  5225. if (!alloc.reserve()) return false;
  5226. for (xml_node_struct* cur = _root->first_child; cur; )
  5227. {
  5228. xml_node_struct* next = cur->next_sibling;
  5229. impl::destroy_node(cur, alloc);
  5230. cur = next;
  5231. }
  5232. _root->first_child = NULL;
  5233. return true;
  5234. }
  5235. PUGI_IMPL_FN xml_parse_result xml_node::append_buffer(const void* contents, size_t size, unsigned int options, xml_encoding encoding)
  5236. {
  5237. // append_buffer is only valid for elements/documents
  5238. if (!impl::allow_insert_child(type(), node_element)) return impl::make_parse_result(status_append_invalid_root);
  5239. // append buffer can not merge PCDATA into existing PCDATA nodes
  5240. if ((options & parse_merge_pcdata) != 0 && last_child().type() == node_pcdata) return impl::make_parse_result(status_append_invalid_root);
  5241. // get document node
  5242. impl::xml_document_struct* doc = &impl::get_document(_root);
  5243. // disable document_buffer_order optimization since in a document with multiple buffers comparing buffer pointers does not make sense
  5244. doc->header |= impl::xml_memory_page_contents_shared_mask;
  5245. // get extra buffer element (we'll store the document fragment buffer there so that we can deallocate it later)
  5246. impl::xml_memory_page* page = NULL;
  5247. impl::xml_extra_buffer* extra = static_cast<impl::xml_extra_buffer*>(doc->allocate_memory(sizeof(impl::xml_extra_buffer) + sizeof(void*), page));
  5248. (void)page;
  5249. if (!extra) return impl::make_parse_result(status_out_of_memory);
  5250. #ifdef PUGIXML_COMPACT
  5251. // align the memory block to a pointer boundary; this is required for compact mode where memory allocations are only 4b aligned
  5252. // note that this requires up to sizeof(void*)-1 additional memory, which the allocation above takes into account
  5253. extra = reinterpret_cast<impl::xml_extra_buffer*>((reinterpret_cast<uintptr_t>(extra) + (sizeof(void*) - 1)) & ~(sizeof(void*) - 1));
  5254. #endif
  5255. // add extra buffer to the list
  5256. extra->buffer = NULL;
  5257. extra->next = doc->extra_buffers;
  5258. doc->extra_buffers = extra;
  5259. // name of the root has to be NULL before parsing - otherwise closing node mismatches will not be detected at the top level
  5260. impl::name_null_sentry sentry(_root);
  5261. return impl::load_buffer_impl(doc, _root, const_cast<void*>(contents), size, options, encoding, false, false, &extra->buffer);
  5262. }
  5263. PUGI_IMPL_FN xml_node xml_node::find_child_by_attribute(const char_t* name_, const char_t* attr_name, const char_t* attr_value) const
  5264. {
  5265. if (!_root) return xml_node();
  5266. for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
  5267. {
  5268. const char_t* iname = i->name;
  5269. if (iname && impl::strequal(name_, iname))
  5270. {
  5271. for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute)
  5272. {
  5273. const char_t* aname = a->name;
  5274. if (aname && impl::strequal(attr_name, aname))
  5275. {
  5276. const char_t* avalue = a->value;
  5277. if (impl::strequal(attr_value, avalue ? avalue : PUGIXML_TEXT("")))
  5278. return xml_node(i);
  5279. }
  5280. }
  5281. }
  5282. }
  5283. return xml_node();
  5284. }
  5285. PUGI_IMPL_FN xml_node xml_node::find_child_by_attribute(const char_t* attr_name, const char_t* attr_value) const
  5286. {
  5287. if (!_root) return xml_node();
  5288. for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
  5289. for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute)
  5290. {
  5291. const char_t* aname = a->name;
  5292. if (aname && impl::strequal(attr_name, aname))
  5293. {
  5294. const char_t* avalue = a->value;
  5295. if (impl::strequal(attr_value, avalue ? avalue : PUGIXML_TEXT("")))
  5296. return xml_node(i);
  5297. }
  5298. }
  5299. return xml_node();
  5300. }
  5301. #ifndef PUGIXML_NO_STL
  5302. PUGI_IMPL_FN string_t xml_node::path(char_t delimiter) const
  5303. {
  5304. if (!_root) return string_t();
  5305. size_t offset = 0;
  5306. for (xml_node_struct* i = _root; i; i = i->parent)
  5307. {
  5308. const char_t* iname = i->name;
  5309. offset += (i != _root);
  5310. offset += iname ? impl::strlength(iname) : 0;
  5311. }
  5312. string_t result;
  5313. result.resize(offset);
  5314. for (xml_node_struct* j = _root; j; j = j->parent)
  5315. {
  5316. if (j != _root)
  5317. result[--offset] = delimiter;
  5318. const char_t* jname = j->name;
  5319. if (jname)
  5320. {
  5321. size_t length = impl::strlength(jname);
  5322. offset -= length;
  5323. memcpy(&result[offset], jname, length * sizeof(char_t));
  5324. }
  5325. }
  5326. assert(offset == 0);
  5327. return result;
  5328. }
  5329. #endif
  5330. PUGI_IMPL_FN xml_node xml_node::first_element_by_path(const char_t* path_, char_t delimiter) const
  5331. {
  5332. xml_node context = path_[0] == delimiter ? root() : *this;
  5333. if (!context._root) return xml_node();
  5334. const char_t* path_segment = path_;
  5335. while (*path_segment == delimiter) ++path_segment;
  5336. const char_t* path_segment_end = path_segment;
  5337. while (*path_segment_end && *path_segment_end != delimiter) ++path_segment_end;
  5338. if (path_segment == path_segment_end) return context;
  5339. const char_t* next_segment = path_segment_end;
  5340. while (*next_segment == delimiter) ++next_segment;
  5341. if (*path_segment == '.' && path_segment + 1 == path_segment_end)
  5342. return context.first_element_by_path(next_segment, delimiter);
  5343. else if (*path_segment == '.' && *(path_segment+1) == '.' && path_segment + 2 == path_segment_end)
  5344. return context.parent().first_element_by_path(next_segment, delimiter);
  5345. else
  5346. {
  5347. for (xml_node_struct* j = context._root->first_child; j; j = j->next_sibling)
  5348. {
  5349. const char_t* jname = j->name;
  5350. if (jname && impl::strequalrange(jname, path_segment, static_cast<size_t>(path_segment_end - path_segment)))
  5351. {
  5352. xml_node subsearch = xml_node(j).first_element_by_path(next_segment, delimiter);
  5353. if (subsearch) return subsearch;
  5354. }
  5355. }
  5356. return xml_node();
  5357. }
  5358. }
  5359. PUGI_IMPL_FN bool xml_node::traverse(xml_tree_walker& walker)
  5360. {
  5361. walker._depth = -1;
  5362. xml_node arg_begin(_root);
  5363. if (!walker.begin(arg_begin)) return false;
  5364. xml_node_struct* cur = _root ? _root->first_child + 0 : NULL;
  5365. if (cur)
  5366. {
  5367. ++walker._depth;
  5368. do
  5369. {
  5370. xml_node arg_for_each(cur);
  5371. if (!walker.for_each(arg_for_each))
  5372. return false;
  5373. if (cur->first_child)
  5374. {
  5375. ++walker._depth;
  5376. cur = cur->first_child;
  5377. }
  5378. else if (cur->next_sibling)
  5379. cur = cur->next_sibling;
  5380. else
  5381. {
  5382. while (!cur->next_sibling && cur != _root && cur->parent)
  5383. {
  5384. --walker._depth;
  5385. cur = cur->parent;
  5386. }
  5387. if (cur != _root)
  5388. cur = cur->next_sibling;
  5389. }
  5390. }
  5391. while (cur && cur != _root);
  5392. }
  5393. assert(walker._depth == -1);
  5394. xml_node arg_end(_root);
  5395. return walker.end(arg_end);
  5396. }
  5397. PUGI_IMPL_FN size_t xml_node::hash_value() const
  5398. {
  5399. return reinterpret_cast<uintptr_t>(_root) / sizeof(xml_node_struct);
  5400. }
  5401. PUGI_IMPL_FN xml_node_struct* xml_node::internal_object() const
  5402. {
  5403. return _root;
  5404. }
  5405. PUGI_IMPL_FN void xml_node::print(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const
  5406. {
  5407. if (!_root) return;
  5408. impl::xml_buffered_writer buffered_writer(writer, encoding);
  5409. impl::node_output(buffered_writer, _root, indent, flags, depth);
  5410. buffered_writer.flush();
  5411. }
  5412. #ifndef PUGIXML_NO_STL
  5413. PUGI_IMPL_FN void xml_node::print(std::basic_ostream<char>& stream, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const
  5414. {
  5415. xml_writer_stream writer(stream);
  5416. print(writer, indent, flags, encoding, depth);
  5417. }
  5418. PUGI_IMPL_FN void xml_node::print(std::basic_ostream<wchar_t>& stream, const char_t* indent, unsigned int flags, unsigned int depth) const
  5419. {
  5420. xml_writer_stream writer(stream);
  5421. print(writer, indent, flags, encoding_wchar, depth);
  5422. }
  5423. #endif
  5424. PUGI_IMPL_FN ptrdiff_t xml_node::offset_debug() const
  5425. {
  5426. if (!_root) return -1;
  5427. impl::xml_document_struct& doc = impl::get_document(_root);
  5428. // we can determine the offset reliably only if there is exactly once parse buffer
  5429. if (!doc.buffer || doc.extra_buffers) return -1;
  5430. switch (type())
  5431. {
  5432. case node_document:
  5433. return 0;
  5434. case node_element:
  5435. case node_declaration:
  5436. case node_pi:
  5437. return _root->name && (_root->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0 ? _root->name - doc.buffer : -1;
  5438. case node_pcdata:
  5439. case node_cdata:
  5440. case node_comment:
  5441. case node_doctype:
  5442. return _root->value && (_root->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0 ? _root->value - doc.buffer : -1;
  5443. default:
  5444. assert(false && "Invalid node type"); // unreachable
  5445. return -1;
  5446. }
  5447. }
  5448. #ifdef __BORLANDC__
  5449. PUGI_IMPL_FN bool operator&&(const xml_node& lhs, bool rhs)
  5450. {
  5451. return (bool)lhs && rhs;
  5452. }
  5453. PUGI_IMPL_FN bool operator||(const xml_node& lhs, bool rhs)
  5454. {
  5455. return (bool)lhs || rhs;
  5456. }
  5457. #endif
  5458. PUGI_IMPL_FN xml_text::xml_text(xml_node_struct* root): _root(root)
  5459. {
  5460. }
  5461. PUGI_IMPL_FN xml_node_struct* xml_text::_data() const
  5462. {
  5463. if (!_root || impl::is_text_node(_root)) return _root;
  5464. // element nodes can have value if parse_embed_pcdata was used
  5465. if (PUGI_IMPL_NODETYPE(_root) == node_element && _root->value)
  5466. return _root;
  5467. for (xml_node_struct* node = _root->first_child; node; node = node->next_sibling)
  5468. if (impl::is_text_node(node))
  5469. return node;
  5470. return NULL;
  5471. }
  5472. PUGI_IMPL_FN xml_node_struct* xml_text::_data_new()
  5473. {
  5474. xml_node_struct* d = _data();
  5475. if (d) return d;
  5476. return xml_node(_root).append_child(node_pcdata).internal_object();
  5477. }
  5478. PUGI_IMPL_FN xml_text::xml_text(): _root(NULL)
  5479. {
  5480. }
  5481. PUGI_IMPL_FN static void unspecified_bool_xml_text(xml_text***)
  5482. {
  5483. }
  5484. PUGI_IMPL_FN xml_text::operator xml_text::unspecified_bool_type() const
  5485. {
  5486. return _data() ? unspecified_bool_xml_text : NULL;
  5487. }
  5488. PUGI_IMPL_FN bool xml_text::operator!() const
  5489. {
  5490. return !_data();
  5491. }
  5492. PUGI_IMPL_FN bool xml_text::empty() const
  5493. {
  5494. return _data() == NULL;
  5495. }
  5496. PUGI_IMPL_FN const char_t* xml_text::get() const
  5497. {
  5498. xml_node_struct* d = _data();
  5499. if (!d) return PUGIXML_TEXT("");
  5500. const char_t* value = d->value;
  5501. return value ? value : PUGIXML_TEXT("");
  5502. }
  5503. PUGI_IMPL_FN const char_t* xml_text::as_string(const char_t* def) const
  5504. {
  5505. xml_node_struct* d = _data();
  5506. if (!d) return def;
  5507. const char_t* value = d->value;
  5508. return value ? value : def;
  5509. }
  5510. PUGI_IMPL_FN int xml_text::as_int(int def) const
  5511. {
  5512. xml_node_struct* d = _data();
  5513. if (!d) return def;
  5514. const char_t* value = d->value;
  5515. return value ? impl::get_value_int(value) : def;
  5516. }
  5517. PUGI_IMPL_FN unsigned int xml_text::as_uint(unsigned int def) const
  5518. {
  5519. xml_node_struct* d = _data();
  5520. if (!d) return def;
  5521. const char_t* value = d->value;
  5522. return value ? impl::get_value_uint(value) : def;
  5523. }
  5524. PUGI_IMPL_FN double xml_text::as_double(double def) const
  5525. {
  5526. xml_node_struct* d = _data();
  5527. if (!d) return def;
  5528. const char_t* value = d->value;
  5529. return value ? impl::get_value_double(value) : def;
  5530. }
  5531. PUGI_IMPL_FN float xml_text::as_float(float def) const
  5532. {
  5533. xml_node_struct* d = _data();
  5534. if (!d) return def;
  5535. const char_t* value = d->value;
  5536. return value ? impl::get_value_float(value) : def;
  5537. }
  5538. PUGI_IMPL_FN bool xml_text::as_bool(bool def) const
  5539. {
  5540. xml_node_struct* d = _data();
  5541. if (!d) return def;
  5542. const char_t* value = d->value;
  5543. return value ? impl::get_value_bool(value) : def;
  5544. }
  5545. #ifdef PUGIXML_HAS_LONG_LONG
  5546. PUGI_IMPL_FN long long xml_text::as_llong(long long def) const
  5547. {
  5548. xml_node_struct* d = _data();
  5549. if (!d) return def;
  5550. const char_t* value = d->value;
  5551. return value ? impl::get_value_llong(value) : def;
  5552. }
  5553. PUGI_IMPL_FN unsigned long long xml_text::as_ullong(unsigned long long def) const
  5554. {
  5555. xml_node_struct* d = _data();
  5556. if (!d) return def;
  5557. const char_t* value = d->value;
  5558. return value ? impl::get_value_ullong(value) : def;
  5559. }
  5560. #endif
  5561. PUGI_IMPL_FN bool xml_text::set(const char_t* rhs)
  5562. {
  5563. xml_node_struct* dn = _data_new();
  5564. return dn ? impl::strcpy_insitu(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, impl::strlength(rhs)) : false;
  5565. }
  5566. PUGI_IMPL_FN bool xml_text::set(const char_t* rhs, size_t size)
  5567. {
  5568. xml_node_struct* dn = _data_new();
  5569. return dn ? impl::strcpy_insitu(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, size) : false;
  5570. }
  5571. #ifdef PUGIXML_HAS_STRING_VIEW
  5572. PUGI_IMPL_FN bool xml_text::set(string_view_t rhs)
  5573. {
  5574. xml_node_struct* dn = _data_new();
  5575. return dn ? impl::strcpy_insitu(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs.data(), rhs.size()) : false;
  5576. }
  5577. #endif
  5578. PUGI_IMPL_FN bool xml_text::set(int rhs)
  5579. {
  5580. xml_node_struct* dn = _data_new();
  5581. return dn ? impl::set_value_integer<unsigned int>(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0) : false;
  5582. }
  5583. PUGI_IMPL_FN bool xml_text::set(unsigned int rhs)
  5584. {
  5585. xml_node_struct* dn = _data_new();
  5586. return dn ? impl::set_value_integer<unsigned int>(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, false) : false;
  5587. }
  5588. PUGI_IMPL_FN bool xml_text::set(long rhs)
  5589. {
  5590. xml_node_struct* dn = _data_new();
  5591. return dn ? impl::set_value_integer<unsigned long>(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0) : false;
  5592. }
  5593. PUGI_IMPL_FN bool xml_text::set(unsigned long rhs)
  5594. {
  5595. xml_node_struct* dn = _data_new();
  5596. return dn ? impl::set_value_integer<unsigned long>(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, false) : false;
  5597. }
  5598. PUGI_IMPL_FN bool xml_text::set(float rhs)
  5599. {
  5600. xml_node_struct* dn = _data_new();
  5601. return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, default_float_precision) : false;
  5602. }
  5603. PUGI_IMPL_FN bool xml_text::set(float rhs, int precision)
  5604. {
  5605. xml_node_struct* dn = _data_new();
  5606. return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, precision) : false;
  5607. }
  5608. PUGI_IMPL_FN bool xml_text::set(double rhs)
  5609. {
  5610. xml_node_struct* dn = _data_new();
  5611. return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, default_double_precision) : false;
  5612. }
  5613. PUGI_IMPL_FN bool xml_text::set(double rhs, int precision)
  5614. {
  5615. xml_node_struct* dn = _data_new();
  5616. return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, precision) : false;
  5617. }
  5618. PUGI_IMPL_FN bool xml_text::set(bool rhs)
  5619. {
  5620. xml_node_struct* dn = _data_new();
  5621. return dn ? impl::set_value_bool(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
  5622. }
  5623. #ifdef PUGIXML_HAS_LONG_LONG
  5624. PUGI_IMPL_FN bool xml_text::set(long long rhs)
  5625. {
  5626. xml_node_struct* dn = _data_new();
  5627. return dn ? impl::set_value_integer<unsigned long long>(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0) : false;
  5628. }
  5629. PUGI_IMPL_FN bool xml_text::set(unsigned long long rhs)
  5630. {
  5631. xml_node_struct* dn = _data_new();
  5632. return dn ? impl::set_value_integer<unsigned long long>(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, false) : false;
  5633. }
  5634. #endif
  5635. PUGI_IMPL_FN xml_text& xml_text::operator=(const char_t* rhs)
  5636. {
  5637. set(rhs);
  5638. return *this;
  5639. }
  5640. PUGI_IMPL_FN xml_text& xml_text::operator=(int rhs)
  5641. {
  5642. set(rhs);
  5643. return *this;
  5644. }
  5645. PUGI_IMPL_FN xml_text& xml_text::operator=(unsigned int rhs)
  5646. {
  5647. set(rhs);
  5648. return *this;
  5649. }
  5650. PUGI_IMPL_FN xml_text& xml_text::operator=(long rhs)
  5651. {
  5652. set(rhs);
  5653. return *this;
  5654. }
  5655. PUGI_IMPL_FN xml_text& xml_text::operator=(unsigned long rhs)
  5656. {
  5657. set(rhs);
  5658. return *this;
  5659. }
  5660. PUGI_IMPL_FN xml_text& xml_text::operator=(double rhs)
  5661. {
  5662. set(rhs);
  5663. return *this;
  5664. }
  5665. PUGI_IMPL_FN xml_text& xml_text::operator=(float rhs)
  5666. {
  5667. set(rhs);
  5668. return *this;
  5669. }
  5670. PUGI_IMPL_FN xml_text& xml_text::operator=(bool rhs)
  5671. {
  5672. set(rhs);
  5673. return *this;
  5674. }
  5675. #ifdef PUGIXML_HAS_STRING_VIEW
  5676. PUGI_IMPL_FN xml_text& xml_text::operator=(string_view_t rhs)
  5677. {
  5678. set(rhs);
  5679. return *this;
  5680. }
  5681. #endif
  5682. #ifdef PUGIXML_HAS_LONG_LONG
  5683. PUGI_IMPL_FN xml_text& xml_text::operator=(long long rhs)
  5684. {
  5685. set(rhs);
  5686. return *this;
  5687. }
  5688. PUGI_IMPL_FN xml_text& xml_text::operator=(unsigned long long rhs)
  5689. {
  5690. set(rhs);
  5691. return *this;
  5692. }
  5693. #endif
  5694. PUGI_IMPL_FN xml_node xml_text::data() const
  5695. {
  5696. return xml_node(_data());
  5697. }
  5698. #ifdef __BORLANDC__
  5699. PUGI_IMPL_FN bool operator&&(const xml_text& lhs, bool rhs)
  5700. {
  5701. return (bool)lhs && rhs;
  5702. }
  5703. PUGI_IMPL_FN bool operator||(const xml_text& lhs, bool rhs)
  5704. {
  5705. return (bool)lhs || rhs;
  5706. }
  5707. #endif
  5708. PUGI_IMPL_FN xml_node_iterator::xml_node_iterator()
  5709. {
  5710. }
  5711. PUGI_IMPL_FN xml_node_iterator::xml_node_iterator(const xml_node& node): _wrap(node), _parent(node.parent())
  5712. {
  5713. }
  5714. PUGI_IMPL_FN xml_node_iterator::xml_node_iterator(xml_node_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent)
  5715. {
  5716. }
  5717. PUGI_IMPL_FN bool xml_node_iterator::operator==(const xml_node_iterator& rhs) const
  5718. {
  5719. return _wrap._root == rhs._wrap._root && _parent._root == rhs._parent._root;
  5720. }
  5721. PUGI_IMPL_FN bool xml_node_iterator::operator!=(const xml_node_iterator& rhs) const
  5722. {
  5723. return _wrap._root != rhs._wrap._root || _parent._root != rhs._parent._root;
  5724. }
  5725. PUGI_IMPL_FN xml_node& xml_node_iterator::operator*() const
  5726. {
  5727. assert(_wrap._root);
  5728. return _wrap;
  5729. }
  5730. PUGI_IMPL_FN xml_node* xml_node_iterator::operator->() const
  5731. {
  5732. assert(_wrap._root);
  5733. return &_wrap;
  5734. }
  5735. PUGI_IMPL_FN xml_node_iterator& xml_node_iterator::operator++()
  5736. {
  5737. assert(_wrap._root);
  5738. _wrap._root = _wrap._root->next_sibling;
  5739. return *this;
  5740. }
  5741. PUGI_IMPL_FN xml_node_iterator xml_node_iterator::operator++(int)
  5742. {
  5743. xml_node_iterator temp = *this;
  5744. ++*this;
  5745. return temp;
  5746. }
  5747. PUGI_IMPL_FN xml_node_iterator& xml_node_iterator::operator--()
  5748. {
  5749. _wrap = _wrap._root ? _wrap.previous_sibling() : _parent.last_child();
  5750. return *this;
  5751. }
  5752. PUGI_IMPL_FN xml_node_iterator xml_node_iterator::operator--(int)
  5753. {
  5754. xml_node_iterator temp = *this;
  5755. --*this;
  5756. return temp;
  5757. }
  5758. PUGI_IMPL_FN xml_attribute_iterator::xml_attribute_iterator()
  5759. {
  5760. }
  5761. PUGI_IMPL_FN xml_attribute_iterator::xml_attribute_iterator(const xml_attribute& attr, const xml_node& parent): _wrap(attr), _parent(parent)
  5762. {
  5763. }
  5764. PUGI_IMPL_FN xml_attribute_iterator::xml_attribute_iterator(xml_attribute_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent)
  5765. {
  5766. }
  5767. PUGI_IMPL_FN bool xml_attribute_iterator::operator==(const xml_attribute_iterator& rhs) const
  5768. {
  5769. return _wrap._attr == rhs._wrap._attr && _parent._root == rhs._parent._root;
  5770. }
  5771. PUGI_IMPL_FN bool xml_attribute_iterator::operator!=(const xml_attribute_iterator& rhs) const
  5772. {
  5773. return _wrap._attr != rhs._wrap._attr || _parent._root != rhs._parent._root;
  5774. }
  5775. PUGI_IMPL_FN xml_attribute& xml_attribute_iterator::operator*() const
  5776. {
  5777. assert(_wrap._attr);
  5778. return _wrap;
  5779. }
  5780. PUGI_IMPL_FN xml_attribute* xml_attribute_iterator::operator->() const
  5781. {
  5782. assert(_wrap._attr);
  5783. return &_wrap;
  5784. }
  5785. PUGI_IMPL_FN xml_attribute_iterator& xml_attribute_iterator::operator++()
  5786. {
  5787. assert(_wrap._attr);
  5788. _wrap._attr = _wrap._attr->next_attribute;
  5789. return *this;
  5790. }
  5791. PUGI_IMPL_FN xml_attribute_iterator xml_attribute_iterator::operator++(int)
  5792. {
  5793. xml_attribute_iterator temp = *this;
  5794. ++*this;
  5795. return temp;
  5796. }
  5797. PUGI_IMPL_FN xml_attribute_iterator& xml_attribute_iterator::operator--()
  5798. {
  5799. _wrap = _wrap._attr ? _wrap.previous_attribute() : _parent.last_attribute();
  5800. return *this;
  5801. }
  5802. PUGI_IMPL_FN xml_attribute_iterator xml_attribute_iterator::operator--(int)
  5803. {
  5804. xml_attribute_iterator temp = *this;
  5805. --*this;
  5806. return temp;
  5807. }
  5808. PUGI_IMPL_FN xml_named_node_iterator::xml_named_node_iterator(): _name(NULL)
  5809. {
  5810. }
  5811. PUGI_IMPL_FN xml_named_node_iterator::xml_named_node_iterator(const xml_node& node, const char_t* name): _wrap(node), _parent(node.parent()), _name(name)
  5812. {
  5813. }
  5814. PUGI_IMPL_FN xml_named_node_iterator::xml_named_node_iterator(xml_node_struct* ref, xml_node_struct* parent, const char_t* name): _wrap(ref), _parent(parent), _name(name)
  5815. {
  5816. }
  5817. PUGI_IMPL_FN bool xml_named_node_iterator::operator==(const xml_named_node_iterator& rhs) const
  5818. {
  5819. return _wrap._root == rhs._wrap._root && _parent._root == rhs._parent._root;
  5820. }
  5821. PUGI_IMPL_FN bool xml_named_node_iterator::operator!=(const xml_named_node_iterator& rhs) const
  5822. {
  5823. return _wrap._root != rhs._wrap._root || _parent._root != rhs._parent._root;
  5824. }
  5825. PUGI_IMPL_FN xml_node& xml_named_node_iterator::operator*() const
  5826. {
  5827. assert(_wrap._root);
  5828. return _wrap;
  5829. }
  5830. PUGI_IMPL_FN xml_node* xml_named_node_iterator::operator->() const
  5831. {
  5832. assert(_wrap._root);
  5833. return &_wrap;
  5834. }
  5835. PUGI_IMPL_FN xml_named_node_iterator& xml_named_node_iterator::operator++()
  5836. {
  5837. assert(_wrap._root);
  5838. _wrap = _wrap.next_sibling(_name);
  5839. return *this;
  5840. }
  5841. PUGI_IMPL_FN xml_named_node_iterator xml_named_node_iterator::operator++(int)
  5842. {
  5843. xml_named_node_iterator temp = *this;
  5844. ++*this;
  5845. return temp;
  5846. }
  5847. PUGI_IMPL_FN xml_named_node_iterator& xml_named_node_iterator::operator--()
  5848. {
  5849. if (_wrap._root)
  5850. _wrap = _wrap.previous_sibling(_name);
  5851. else
  5852. {
  5853. _wrap = _parent.last_child();
  5854. if (!impl::strequal(_wrap.name(), _name))
  5855. _wrap = _wrap.previous_sibling(_name);
  5856. }
  5857. return *this;
  5858. }
  5859. PUGI_IMPL_FN xml_named_node_iterator xml_named_node_iterator::operator--(int)
  5860. {
  5861. xml_named_node_iterator temp = *this;
  5862. --*this;
  5863. return temp;
  5864. }
  5865. PUGI_IMPL_FN xml_parse_result::xml_parse_result(): status(status_internal_error), offset(0), encoding(encoding_auto)
  5866. {
  5867. }
  5868. PUGI_IMPL_FN xml_parse_result::operator bool() const
  5869. {
  5870. return status == status_ok;
  5871. }
  5872. PUGI_IMPL_FN const char* xml_parse_result::description() const
  5873. {
  5874. switch (status)
  5875. {
  5876. case status_ok: return "No error";
  5877. case status_file_not_found: return "File was not found";
  5878. case status_io_error: return "Error reading from file/stream";
  5879. case status_out_of_memory: return "Could not allocate memory";
  5880. case status_internal_error: return "Internal error occurred";
  5881. case status_unrecognized_tag: return "Could not determine tag type";
  5882. case status_bad_pi: return "Error parsing document declaration/processing instruction";
  5883. case status_bad_comment: return "Error parsing comment";
  5884. case status_bad_cdata: return "Error parsing CDATA section";
  5885. case status_bad_doctype: return "Error parsing document type declaration";
  5886. case status_bad_pcdata: return "Error parsing PCDATA section";
  5887. case status_bad_start_element: return "Error parsing start element tag";
  5888. case status_bad_attribute: return "Error parsing element attribute";
  5889. case status_bad_end_element: return "Error parsing end element tag";
  5890. case status_end_element_mismatch: return "Start-end tags mismatch";
  5891. case status_append_invalid_root: return "Unable to append nodes: root is not an element or document";
  5892. case status_no_document_element: return "No document element found";
  5893. default: return "Unknown error";
  5894. }
  5895. }
  5896. PUGI_IMPL_FN xml_document::xml_document(): _buffer(NULL)
  5897. {
  5898. _create();
  5899. }
  5900. PUGI_IMPL_FN xml_document::~xml_document()
  5901. {
  5902. _destroy();
  5903. }
  5904. #ifdef PUGIXML_HAS_MOVE
  5905. PUGI_IMPL_FN xml_document::xml_document(xml_document&& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT: _buffer(NULL)
  5906. {
  5907. _create();
  5908. _move(rhs);
  5909. }
  5910. PUGI_IMPL_FN xml_document& xml_document::operator=(xml_document&& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT
  5911. {
  5912. if (this == &rhs) return *this;
  5913. _destroy();
  5914. _create();
  5915. _move(rhs);
  5916. return *this;
  5917. }
  5918. #endif
  5919. PUGI_IMPL_FN void xml_document::reset()
  5920. {
  5921. _destroy();
  5922. _create();
  5923. }
  5924. PUGI_IMPL_FN void xml_document::reset(const xml_document& proto)
  5925. {
  5926. reset();
  5927. impl::node_copy_tree(_root, proto._root);
  5928. }
  5929. PUGI_IMPL_FN void xml_document::_create()
  5930. {
  5931. assert(!_root);
  5932. #ifdef PUGIXML_COMPACT
  5933. // space for page marker for the first page (uint32_t), rounded up to pointer size; assumes pointers are at least 32-bit
  5934. const size_t page_offset = sizeof(void*);
  5935. #else
  5936. const size_t page_offset = 0;
  5937. #endif
  5938. // initialize sentinel page
  5939. PUGI_IMPL_STATIC_ASSERT(sizeof(impl::xml_memory_page) + sizeof(impl::xml_document_struct) + page_offset <= sizeof(_memory));
  5940. // prepare page structure
  5941. impl::xml_memory_page* page = impl::xml_memory_page::construct(_memory);
  5942. assert(page);
  5943. page->busy_size = impl::xml_memory_page_size;
  5944. // setup first page marker
  5945. #ifdef PUGIXML_COMPACT
  5946. // round-trip through void* to avoid 'cast increases required alignment of target type' warning
  5947. page->compact_page_marker = reinterpret_cast<uint32_t*>(static_cast<void*>(reinterpret_cast<char*>(page) + sizeof(impl::xml_memory_page)));
  5948. *page->compact_page_marker = sizeof(impl::xml_memory_page);
  5949. #endif
  5950. // allocate new root
  5951. _root = new (reinterpret_cast<char*>(page) + sizeof(impl::xml_memory_page) + page_offset) impl::xml_document_struct(page);
  5952. _root->prev_sibling_c = _root;
  5953. // setup sentinel page
  5954. page->allocator = static_cast<impl::xml_document_struct*>(_root);
  5955. // setup hash table pointer in allocator
  5956. #ifdef PUGIXML_COMPACT
  5957. page->allocator->_hash = &static_cast<impl::xml_document_struct*>(_root)->hash;
  5958. #endif
  5959. // verify the document allocation
  5960. assert(reinterpret_cast<char*>(_root) + sizeof(impl::xml_document_struct) <= _memory + sizeof(_memory));
  5961. }
  5962. PUGI_IMPL_FN void xml_document::_destroy()
  5963. {
  5964. assert(_root);
  5965. // destroy static storage
  5966. if (_buffer)
  5967. {
  5968. impl::xml_memory::deallocate(_buffer);
  5969. _buffer = NULL;
  5970. }
  5971. // destroy extra buffers (note: no need to destroy linked list nodes, they're allocated using document allocator)
  5972. for (impl::xml_extra_buffer* extra = static_cast<impl::xml_document_struct*>(_root)->extra_buffers; extra; extra = extra->next)
  5973. {
  5974. if (extra->buffer) impl::xml_memory::deallocate(extra->buffer);
  5975. }
  5976. // destroy dynamic storage, leave sentinel page (it's in static memory)
  5977. impl::xml_memory_page* root_page = PUGI_IMPL_GETPAGE(_root);
  5978. assert(root_page && !root_page->prev);
  5979. assert(reinterpret_cast<char*>(root_page) >= _memory && reinterpret_cast<char*>(root_page) < _memory + sizeof(_memory));
  5980. for (impl::xml_memory_page* page = root_page->next; page; )
  5981. {
  5982. impl::xml_memory_page* next = page->next;
  5983. impl::xml_allocator::deallocate_page(page);
  5984. page = next;
  5985. }
  5986. #ifdef PUGIXML_COMPACT
  5987. // destroy hash table
  5988. static_cast<impl::xml_document_struct*>(_root)->hash.clear();
  5989. #endif
  5990. _root = NULL;
  5991. }
  5992. #ifdef PUGIXML_HAS_MOVE
  5993. PUGI_IMPL_FN void xml_document::_move(xml_document& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT
  5994. {
  5995. impl::xml_document_struct* doc = static_cast<impl::xml_document_struct*>(_root);
  5996. impl::xml_document_struct* other = static_cast<impl::xml_document_struct*>(rhs._root);
  5997. // save first child pointer for later; this needs hash access
  5998. xml_node_struct* other_first_child = other->first_child;
  5999. #ifdef PUGIXML_COMPACT
  6000. // reserve space for the hash table up front; this is the only operation that can fail
  6001. // if it does, we have no choice but to throw (if we have exceptions)
  6002. if (other_first_child)
  6003. {
  6004. size_t other_children = 0;
  6005. for (xml_node_struct* node = other_first_child; node; node = node->next_sibling)
  6006. other_children++;
  6007. // in compact mode, each pointer assignment could result in a hash table request
  6008. // during move, we have to relocate document first_child and parents of all children
  6009. // normally there's just one child and its parent has a pointerless encoding but
  6010. // we assume the worst here
  6011. if (!other->_hash->reserve(other_children + 1))
  6012. {
  6013. #ifdef PUGIXML_NO_EXCEPTIONS
  6014. return;
  6015. #else
  6016. throw std::bad_alloc();
  6017. #endif
  6018. }
  6019. }
  6020. #endif
  6021. // move allocation state
  6022. // note that other->_root may point to the embedded document page, in which case we should keep original (empty) state
  6023. if (other->_root != PUGI_IMPL_GETPAGE(other))
  6024. {
  6025. doc->_root = other->_root;
  6026. doc->_busy_size = other->_busy_size;
  6027. }
  6028. // move buffer state
  6029. doc->buffer = other->buffer;
  6030. doc->extra_buffers = other->extra_buffers;
  6031. _buffer = rhs._buffer;
  6032. #ifdef PUGIXML_COMPACT
  6033. // move compact hash; note that the hash table can have pointers to other but they will be "inactive", similarly to nodes removed with remove_child
  6034. doc->hash = other->hash;
  6035. doc->_hash = &doc->hash;
  6036. // make sure we don't access other hash up until the end when we reinitialize other document
  6037. other->_hash = NULL;
  6038. #endif
  6039. // move page structure
  6040. impl::xml_memory_page* doc_page = PUGI_IMPL_GETPAGE(doc);
  6041. assert(doc_page && !doc_page->prev && !doc_page->next);
  6042. impl::xml_memory_page* other_page = PUGI_IMPL_GETPAGE(other);
  6043. assert(other_page && !other_page->prev);
  6044. // relink pages since root page is embedded into xml_document
  6045. if (impl::xml_memory_page* page = other_page->next)
  6046. {
  6047. assert(page->prev == other_page);
  6048. page->prev = doc_page;
  6049. doc_page->next = page;
  6050. other_page->next = NULL;
  6051. }
  6052. // make sure pages point to the correct document state
  6053. for (impl::xml_memory_page* page = doc_page->next; page; page = page->next)
  6054. {
  6055. assert(page->allocator == other);
  6056. page->allocator = doc;
  6057. #ifdef PUGIXML_COMPACT
  6058. // this automatically migrates most children between documents and prevents ->parent assignment from allocating
  6059. if (page->compact_shared_parent == other)
  6060. page->compact_shared_parent = doc;
  6061. #endif
  6062. }
  6063. // move tree structure
  6064. assert(!doc->first_child);
  6065. doc->first_child = other_first_child;
  6066. for (xml_node_struct* node = other_first_child; node; node = node->next_sibling)
  6067. {
  6068. #ifdef PUGIXML_COMPACT
  6069. // most children will have migrated when we reassigned compact_shared_parent
  6070. assert(node->parent == other || node->parent == doc);
  6071. node->parent = doc;
  6072. #else
  6073. assert(node->parent == other);
  6074. node->parent = doc;
  6075. #endif
  6076. }
  6077. // reset other document
  6078. new (other) impl::xml_document_struct(PUGI_IMPL_GETPAGE(other));
  6079. rhs._buffer = NULL;
  6080. }
  6081. #endif
  6082. #ifndef PUGIXML_NO_STL
  6083. PUGI_IMPL_FN xml_parse_result xml_document::load(std::basic_istream<char>& stream, unsigned int options, xml_encoding encoding)
  6084. {
  6085. reset();
  6086. return impl::load_stream_impl(static_cast<impl::xml_document_struct*>(_root), stream, options, encoding, &_buffer);
  6087. }
  6088. PUGI_IMPL_FN xml_parse_result xml_document::load(std::basic_istream<wchar_t>& stream, unsigned int options)
  6089. {
  6090. reset();
  6091. return impl::load_stream_impl(static_cast<impl::xml_document_struct*>(_root), stream, options, encoding_wchar, &_buffer);
  6092. }
  6093. #endif
  6094. PUGI_IMPL_FN xml_parse_result xml_document::load_string(const char_t* contents, unsigned int options)
  6095. {
  6096. // Force native encoding (skip autodetection)
  6097. #ifdef PUGIXML_WCHAR_MODE
  6098. xml_encoding encoding = encoding_wchar;
  6099. #else
  6100. xml_encoding encoding = encoding_utf8;
  6101. #endif
  6102. return load_buffer(contents, impl::strlength(contents) * sizeof(char_t), options, encoding);
  6103. }
  6104. PUGI_IMPL_FN xml_parse_result xml_document::load(const char_t* contents, unsigned int options)
  6105. {
  6106. return load_string(contents, options);
  6107. }
  6108. PUGI_IMPL_FN xml_parse_result xml_document::load_file(const char* path_, unsigned int options, xml_encoding encoding)
  6109. {
  6110. reset();
  6111. using impl::auto_deleter; // MSVC7 workaround
  6112. auto_deleter<FILE> file(impl::open_file(path_, "rb"), impl::close_file);
  6113. return impl::load_file_impl(static_cast<impl::xml_document_struct*>(_root), file.data, options, encoding, &_buffer);
  6114. }
  6115. PUGI_IMPL_FN xml_parse_result xml_document::load_file(const wchar_t* path_, unsigned int options, xml_encoding encoding)
  6116. {
  6117. reset();
  6118. using impl::auto_deleter; // MSVC7 workaround
  6119. auto_deleter<FILE> file(impl::open_file_wide(path_, L"rb"), impl::close_file);
  6120. return impl::load_file_impl(static_cast<impl::xml_document_struct*>(_root), file.data, options, encoding, &_buffer);
  6121. }
  6122. PUGI_IMPL_FN xml_parse_result xml_document::load_buffer(const void* contents, size_t size, unsigned int options, xml_encoding encoding)
  6123. {
  6124. reset();
  6125. return impl::load_buffer_impl(static_cast<impl::xml_document_struct*>(_root), _root, const_cast<void*>(contents), size, options, encoding, false, false, &_buffer);
  6126. }
  6127. PUGI_IMPL_FN xml_parse_result xml_document::load_buffer_inplace(void* contents, size_t size, unsigned int options, xml_encoding encoding)
  6128. {
  6129. reset();
  6130. return impl::load_buffer_impl(static_cast<impl::xml_document_struct*>(_root), _root, contents, size, options, encoding, true, false, &_buffer);
  6131. }
  6132. PUGI_IMPL_FN xml_parse_result xml_document::load_buffer_inplace_own(void* contents, size_t size, unsigned int options, xml_encoding encoding)
  6133. {
  6134. reset();
  6135. return impl::load_buffer_impl(static_cast<impl::xml_document_struct*>(_root), _root, contents, size, options, encoding, true, true, &_buffer);
  6136. }
  6137. PUGI_IMPL_FN void xml_document::save(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding) const
  6138. {
  6139. impl::xml_buffered_writer buffered_writer(writer, encoding);
  6140. if ((flags & format_write_bom) && buffered_writer.encoding != encoding_latin1)
  6141. {
  6142. // BOM always represents the codepoint U+FEFF, so just write it in native encoding
  6143. #ifdef PUGIXML_WCHAR_MODE
  6144. unsigned int bom = 0xfeff;
  6145. buffered_writer.write(static_cast<wchar_t>(bom));
  6146. #else
  6147. buffered_writer.write('\xef', '\xbb', '\xbf');
  6148. #endif
  6149. }
  6150. if (!(flags & format_no_declaration) && !impl::has_declaration(_root))
  6151. {
  6152. buffered_writer.write_string(PUGIXML_TEXT("<?xml version=\"1.0\""));
  6153. if (buffered_writer.encoding == encoding_latin1) buffered_writer.write_string(PUGIXML_TEXT(" encoding=\"ISO-8859-1\""));
  6154. buffered_writer.write('?', '>');
  6155. if (!(flags & format_raw)) buffered_writer.write('\n');
  6156. }
  6157. impl::node_output(buffered_writer, _root, indent, flags, 0);
  6158. buffered_writer.flush();
  6159. }
  6160. #ifndef PUGIXML_NO_STL
  6161. PUGI_IMPL_FN void xml_document::save(std::basic_ostream<char>& stream, const char_t* indent, unsigned int flags, xml_encoding encoding) const
  6162. {
  6163. xml_writer_stream writer(stream);
  6164. save(writer, indent, flags, encoding);
  6165. }
  6166. PUGI_IMPL_FN void xml_document::save(std::basic_ostream<wchar_t>& stream, const char_t* indent, unsigned int flags) const
  6167. {
  6168. xml_writer_stream writer(stream);
  6169. save(writer, indent, flags, encoding_wchar);
  6170. }
  6171. #endif
  6172. PUGI_IMPL_FN bool xml_document::save_file(const char* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const
  6173. {
  6174. using impl::auto_deleter; // MSVC7 workaround
  6175. auto_deleter<FILE> file(impl::open_file(path_, (flags & format_save_file_text) ? "w" : "wb"), impl::close_file);
  6176. return impl::save_file_impl(*this, file.data, indent, flags, encoding) && fclose(file.release()) == 0;
  6177. }
  6178. PUGI_IMPL_FN bool xml_document::save_file(const wchar_t* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const
  6179. {
  6180. using impl::auto_deleter; // MSVC7 workaround
  6181. auto_deleter<FILE> file(impl::open_file_wide(path_, (flags & format_save_file_text) ? L"w" : L"wb"), impl::close_file);
  6182. return impl::save_file_impl(*this, file.data, indent, flags, encoding) && fclose(file.release()) == 0;
  6183. }
  6184. PUGI_IMPL_FN xml_node xml_document::document_element() const
  6185. {
  6186. assert(_root);
  6187. for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
  6188. if (PUGI_IMPL_NODETYPE(i) == node_element)
  6189. return xml_node(i);
  6190. return xml_node();
  6191. }
  6192. #ifndef PUGIXML_NO_STL
  6193. PUGI_IMPL_FN std::string PUGIXML_FUNCTION as_utf8(const wchar_t* str)
  6194. {
  6195. assert(str);
  6196. return impl::as_utf8_impl(str, impl::strlength_wide(str));
  6197. }
  6198. PUGI_IMPL_FN std::string PUGIXML_FUNCTION as_utf8(const std::basic_string<wchar_t>& str)
  6199. {
  6200. return impl::as_utf8_impl(str.c_str(), str.size());
  6201. }
  6202. PUGI_IMPL_FN std::basic_string<wchar_t> PUGIXML_FUNCTION as_wide(const char* str)
  6203. {
  6204. assert(str);
  6205. return impl::as_wide_impl(str, strlen(str));
  6206. }
  6207. PUGI_IMPL_FN std::basic_string<wchar_t> PUGIXML_FUNCTION as_wide(const std::string& str)
  6208. {
  6209. return impl::as_wide_impl(str.c_str(), str.size());
  6210. }
  6211. #endif
  6212. PUGI_IMPL_FN void PUGIXML_FUNCTION set_memory_management_functions(allocation_function allocate, deallocation_function deallocate)
  6213. {
  6214. impl::xml_memory::allocate = allocate;
  6215. impl::xml_memory::deallocate = deallocate;
  6216. }
  6217. PUGI_IMPL_FN allocation_function PUGIXML_FUNCTION get_memory_allocation_function()
  6218. {
  6219. return impl::xml_memory::allocate;
  6220. }
  6221. PUGI_IMPL_FN deallocation_function PUGIXML_FUNCTION get_memory_deallocation_function()
  6222. {
  6223. return impl::xml_memory::deallocate;
  6224. }
  6225. }
  6226. #if !defined(PUGIXML_NO_STL) && (defined(_MSC_VER) || defined(__ICC))
  6227. namespace std
  6228. {
  6229. // Workarounds for (non-standard) iterator category detection for older versions (MSVC7/IC8 and earlier)
  6230. PUGI_IMPL_FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_node_iterator&)
  6231. {
  6232. return std::bidirectional_iterator_tag();
  6233. }
  6234. PUGI_IMPL_FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_attribute_iterator&)
  6235. {
  6236. return std::bidirectional_iterator_tag();
  6237. }
  6238. PUGI_IMPL_FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_named_node_iterator&)
  6239. {
  6240. return std::bidirectional_iterator_tag();
  6241. }
  6242. }
  6243. #endif
  6244. #if !defined(PUGIXML_NO_STL) && defined(__SUNPRO_CC)
  6245. namespace std
  6246. {
  6247. // Workarounds for (non-standard) iterator category detection
  6248. PUGI_IMPL_FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_node_iterator&)
  6249. {
  6250. return std::bidirectional_iterator_tag();
  6251. }
  6252. PUGI_IMPL_FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_attribute_iterator&)
  6253. {
  6254. return std::bidirectional_iterator_tag();
  6255. }
  6256. PUGI_IMPL_FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_named_node_iterator&)
  6257. {
  6258. return std::bidirectional_iterator_tag();
  6259. }
  6260. }
  6261. #endif
  6262. #ifndef PUGIXML_NO_XPATH
  6263. // STL replacements
  6264. PUGI_IMPL_NS_BEGIN
  6265. struct equal_to
  6266. {
  6267. template <typename T> bool operator()(const T& lhs, const T& rhs) const
  6268. {
  6269. return lhs == rhs;
  6270. }
  6271. };
  6272. struct not_equal_to
  6273. {
  6274. template <typename T> bool operator()(const T& lhs, const T& rhs) const
  6275. {
  6276. return lhs != rhs;
  6277. }
  6278. };
  6279. struct less
  6280. {
  6281. template <typename T> bool operator()(const T& lhs, const T& rhs) const
  6282. {
  6283. return lhs < rhs;
  6284. }
  6285. };
  6286. struct less_equal
  6287. {
  6288. template <typename T> bool operator()(const T& lhs, const T& rhs) const
  6289. {
  6290. return lhs <= rhs;
  6291. }
  6292. };
  6293. template <typename T> inline void swap(T& lhs, T& rhs)
  6294. {
  6295. T temp = lhs;
  6296. lhs = rhs;
  6297. rhs = temp;
  6298. }
  6299. template <typename I, typename Pred> PUGI_IMPL_FN I min_element(I begin, I end, const Pred& pred)
  6300. {
  6301. I result = begin;
  6302. for (I it = begin + 1; it != end; ++it)
  6303. if (pred(*it, *result))
  6304. result = it;
  6305. return result;
  6306. }
  6307. template <typename I> PUGI_IMPL_FN void reverse(I begin, I end)
  6308. {
  6309. while (end - begin > 1)
  6310. swap(*begin++, *--end);
  6311. }
  6312. template <typename I> PUGI_IMPL_FN I unique(I begin, I end)
  6313. {
  6314. // fast skip head
  6315. while (end - begin > 1 && *begin != *(begin + 1))
  6316. begin++;
  6317. if (begin == end)
  6318. return begin;
  6319. // last written element
  6320. I write = begin++;
  6321. // merge unique elements
  6322. while (begin != end)
  6323. {
  6324. if (*begin != *write)
  6325. *++write = *begin++;
  6326. else
  6327. begin++;
  6328. }
  6329. // past-the-end (write points to live element)
  6330. return write + 1;
  6331. }
  6332. template <typename T, typename Pred> PUGI_IMPL_FN void insertion_sort(T* begin, T* end, const Pred& pred)
  6333. {
  6334. if (begin == end)
  6335. return;
  6336. for (T* it = begin + 1; it != end; ++it)
  6337. {
  6338. T val = *it;
  6339. T* hole = it;
  6340. // move hole backwards
  6341. while (hole > begin && pred(val, *(hole - 1)))
  6342. {
  6343. *hole = *(hole - 1);
  6344. hole--;
  6345. }
  6346. // fill hole with element
  6347. *hole = val;
  6348. }
  6349. }
  6350. template <typename I, typename Pred> inline I median3(I first, I middle, I last, const Pred& pred)
  6351. {
  6352. if (pred(*middle, *first))
  6353. swap(middle, first);
  6354. if (pred(*last, *middle))
  6355. swap(last, middle);
  6356. if (pred(*middle, *first))
  6357. swap(middle, first);
  6358. return middle;
  6359. }
  6360. template <typename T, typename Pred> PUGI_IMPL_FN void partition3(T* begin, T* end, T pivot, const Pred& pred, T** out_eqbeg, T** out_eqend)
  6361. {
  6362. // invariant: array is split into 4 groups: = < ? > (each variable denotes the boundary between the groups)
  6363. T* eq = begin;
  6364. T* lt = begin;
  6365. T* gt = end;
  6366. while (lt < gt)
  6367. {
  6368. if (pred(*lt, pivot))
  6369. lt++;
  6370. else if (*lt == pivot)
  6371. swap(*eq++, *lt++);
  6372. else
  6373. swap(*lt, *--gt);
  6374. }
  6375. // we now have just 4 groups: = < >; move equal elements to the middle
  6376. T* eqbeg = gt;
  6377. for (T* it = begin; it != eq; ++it)
  6378. swap(*it, *--eqbeg);
  6379. *out_eqbeg = eqbeg;
  6380. *out_eqend = gt;
  6381. }
  6382. template <typename I, typename Pred> PUGI_IMPL_FN void sort(I begin, I end, const Pred& pred)
  6383. {
  6384. // sort large chunks
  6385. while (end - begin > 16)
  6386. {
  6387. // find median element
  6388. I middle = begin + (end - begin) / 2;
  6389. I median = median3(begin, middle, end - 1, pred);
  6390. // partition in three chunks (< = >)
  6391. I eqbeg, eqend;
  6392. partition3(begin, end, *median, pred, &eqbeg, &eqend);
  6393. // loop on larger half
  6394. if (eqbeg - begin > end - eqend)
  6395. {
  6396. sort(eqend, end, pred);
  6397. end = eqbeg;
  6398. }
  6399. else
  6400. {
  6401. sort(begin, eqbeg, pred);
  6402. begin = eqend;
  6403. }
  6404. }
  6405. // insertion sort small chunk
  6406. insertion_sort(begin, end, pred);
  6407. }
  6408. PUGI_IMPL_FN bool hash_insert(const void** table, size_t size, const void* key)
  6409. {
  6410. assert(key);
  6411. unsigned int h = static_cast<unsigned int>(reinterpret_cast<uintptr_t>(key));
  6412. // MurmurHash3 32-bit finalizer
  6413. h ^= h >> 16;
  6414. h *= 0x85ebca6bu;
  6415. h ^= h >> 13;
  6416. h *= 0xc2b2ae35u;
  6417. h ^= h >> 16;
  6418. size_t hashmod = size - 1;
  6419. size_t bucket = h & hashmod;
  6420. for (size_t probe = 0; probe <= hashmod; ++probe)
  6421. {
  6422. if (table[bucket] == NULL)
  6423. {
  6424. table[bucket] = key;
  6425. return true;
  6426. }
  6427. if (table[bucket] == key)
  6428. return false;
  6429. // hash collision, quadratic probing
  6430. bucket = (bucket + probe + 1) & hashmod;
  6431. }
  6432. assert(false && "Hash table is full"); // unreachable
  6433. return false;
  6434. }
  6435. PUGI_IMPL_NS_END
  6436. // Allocator used for AST and evaluation stacks
  6437. PUGI_IMPL_NS_BEGIN
  6438. static const size_t xpath_memory_page_size =
  6439. #ifdef PUGIXML_MEMORY_XPATH_PAGE_SIZE
  6440. PUGIXML_MEMORY_XPATH_PAGE_SIZE
  6441. #else
  6442. 4096
  6443. #endif
  6444. ;
  6445. static const uintptr_t xpath_memory_block_alignment = sizeof(double) > sizeof(void*) ? sizeof(double) : sizeof(void*);
  6446. struct xpath_memory_block
  6447. {
  6448. xpath_memory_block* next;
  6449. size_t capacity;
  6450. union
  6451. {
  6452. char data[xpath_memory_page_size];
  6453. double alignment;
  6454. };
  6455. };
  6456. struct xpath_allocator
  6457. {
  6458. xpath_memory_block* _root;
  6459. size_t _root_size;
  6460. bool* _error;
  6461. xpath_allocator(xpath_memory_block* root, bool* error = NULL): _root(root), _root_size(0), _error(error)
  6462. {
  6463. }
  6464. void* allocate(size_t size)
  6465. {
  6466. // round size up to block alignment boundary
  6467. size = (size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1);
  6468. if (_root_size + size <= _root->capacity)
  6469. {
  6470. void* buf = &_root->data[0] + _root_size;
  6471. _root_size += size;
  6472. return buf;
  6473. }
  6474. else
  6475. {
  6476. // make sure we have at least 1/4th of the page free after allocation to satisfy subsequent allocation requests
  6477. size_t block_capacity_base = sizeof(_root->data);
  6478. size_t block_capacity_req = size + block_capacity_base / 4;
  6479. size_t block_capacity = (block_capacity_base > block_capacity_req) ? block_capacity_base : block_capacity_req;
  6480. size_t block_size = block_capacity + offsetof(xpath_memory_block, data);
  6481. xpath_memory_block* block = static_cast<xpath_memory_block*>(xml_memory::allocate(block_size));
  6482. if (!block)
  6483. {
  6484. if (_error) *_error = true;
  6485. return NULL;
  6486. }
  6487. block->next = _root;
  6488. block->capacity = block_capacity;
  6489. _root = block;
  6490. _root_size = size;
  6491. return block->data;
  6492. }
  6493. }
  6494. void* reallocate(void* ptr, size_t old_size, size_t new_size)
  6495. {
  6496. // round size up to block alignment boundary
  6497. old_size = (old_size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1);
  6498. new_size = (new_size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1);
  6499. // we can only reallocate the last object
  6500. assert(ptr == NULL || static_cast<char*>(ptr) + old_size == &_root->data[0] + _root_size);
  6501. // try to reallocate the object inplace
  6502. if (ptr && _root_size - old_size + new_size <= _root->capacity)
  6503. {
  6504. _root_size = _root_size - old_size + new_size;
  6505. return ptr;
  6506. }
  6507. // allocate a new block
  6508. void* result = allocate(new_size);
  6509. if (!result) return NULL;
  6510. // we have a new block
  6511. if (ptr)
  6512. {
  6513. // copy old data (we only support growing)
  6514. assert(new_size >= old_size);
  6515. memcpy(result, ptr, old_size);
  6516. // free the previous page if it had no other objects
  6517. assert(_root->data == result);
  6518. assert(_root->next);
  6519. if (_root->next->data == ptr)
  6520. {
  6521. // deallocate the whole page, unless it was the first one
  6522. xpath_memory_block* next = _root->next->next;
  6523. if (next)
  6524. {
  6525. xml_memory::deallocate(_root->next);
  6526. _root->next = next;
  6527. }
  6528. }
  6529. }
  6530. return result;
  6531. }
  6532. void revert(const xpath_allocator& state)
  6533. {
  6534. // free all new pages
  6535. xpath_memory_block* cur = _root;
  6536. while (cur != state._root)
  6537. {
  6538. xpath_memory_block* next = cur->next;
  6539. xml_memory::deallocate(cur);
  6540. cur = next;
  6541. }
  6542. // restore state
  6543. _root = state._root;
  6544. _root_size = state._root_size;
  6545. }
  6546. void release()
  6547. {
  6548. xpath_memory_block* cur = _root;
  6549. assert(cur);
  6550. while (cur->next)
  6551. {
  6552. xpath_memory_block* next = cur->next;
  6553. xml_memory::deallocate(cur);
  6554. cur = next;
  6555. }
  6556. }
  6557. };
  6558. struct xpath_allocator_capture
  6559. {
  6560. xpath_allocator_capture(xpath_allocator* alloc): _target(alloc), _state(*alloc)
  6561. {
  6562. }
  6563. ~xpath_allocator_capture()
  6564. {
  6565. _target->revert(_state);
  6566. }
  6567. xpath_allocator* _target;
  6568. xpath_allocator _state;
  6569. };
  6570. struct xpath_stack
  6571. {
  6572. xpath_allocator* result;
  6573. xpath_allocator* temp;
  6574. };
  6575. struct xpath_stack_data
  6576. {
  6577. xpath_memory_block blocks[2];
  6578. xpath_allocator result;
  6579. xpath_allocator temp;
  6580. xpath_stack stack;
  6581. bool oom;
  6582. xpath_stack_data(): result(blocks + 0, &oom), temp(blocks + 1, &oom), oom(false)
  6583. {
  6584. blocks[0].next = blocks[1].next = NULL;
  6585. blocks[0].capacity = blocks[1].capacity = sizeof(blocks[0].data);
  6586. stack.result = &result;
  6587. stack.temp = &temp;
  6588. }
  6589. ~xpath_stack_data()
  6590. {
  6591. result.release();
  6592. temp.release();
  6593. }
  6594. };
  6595. PUGI_IMPL_NS_END
  6596. // String class
  6597. PUGI_IMPL_NS_BEGIN
  6598. class xpath_string
  6599. {
  6600. const char_t* _buffer;
  6601. bool _uses_heap;
  6602. size_t _length_heap;
  6603. static char_t* duplicate_string(const char_t* string, size_t length, xpath_allocator* alloc)
  6604. {
  6605. char_t* result = static_cast<char_t*>(alloc->allocate((length + 1) * sizeof(char_t)));
  6606. if (!result) return NULL;
  6607. memcpy(result, string, length * sizeof(char_t));
  6608. result[length] = 0;
  6609. return result;
  6610. }
  6611. xpath_string(const char_t* buffer, bool uses_heap_, size_t length_heap): _buffer(buffer), _uses_heap(uses_heap_), _length_heap(length_heap)
  6612. {
  6613. }
  6614. public:
  6615. static xpath_string from_const(const char_t* str)
  6616. {
  6617. return xpath_string(str, false, 0);
  6618. }
  6619. static xpath_string from_heap_preallocated(const char_t* begin, const char_t* end)
  6620. {
  6621. assert(begin <= end && *end == 0);
  6622. return xpath_string(begin, true, static_cast<size_t>(end - begin));
  6623. }
  6624. static xpath_string from_heap(const char_t* begin, const char_t* end, xpath_allocator* alloc)
  6625. {
  6626. assert(begin <= end);
  6627. if (begin == end)
  6628. return xpath_string();
  6629. size_t length = static_cast<size_t>(end - begin);
  6630. const char_t* data = duplicate_string(begin, length, alloc);
  6631. return data ? xpath_string(data, true, length) : xpath_string();
  6632. }
  6633. xpath_string(): _buffer(PUGIXML_TEXT("")), _uses_heap(false), _length_heap(0)
  6634. {
  6635. }
  6636. void append(const xpath_string& o, xpath_allocator* alloc)
  6637. {
  6638. // skip empty sources
  6639. if (!*o._buffer) return;
  6640. // fast append for constant empty target and constant source
  6641. if (!*_buffer && !_uses_heap && !o._uses_heap)
  6642. {
  6643. _buffer = o._buffer;
  6644. }
  6645. else
  6646. {
  6647. // need to make heap copy
  6648. size_t target_length = length();
  6649. size_t source_length = o.length();
  6650. size_t result_length = target_length + source_length;
  6651. // allocate new buffer
  6652. char_t* result = static_cast<char_t*>(alloc->reallocate(_uses_heap ? const_cast<char_t*>(_buffer) : NULL, (target_length + 1) * sizeof(char_t), (result_length + 1) * sizeof(char_t)));
  6653. if (!result) return;
  6654. // append first string to the new buffer in case there was no reallocation
  6655. if (!_uses_heap) memcpy(result, _buffer, target_length * sizeof(char_t));
  6656. // append second string to the new buffer
  6657. memcpy(result + target_length, o._buffer, source_length * sizeof(char_t));
  6658. result[result_length] = 0;
  6659. // finalize
  6660. _buffer = result;
  6661. _uses_heap = true;
  6662. _length_heap = result_length;
  6663. }
  6664. }
  6665. const char_t* c_str() const
  6666. {
  6667. return _buffer;
  6668. }
  6669. size_t length() const
  6670. {
  6671. return _uses_heap ? _length_heap : strlength(_buffer);
  6672. }
  6673. char_t* data(xpath_allocator* alloc)
  6674. {
  6675. // make private heap copy
  6676. if (!_uses_heap)
  6677. {
  6678. size_t length_ = strlength(_buffer);
  6679. const char_t* data_ = duplicate_string(_buffer, length_, alloc);
  6680. if (!data_) return NULL;
  6681. _buffer = data_;
  6682. _uses_heap = true;
  6683. _length_heap = length_;
  6684. }
  6685. return const_cast<char_t*>(_buffer);
  6686. }
  6687. bool empty() const
  6688. {
  6689. return *_buffer == 0;
  6690. }
  6691. bool operator==(const xpath_string& o) const
  6692. {
  6693. return strequal(_buffer, o._buffer);
  6694. }
  6695. bool operator!=(const xpath_string& o) const
  6696. {
  6697. return !strequal(_buffer, o._buffer);
  6698. }
  6699. bool uses_heap() const
  6700. {
  6701. return _uses_heap;
  6702. }
  6703. };
  6704. PUGI_IMPL_NS_END
  6705. PUGI_IMPL_NS_BEGIN
  6706. PUGI_IMPL_FN bool starts_with(const char_t* string, const char_t* pattern)
  6707. {
  6708. while (*pattern && *string == *pattern)
  6709. {
  6710. string++;
  6711. pattern++;
  6712. }
  6713. return *pattern == 0;
  6714. }
  6715. PUGI_IMPL_FN const char_t* find_char(const char_t* s, char_t c)
  6716. {
  6717. #ifdef PUGIXML_WCHAR_MODE
  6718. return wcschr(s, c);
  6719. #else
  6720. return strchr(s, c);
  6721. #endif
  6722. }
  6723. PUGI_IMPL_FN const char_t* find_substring(const char_t* s, const char_t* p)
  6724. {
  6725. #ifdef PUGIXML_WCHAR_MODE
  6726. // MSVC6 wcsstr bug workaround (if s is empty it always returns 0)
  6727. return (*p == 0) ? s : wcsstr(s, p);
  6728. #else
  6729. return strstr(s, p);
  6730. #endif
  6731. }
  6732. // Converts symbol to lower case, if it is an ASCII one
  6733. PUGI_IMPL_FN char_t tolower_ascii(char_t ch)
  6734. {
  6735. return static_cast<unsigned int>(ch - 'A') < 26 ? static_cast<char_t>(ch | ' ') : ch;
  6736. }
  6737. PUGI_IMPL_FN xpath_string string_value(const xpath_node& na, xpath_allocator* alloc)
  6738. {
  6739. if (na.attribute())
  6740. return xpath_string::from_const(na.attribute().value());
  6741. else
  6742. {
  6743. xml_node n = na.node();
  6744. switch (n.type())
  6745. {
  6746. case node_pcdata:
  6747. case node_cdata:
  6748. case node_comment:
  6749. case node_pi:
  6750. return xpath_string::from_const(n.value());
  6751. case node_document:
  6752. case node_element:
  6753. {
  6754. xpath_string result;
  6755. // element nodes can have value if parse_embed_pcdata was used
  6756. if (n.value()[0])
  6757. result.append(xpath_string::from_const(n.value()), alloc);
  6758. xml_node cur = n.first_child();
  6759. while (cur && cur != n)
  6760. {
  6761. if (cur.type() == node_pcdata || cur.type() == node_cdata)
  6762. result.append(xpath_string::from_const(cur.value()), alloc);
  6763. if (cur.first_child())
  6764. cur = cur.first_child();
  6765. else if (cur.next_sibling())
  6766. cur = cur.next_sibling();
  6767. else
  6768. {
  6769. while (!cur.next_sibling() && cur != n)
  6770. cur = cur.parent();
  6771. if (cur != n) cur = cur.next_sibling();
  6772. }
  6773. }
  6774. return result;
  6775. }
  6776. default:
  6777. return xpath_string();
  6778. }
  6779. }
  6780. }
  6781. PUGI_IMPL_FN bool node_is_before_sibling(xml_node_struct* ln, xml_node_struct* rn)
  6782. {
  6783. assert(ln->parent == rn->parent);
  6784. // there is no common ancestor (the shared parent is null), nodes are from different documents
  6785. if (!ln->parent) return ln < rn;
  6786. // determine sibling order
  6787. xml_node_struct* ls = ln;
  6788. xml_node_struct* rs = rn;
  6789. while (ls && rs)
  6790. {
  6791. if (ls == rn) return true;
  6792. if (rs == ln) return false;
  6793. ls = ls->next_sibling;
  6794. rs = rs->next_sibling;
  6795. }
  6796. // if rn sibling chain ended ln must be before rn
  6797. return !rs;
  6798. }
  6799. PUGI_IMPL_FN bool node_is_before(xml_node_struct* ln, xml_node_struct* rn)
  6800. {
  6801. // find common ancestor at the same depth, if any
  6802. xml_node_struct* lp = ln;
  6803. xml_node_struct* rp = rn;
  6804. while (lp && rp && lp->parent != rp->parent)
  6805. {
  6806. lp = lp->parent;
  6807. rp = rp->parent;
  6808. }
  6809. // parents are the same!
  6810. if (lp && rp) return node_is_before_sibling(lp, rp);
  6811. // nodes are at different depths, need to normalize heights
  6812. bool left_higher = !lp;
  6813. while (lp)
  6814. {
  6815. lp = lp->parent;
  6816. ln = ln->parent;
  6817. }
  6818. while (rp)
  6819. {
  6820. rp = rp->parent;
  6821. rn = rn->parent;
  6822. }
  6823. // one node is the ancestor of the other
  6824. if (ln == rn) return left_higher;
  6825. // find common ancestor... again
  6826. while (ln->parent != rn->parent)
  6827. {
  6828. ln = ln->parent;
  6829. rn = rn->parent;
  6830. }
  6831. return node_is_before_sibling(ln, rn);
  6832. }
  6833. PUGI_IMPL_FN bool node_is_ancestor(xml_node_struct* parent, xml_node_struct* node)
  6834. {
  6835. while (node && node != parent) node = node->parent;
  6836. return parent && node == parent;
  6837. }
  6838. PUGI_IMPL_FN const void* document_buffer_order(const xpath_node& xnode)
  6839. {
  6840. xml_node_struct* node = xnode.node().internal_object();
  6841. if (node)
  6842. {
  6843. if ((get_document(node).header & xml_memory_page_contents_shared_mask) == 0)
  6844. {
  6845. if (node->name && (node->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0) return node->name;
  6846. if (node->value && (node->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0) return node->value;
  6847. }
  6848. return NULL;
  6849. }
  6850. xml_attribute_struct* attr = xnode.attribute().internal_object();
  6851. if (attr)
  6852. {
  6853. if ((get_document(attr).header & xml_memory_page_contents_shared_mask) == 0)
  6854. {
  6855. if ((attr->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0) return attr->name;
  6856. if ((attr->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0) return attr->value;
  6857. }
  6858. return NULL;
  6859. }
  6860. return NULL;
  6861. }
  6862. struct document_order_comparator
  6863. {
  6864. bool operator()(const xpath_node& lhs, const xpath_node& rhs) const
  6865. {
  6866. // optimized document order based check
  6867. const void* lo = document_buffer_order(lhs);
  6868. const void* ro = document_buffer_order(rhs);
  6869. if (lo && ro) return lo < ro;
  6870. // slow comparison
  6871. xml_node ln = lhs.node(), rn = rhs.node();
  6872. // compare attributes
  6873. if (lhs.attribute() && rhs.attribute())
  6874. {
  6875. // shared parent
  6876. if (lhs.parent() == rhs.parent())
  6877. {
  6878. // determine sibling order
  6879. for (xml_attribute a = lhs.attribute(); a; a = a.next_attribute())
  6880. if (a == rhs.attribute())
  6881. return true;
  6882. return false;
  6883. }
  6884. // compare attribute parents
  6885. ln = lhs.parent();
  6886. rn = rhs.parent();
  6887. }
  6888. else if (lhs.attribute())
  6889. {
  6890. // attributes go after the parent element
  6891. if (lhs.parent() == rhs.node()) return false;
  6892. ln = lhs.parent();
  6893. }
  6894. else if (rhs.attribute())
  6895. {
  6896. // attributes go after the parent element
  6897. if (rhs.parent() == lhs.node()) return true;
  6898. rn = rhs.parent();
  6899. }
  6900. if (ln == rn) return false;
  6901. if (!ln || !rn) return ln < rn;
  6902. return node_is_before(ln.internal_object(), rn.internal_object());
  6903. }
  6904. };
  6905. PUGI_IMPL_FN double gen_nan()
  6906. {
  6907. #if defined(__STDC_IEC_559__) || ((FLT_RADIX - 0 == 2) && (FLT_MAX_EXP - 0 == 128) && (FLT_MANT_DIG - 0 == 24))
  6908. PUGI_IMPL_STATIC_ASSERT(sizeof(float) == sizeof(uint32_t));
  6909. typedef uint32_t UI; // BCC5 workaround
  6910. union { float f; UI i; } u;
  6911. u.i = 0x7fc00000;
  6912. return double(u.f);
  6913. #else
  6914. // fallback
  6915. const volatile double zero = 0.0;
  6916. return zero / zero;
  6917. #endif
  6918. }
  6919. PUGI_IMPL_FN bool is_nan(double value)
  6920. {
  6921. #if defined(PUGI_IMPL_MSVC_CRT_VERSION) || defined(__BORLANDC__)
  6922. return !!_isnan(value);
  6923. #elif defined(fpclassify) && defined(FP_NAN)
  6924. return fpclassify(value) == FP_NAN;
  6925. #else
  6926. // fallback
  6927. const volatile double v = value;
  6928. return v != v;
  6929. #endif
  6930. }
  6931. PUGI_IMPL_FN const char_t* convert_number_to_string_special(double value)
  6932. {
  6933. #if defined(PUGI_IMPL_MSVC_CRT_VERSION) || defined(__BORLANDC__)
  6934. if (_finite(value)) return (value == 0) ? PUGIXML_TEXT("0") : 0;
  6935. if (_isnan(value)) return PUGIXML_TEXT("NaN");
  6936. return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity");
  6937. #elif defined(fpclassify) && defined(FP_NAN) && defined(FP_INFINITE) && defined(FP_ZERO)
  6938. switch (fpclassify(value))
  6939. {
  6940. case FP_NAN:
  6941. return PUGIXML_TEXT("NaN");
  6942. case FP_INFINITE:
  6943. return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity");
  6944. case FP_ZERO:
  6945. return PUGIXML_TEXT("0");
  6946. default:
  6947. return 0;
  6948. }
  6949. #else
  6950. // fallback
  6951. const volatile double v = value;
  6952. if (v == 0) return PUGIXML_TEXT("0");
  6953. if (v != v) return PUGIXML_TEXT("NaN");
  6954. if (v * 2 == v) return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity");
  6955. return NULL;
  6956. #endif
  6957. }
  6958. PUGI_IMPL_FN bool convert_number_to_boolean(double value)
  6959. {
  6960. return (value != 0 && !is_nan(value));
  6961. }
  6962. PUGI_IMPL_FN void truncate_zeros(char* begin, char* end)
  6963. {
  6964. while (begin != end && end[-1] == '0') end--;
  6965. *end = 0;
  6966. }
  6967. // gets mantissa digits in the form of 0.xxxxx with 0. implied and the exponent
  6968. #if defined(PUGI_IMPL_MSVC_CRT_VERSION) && PUGI_IMPL_MSVC_CRT_VERSION >= 1400
  6969. PUGI_IMPL_FN void convert_number_to_mantissa_exponent(double value, char (&buffer)[32], char** out_mantissa, int* out_exponent)
  6970. {
  6971. // get base values
  6972. int sign, exponent;
  6973. _ecvt_s(buffer, sizeof(buffer), value, DBL_DIG + 1, &exponent, &sign);
  6974. // truncate redundant zeros
  6975. truncate_zeros(buffer, buffer + strlen(buffer));
  6976. // fill results
  6977. *out_mantissa = buffer;
  6978. *out_exponent = exponent;
  6979. }
  6980. #else
  6981. PUGI_IMPL_FN void convert_number_to_mantissa_exponent(double value, char (&buffer)[32], char** out_mantissa, int* out_exponent)
  6982. {
  6983. // get a scientific notation value with IEEE DBL_DIG decimals
  6984. PUGI_IMPL_SNPRINTF(buffer, "%.*e", DBL_DIG, value);
  6985. // get the exponent (possibly negative)
  6986. char* exponent_string = strchr(buffer, 'e');
  6987. assert(exponent_string);
  6988. int exponent = atoi(exponent_string + 1);
  6989. // extract mantissa string: skip sign
  6990. char* mantissa = buffer[0] == '-' ? buffer + 1 : buffer;
  6991. assert(mantissa[0] != '0' && (mantissa[1] == '.' || mantissa[1] == ','));
  6992. // divide mantissa by 10 to eliminate integer part
  6993. mantissa[1] = mantissa[0];
  6994. mantissa++;
  6995. exponent++;
  6996. // remove extra mantissa digits and zero-terminate mantissa
  6997. truncate_zeros(mantissa, exponent_string);
  6998. // fill results
  6999. *out_mantissa = mantissa;
  7000. *out_exponent = exponent;
  7001. }
  7002. #endif
  7003. PUGI_IMPL_FN xpath_string convert_number_to_string(double value, xpath_allocator* alloc)
  7004. {
  7005. // try special number conversion
  7006. const char_t* special = convert_number_to_string_special(value);
  7007. if (special) return xpath_string::from_const(special);
  7008. // get mantissa + exponent form
  7009. char mantissa_buffer[32];
  7010. char* mantissa;
  7011. int exponent;
  7012. convert_number_to_mantissa_exponent(value, mantissa_buffer, &mantissa, &exponent);
  7013. // allocate a buffer of suitable length for the number
  7014. size_t result_size = strlen(mantissa_buffer) + (exponent > 0 ? exponent : -exponent) + 4;
  7015. char_t* result = static_cast<char_t*>(alloc->allocate(sizeof(char_t) * result_size));
  7016. if (!result) return xpath_string();
  7017. // make the number!
  7018. char_t* s = result;
  7019. // sign
  7020. if (value < 0) *s++ = '-';
  7021. // integer part
  7022. if (exponent <= 0)
  7023. {
  7024. *s++ = '0';
  7025. }
  7026. else
  7027. {
  7028. while (exponent > 0)
  7029. {
  7030. assert(*mantissa == 0 || static_cast<unsigned int>(*mantissa - '0') <= 9);
  7031. *s++ = *mantissa ? *mantissa++ : '0';
  7032. exponent--;
  7033. }
  7034. }
  7035. // fractional part
  7036. if (*mantissa)
  7037. {
  7038. // decimal point
  7039. *s++ = '.';
  7040. // extra zeroes from negative exponent
  7041. while (exponent < 0)
  7042. {
  7043. *s++ = '0';
  7044. exponent++;
  7045. }
  7046. // extra mantissa digits
  7047. while (*mantissa)
  7048. {
  7049. assert(static_cast<unsigned int>(*mantissa - '0') <= 9);
  7050. *s++ = *mantissa++;
  7051. }
  7052. }
  7053. // zero-terminate
  7054. assert(s < result + result_size);
  7055. *s = 0;
  7056. return xpath_string::from_heap_preallocated(result, s);
  7057. }
  7058. PUGI_IMPL_FN bool check_string_to_number_format(const char_t* string)
  7059. {
  7060. // parse leading whitespace
  7061. while (PUGI_IMPL_IS_CHARTYPE(*string, ct_space)) ++string;
  7062. // parse sign
  7063. if (*string == '-') ++string;
  7064. if (!*string) return false;
  7065. // if there is no integer part, there should be a decimal part with at least one digit
  7066. if (!PUGI_IMPL_IS_CHARTYPEX(string[0], ctx_digit) && (string[0] != '.' || !PUGI_IMPL_IS_CHARTYPEX(string[1], ctx_digit))) return false;
  7067. // parse integer part
  7068. while (PUGI_IMPL_IS_CHARTYPEX(*string, ctx_digit)) ++string;
  7069. // parse decimal part
  7070. if (*string == '.')
  7071. {
  7072. ++string;
  7073. while (PUGI_IMPL_IS_CHARTYPEX(*string, ctx_digit)) ++string;
  7074. }
  7075. // parse trailing whitespace
  7076. while (PUGI_IMPL_IS_CHARTYPE(*string, ct_space)) ++string;
  7077. return *string == 0;
  7078. }
  7079. PUGI_IMPL_FN double convert_string_to_number(const char_t* string)
  7080. {
  7081. // check string format
  7082. if (!check_string_to_number_format(string)) return gen_nan();
  7083. // parse string
  7084. #ifdef PUGIXML_WCHAR_MODE
  7085. return wcstod(string, NULL);
  7086. #else
  7087. return strtod(string, NULL);
  7088. #endif
  7089. }
  7090. PUGI_IMPL_FN bool convert_string_to_number_scratch(char_t (&buffer)[32], const char_t* begin, const char_t* end, double* out_result)
  7091. {
  7092. size_t length = static_cast<size_t>(end - begin);
  7093. char_t* scratch = buffer;
  7094. if (length >= sizeof(buffer) / sizeof(buffer[0]))
  7095. {
  7096. // need to make dummy on-heap copy
  7097. scratch = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
  7098. if (!scratch) return false;
  7099. }
  7100. // copy string to zero-terminated buffer and perform conversion
  7101. memcpy(scratch, begin, length * sizeof(char_t));
  7102. scratch[length] = 0;
  7103. *out_result = convert_string_to_number(scratch);
  7104. // free dummy buffer
  7105. if (scratch != buffer) xml_memory::deallocate(scratch);
  7106. return true;
  7107. }
  7108. PUGI_IMPL_FN double round_nearest(double value)
  7109. {
  7110. return floor(value + 0.5);
  7111. }
  7112. PUGI_IMPL_FN double round_nearest_nzero(double value)
  7113. {
  7114. // same as round_nearest, but returns -0 for [-0.5, -0]
  7115. // ceil is used to differentiate between +0 and -0 (we return -0 for [-0.5, -0] and +0 for +0)
  7116. return (value >= -0.5 && value <= 0) ? ceil(value) : floor(value + 0.5);
  7117. }
  7118. PUGI_IMPL_FN const char_t* qualified_name(const xpath_node& node)
  7119. {
  7120. return node.attribute() ? node.attribute().name() : node.node().name();
  7121. }
  7122. PUGI_IMPL_FN const char_t* local_name(const xpath_node& node)
  7123. {
  7124. const char_t* name = qualified_name(node);
  7125. const char_t* p = find_char(name, ':');
  7126. return p ? p + 1 : name;
  7127. }
  7128. struct namespace_uri_predicate
  7129. {
  7130. const char_t* prefix;
  7131. size_t prefix_length;
  7132. namespace_uri_predicate(const char_t* name)
  7133. {
  7134. const char_t* pos = find_char(name, ':');
  7135. prefix = pos ? name : NULL;
  7136. prefix_length = pos ? static_cast<size_t>(pos - name) : 0;
  7137. }
  7138. bool operator()(xml_attribute a) const
  7139. {
  7140. const char_t* name = a.name();
  7141. if (!starts_with(name, PUGIXML_TEXT("xmlns"))) return false;
  7142. return prefix ? name[5] == ':' && strequalrange(name + 6, prefix, prefix_length) : name[5] == 0;
  7143. }
  7144. };
  7145. PUGI_IMPL_FN const char_t* namespace_uri(xml_node node)
  7146. {
  7147. namespace_uri_predicate pred = node.name();
  7148. xml_node p = node;
  7149. while (p)
  7150. {
  7151. xml_attribute a = p.find_attribute(pred);
  7152. if (a) return a.value();
  7153. p = p.parent();
  7154. }
  7155. return PUGIXML_TEXT("");
  7156. }
  7157. PUGI_IMPL_FN const char_t* namespace_uri(xml_attribute attr, xml_node parent)
  7158. {
  7159. namespace_uri_predicate pred = attr.name();
  7160. // Default namespace does not apply to attributes
  7161. if (!pred.prefix) return PUGIXML_TEXT("");
  7162. xml_node p = parent;
  7163. while (p)
  7164. {
  7165. xml_attribute a = p.find_attribute(pred);
  7166. if (a) return a.value();
  7167. p = p.parent();
  7168. }
  7169. return PUGIXML_TEXT("");
  7170. }
  7171. PUGI_IMPL_FN const char_t* namespace_uri(const xpath_node& node)
  7172. {
  7173. return node.attribute() ? namespace_uri(node.attribute(), node.parent()) : namespace_uri(node.node());
  7174. }
  7175. PUGI_IMPL_FN char_t* normalize_space(char_t* buffer)
  7176. {
  7177. char_t* write = buffer;
  7178. for (char_t* it = buffer; *it; )
  7179. {
  7180. char_t ch = *it++;
  7181. if (PUGI_IMPL_IS_CHARTYPE(ch, ct_space))
  7182. {
  7183. // replace whitespace sequence with single space
  7184. while (PUGI_IMPL_IS_CHARTYPE(*it, ct_space)) it++;
  7185. // avoid leading spaces
  7186. if (write != buffer) *write++ = ' ';
  7187. }
  7188. else *write++ = ch;
  7189. }
  7190. // remove trailing space
  7191. if (write != buffer && PUGI_IMPL_IS_CHARTYPE(write[-1], ct_space)) write--;
  7192. // zero-terminate
  7193. *write = 0;
  7194. return write;
  7195. }
  7196. PUGI_IMPL_FN char_t* translate(char_t* buffer, const char_t* from, const char_t* to, size_t to_length)
  7197. {
  7198. char_t* write = buffer;
  7199. while (*buffer)
  7200. {
  7201. PUGI_IMPL_DMC_VOLATILE char_t ch = *buffer++;
  7202. const char_t* pos = find_char(from, ch);
  7203. if (!pos)
  7204. *write++ = ch; // do not process
  7205. else if (static_cast<size_t>(pos - from) < to_length)
  7206. *write++ = to[pos - from]; // replace
  7207. }
  7208. // zero-terminate
  7209. *write = 0;
  7210. return write;
  7211. }
  7212. PUGI_IMPL_FN unsigned char* translate_table_generate(xpath_allocator* alloc, const char_t* from, const char_t* to)
  7213. {
  7214. unsigned char table[128] = {0};
  7215. while (*from)
  7216. {
  7217. unsigned int fc = static_cast<unsigned int>(*from);
  7218. unsigned int tc = static_cast<unsigned int>(*to);
  7219. if (fc >= 128 || tc >= 128)
  7220. return NULL;
  7221. // code=128 means "skip character"
  7222. if (!table[fc])
  7223. table[fc] = static_cast<unsigned char>(tc ? tc : 128);
  7224. from++;
  7225. if (tc) to++;
  7226. }
  7227. for (int i = 0; i < 128; ++i)
  7228. if (!table[i])
  7229. table[i] = static_cast<unsigned char>(i);
  7230. void* result = alloc->allocate(sizeof(table));
  7231. if (!result) return NULL;
  7232. memcpy(result, table, sizeof(table));
  7233. return static_cast<unsigned char*>(result);
  7234. }
  7235. PUGI_IMPL_FN char_t* translate_table(char_t* buffer, const unsigned char* table)
  7236. {
  7237. char_t* write = buffer;
  7238. while (*buffer)
  7239. {
  7240. char_t ch = *buffer++;
  7241. unsigned int index = static_cast<unsigned int>(ch);
  7242. if (index < 128)
  7243. {
  7244. unsigned char code = table[index];
  7245. // code=128 means "skip character" (table size is 128 so 128 can be a special value)
  7246. // this code skips these characters without extra branches
  7247. *write = static_cast<char_t>(code);
  7248. write += 1 - (code >> 7);
  7249. }
  7250. else
  7251. {
  7252. *write++ = ch;
  7253. }
  7254. }
  7255. // zero-terminate
  7256. *write = 0;
  7257. return write;
  7258. }
  7259. inline bool is_xpath_attribute(const char_t* name)
  7260. {
  7261. return !(starts_with(name, PUGIXML_TEXT("xmlns")) && (name[5] == 0 || name[5] == ':'));
  7262. }
  7263. struct xpath_variable_boolean: xpath_variable
  7264. {
  7265. xpath_variable_boolean(): xpath_variable(xpath_type_boolean), value(false)
  7266. {
  7267. }
  7268. bool value;
  7269. char_t name[1];
  7270. };
  7271. struct xpath_variable_number: xpath_variable
  7272. {
  7273. xpath_variable_number(): xpath_variable(xpath_type_number), value(0)
  7274. {
  7275. }
  7276. double value;
  7277. char_t name[1];
  7278. };
  7279. struct xpath_variable_string: xpath_variable
  7280. {
  7281. xpath_variable_string(): xpath_variable(xpath_type_string), value(NULL)
  7282. {
  7283. }
  7284. ~xpath_variable_string()
  7285. {
  7286. if (value) xml_memory::deallocate(value);
  7287. }
  7288. char_t* value;
  7289. char_t name[1];
  7290. };
  7291. struct xpath_variable_node_set: xpath_variable
  7292. {
  7293. xpath_variable_node_set(): xpath_variable(xpath_type_node_set)
  7294. {
  7295. }
  7296. xpath_node_set value;
  7297. char_t name[1];
  7298. };
  7299. PUGI_IMPL_FN PUGI_IMPL_UNSIGNED_OVERFLOW unsigned int hash_string(const char_t* str)
  7300. {
  7301. // Jenkins one-at-a-time hash (http://en.wikipedia.org/wiki/Jenkins_hash_function#one-at-a-time)
  7302. unsigned int result = 0;
  7303. while (*str)
  7304. {
  7305. result += static_cast<unsigned int>(*str++);
  7306. result += result << 10;
  7307. result ^= result >> 6;
  7308. }
  7309. result += result << 3;
  7310. result ^= result >> 11;
  7311. result += result << 15;
  7312. return result;
  7313. }
  7314. template <typename T> PUGI_IMPL_FN T* new_xpath_variable(const char_t* name)
  7315. {
  7316. size_t length = strlength(name);
  7317. if (length == 0) return NULL; // empty variable names are invalid
  7318. // $$ we can't use offsetof(T, name) because T is non-POD, so we just allocate additional length characters
  7319. void* memory = xml_memory::allocate(sizeof(T) + length * sizeof(char_t));
  7320. if (!memory) return NULL;
  7321. T* result = new (memory) T();
  7322. memcpy(result->name, name, (length + 1) * sizeof(char_t));
  7323. return result;
  7324. }
  7325. PUGI_IMPL_FN xpath_variable* new_xpath_variable(xpath_value_type type, const char_t* name)
  7326. {
  7327. switch (type)
  7328. {
  7329. case xpath_type_node_set:
  7330. return new_xpath_variable<xpath_variable_node_set>(name);
  7331. case xpath_type_number:
  7332. return new_xpath_variable<xpath_variable_number>(name);
  7333. case xpath_type_string:
  7334. return new_xpath_variable<xpath_variable_string>(name);
  7335. case xpath_type_boolean:
  7336. return new_xpath_variable<xpath_variable_boolean>(name);
  7337. default:
  7338. return NULL;
  7339. }
  7340. }
  7341. template <typename T> PUGI_IMPL_FN void delete_xpath_variable(T* var)
  7342. {
  7343. var->~T();
  7344. xml_memory::deallocate(var);
  7345. }
  7346. PUGI_IMPL_FN void delete_xpath_variable(xpath_value_type type, xpath_variable* var)
  7347. {
  7348. switch (type)
  7349. {
  7350. case xpath_type_node_set:
  7351. delete_xpath_variable(static_cast<xpath_variable_node_set*>(var));
  7352. break;
  7353. case xpath_type_number:
  7354. delete_xpath_variable(static_cast<xpath_variable_number*>(var));
  7355. break;
  7356. case xpath_type_string:
  7357. delete_xpath_variable(static_cast<xpath_variable_string*>(var));
  7358. break;
  7359. case xpath_type_boolean:
  7360. delete_xpath_variable(static_cast<xpath_variable_boolean*>(var));
  7361. break;
  7362. default:
  7363. assert(false && "Invalid variable type"); // unreachable
  7364. }
  7365. }
  7366. PUGI_IMPL_FN bool copy_xpath_variable(xpath_variable* lhs, const xpath_variable* rhs)
  7367. {
  7368. switch (rhs->type())
  7369. {
  7370. case xpath_type_node_set:
  7371. return lhs->set(static_cast<const xpath_variable_node_set*>(rhs)->value);
  7372. case xpath_type_number:
  7373. return lhs->set(static_cast<const xpath_variable_number*>(rhs)->value);
  7374. case xpath_type_string:
  7375. return lhs->set(static_cast<const xpath_variable_string*>(rhs)->value);
  7376. case xpath_type_boolean:
  7377. return lhs->set(static_cast<const xpath_variable_boolean*>(rhs)->value);
  7378. default:
  7379. assert(false && "Invalid variable type"); // unreachable
  7380. return false;
  7381. }
  7382. }
  7383. PUGI_IMPL_FN bool get_variable_scratch(char_t (&buffer)[32], xpath_variable_set* set, const char_t* begin, const char_t* end, xpath_variable** out_result)
  7384. {
  7385. size_t length = static_cast<size_t>(end - begin);
  7386. char_t* scratch = buffer;
  7387. if (length >= sizeof(buffer) / sizeof(buffer[0]))
  7388. {
  7389. // need to make dummy on-heap copy
  7390. scratch = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
  7391. if (!scratch) return false;
  7392. }
  7393. // copy string to zero-terminated buffer and perform lookup
  7394. memcpy(scratch, begin, length * sizeof(char_t));
  7395. scratch[length] = 0;
  7396. *out_result = set->get(scratch);
  7397. // free dummy buffer
  7398. if (scratch != buffer) xml_memory::deallocate(scratch);
  7399. return true;
  7400. }
  7401. PUGI_IMPL_NS_END
  7402. // Internal node set class
  7403. PUGI_IMPL_NS_BEGIN
  7404. PUGI_IMPL_FN xpath_node_set::type_t xpath_get_order(const xpath_node* begin, const xpath_node* end)
  7405. {
  7406. if (end - begin < 2)
  7407. return xpath_node_set::type_sorted;
  7408. document_order_comparator cmp;
  7409. bool first = cmp(begin[0], begin[1]);
  7410. for (const xpath_node* it = begin + 1; it + 1 < end; ++it)
  7411. if (cmp(it[0], it[1]) != first)
  7412. return xpath_node_set::type_unsorted;
  7413. return first ? xpath_node_set::type_sorted : xpath_node_set::type_sorted_reverse;
  7414. }
  7415. PUGI_IMPL_FN xpath_node_set::type_t xpath_sort(xpath_node* begin, xpath_node* end, xpath_node_set::type_t type, bool rev)
  7416. {
  7417. xpath_node_set::type_t order = rev ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted;
  7418. if (type == xpath_node_set::type_unsorted)
  7419. {
  7420. xpath_node_set::type_t sorted = xpath_get_order(begin, end);
  7421. if (sorted == xpath_node_set::type_unsorted)
  7422. {
  7423. sort(begin, end, document_order_comparator());
  7424. type = xpath_node_set::type_sorted;
  7425. }
  7426. else
  7427. type = sorted;
  7428. }
  7429. if (type != order) reverse(begin, end);
  7430. return order;
  7431. }
  7432. PUGI_IMPL_FN xpath_node xpath_first(const xpath_node* begin, const xpath_node* end, xpath_node_set::type_t type)
  7433. {
  7434. if (begin == end) return xpath_node();
  7435. switch (type)
  7436. {
  7437. case xpath_node_set::type_sorted:
  7438. return *begin;
  7439. case xpath_node_set::type_sorted_reverse:
  7440. return *(end - 1);
  7441. case xpath_node_set::type_unsorted:
  7442. return *min_element(begin, end, document_order_comparator());
  7443. default:
  7444. assert(false && "Invalid node set type"); // unreachable
  7445. return xpath_node();
  7446. }
  7447. }
  7448. class xpath_node_set_raw
  7449. {
  7450. xpath_node_set::type_t _type;
  7451. xpath_node* _begin;
  7452. xpath_node* _end;
  7453. xpath_node* _eos;
  7454. public:
  7455. xpath_node_set_raw(): _type(xpath_node_set::type_unsorted), _begin(NULL), _end(NULL), _eos(NULL)
  7456. {
  7457. }
  7458. xpath_node* begin() const
  7459. {
  7460. return _begin;
  7461. }
  7462. xpath_node* end() const
  7463. {
  7464. return _end;
  7465. }
  7466. bool empty() const
  7467. {
  7468. return _begin == _end;
  7469. }
  7470. size_t size() const
  7471. {
  7472. return static_cast<size_t>(_end - _begin);
  7473. }
  7474. xpath_node first() const
  7475. {
  7476. return xpath_first(_begin, _end, _type);
  7477. }
  7478. void push_back_grow(const xpath_node& node, xpath_allocator* alloc);
  7479. void push_back(const xpath_node& node, xpath_allocator* alloc)
  7480. {
  7481. if (_end != _eos)
  7482. *_end++ = node;
  7483. else
  7484. push_back_grow(node, alloc);
  7485. }
  7486. void append(const xpath_node* begin_, const xpath_node* end_, xpath_allocator* alloc)
  7487. {
  7488. if (begin_ == end_) return;
  7489. size_t size_ = static_cast<size_t>(_end - _begin);
  7490. size_t capacity = static_cast<size_t>(_eos - _begin);
  7491. size_t count = static_cast<size_t>(end_ - begin_);
  7492. if (size_ + count > capacity)
  7493. {
  7494. // reallocate the old array or allocate a new one
  7495. xpath_node* data = static_cast<xpath_node*>(alloc->reallocate(_begin, capacity * sizeof(xpath_node), (size_ + count) * sizeof(xpath_node)));
  7496. if (!data) return;
  7497. // finalize
  7498. _begin = data;
  7499. _end = data + size_;
  7500. _eos = data + size_ + count;
  7501. }
  7502. memcpy(_end, begin_, count * sizeof(xpath_node));
  7503. _end += count;
  7504. }
  7505. void sort_do()
  7506. {
  7507. _type = xpath_sort(_begin, _end, _type, false);
  7508. }
  7509. void truncate(xpath_node* pos)
  7510. {
  7511. assert(_begin <= pos && pos <= _end);
  7512. _end = pos;
  7513. }
  7514. void remove_duplicates(xpath_allocator* alloc)
  7515. {
  7516. if (_type == xpath_node_set::type_unsorted && _end - _begin > 2)
  7517. {
  7518. xpath_allocator_capture cr(alloc);
  7519. size_t size_ = static_cast<size_t>(_end - _begin);
  7520. size_t hash_size = 1;
  7521. while (hash_size < size_ + size_ / 2) hash_size *= 2;
  7522. const void** hash_data = static_cast<const void**>(alloc->allocate(hash_size * sizeof(void*)));
  7523. if (!hash_data) return;
  7524. memset(hash_data, 0, hash_size * sizeof(void*));
  7525. xpath_node* write = _begin;
  7526. for (xpath_node* it = _begin; it != _end; ++it)
  7527. {
  7528. const void* attr = it->attribute().internal_object();
  7529. const void* node = it->node().internal_object();
  7530. const void* key = attr ? attr : node;
  7531. if (key && hash_insert(hash_data, hash_size, key))
  7532. {
  7533. *write++ = *it;
  7534. }
  7535. }
  7536. _end = write;
  7537. }
  7538. else
  7539. {
  7540. _end = unique(_begin, _end);
  7541. }
  7542. }
  7543. xpath_node_set::type_t type() const
  7544. {
  7545. return _type;
  7546. }
  7547. void set_type(xpath_node_set::type_t value)
  7548. {
  7549. _type = value;
  7550. }
  7551. };
  7552. PUGI_IMPL_FN_NO_INLINE void xpath_node_set_raw::push_back_grow(const xpath_node& node, xpath_allocator* alloc)
  7553. {
  7554. size_t capacity = static_cast<size_t>(_eos - _begin);
  7555. // get new capacity (1.5x rule)
  7556. size_t new_capacity = capacity + capacity / 2 + 1;
  7557. // reallocate the old array or allocate a new one
  7558. xpath_node* data = static_cast<xpath_node*>(alloc->reallocate(_begin, capacity * sizeof(xpath_node), new_capacity * sizeof(xpath_node)));
  7559. if (!data) return;
  7560. // finalize
  7561. _begin = data;
  7562. _end = data + capacity;
  7563. _eos = data + new_capacity;
  7564. // push
  7565. *_end++ = node;
  7566. }
  7567. PUGI_IMPL_NS_END
  7568. PUGI_IMPL_NS_BEGIN
  7569. struct xpath_context
  7570. {
  7571. xpath_node n;
  7572. size_t position, size;
  7573. xpath_context(const xpath_node& n_, size_t position_, size_t size_): n(n_), position(position_), size(size_)
  7574. {
  7575. }
  7576. };
  7577. enum lexeme_t
  7578. {
  7579. lex_none = 0,
  7580. lex_equal,
  7581. lex_not_equal,
  7582. lex_less,
  7583. lex_greater,
  7584. lex_less_or_equal,
  7585. lex_greater_or_equal,
  7586. lex_plus,
  7587. lex_minus,
  7588. lex_multiply,
  7589. lex_union,
  7590. lex_var_ref,
  7591. lex_open_brace,
  7592. lex_close_brace,
  7593. lex_quoted_string,
  7594. lex_number,
  7595. lex_slash,
  7596. lex_double_slash,
  7597. lex_open_square_brace,
  7598. lex_close_square_brace,
  7599. lex_string,
  7600. lex_comma,
  7601. lex_axis_attribute,
  7602. lex_dot,
  7603. lex_double_dot,
  7604. lex_double_colon,
  7605. lex_eof
  7606. };
  7607. struct xpath_lexer_string
  7608. {
  7609. const char_t* begin;
  7610. const char_t* end;
  7611. xpath_lexer_string(): begin(NULL), end(NULL)
  7612. {
  7613. }
  7614. bool operator==(const char_t* other) const
  7615. {
  7616. size_t length = static_cast<size_t>(end - begin);
  7617. return strequalrange(other, begin, length);
  7618. }
  7619. };
  7620. class xpath_lexer
  7621. {
  7622. const char_t* _cur;
  7623. const char_t* _cur_lexeme_pos;
  7624. xpath_lexer_string _cur_lexeme_contents;
  7625. lexeme_t _cur_lexeme;
  7626. public:
  7627. explicit xpath_lexer(const char_t* query): _cur(query)
  7628. {
  7629. next();
  7630. }
  7631. const char_t* state() const
  7632. {
  7633. return _cur;
  7634. }
  7635. void next()
  7636. {
  7637. const char_t* cur = _cur;
  7638. while (PUGI_IMPL_IS_CHARTYPE(*cur, ct_space)) ++cur;
  7639. // save lexeme position for error reporting
  7640. _cur_lexeme_pos = cur;
  7641. switch (*cur)
  7642. {
  7643. case 0:
  7644. _cur_lexeme = lex_eof;
  7645. break;
  7646. case '>':
  7647. if (*(cur+1) == '=')
  7648. {
  7649. cur += 2;
  7650. _cur_lexeme = lex_greater_or_equal;
  7651. }
  7652. else
  7653. {
  7654. cur += 1;
  7655. _cur_lexeme = lex_greater;
  7656. }
  7657. break;
  7658. case '<':
  7659. if (*(cur+1) == '=')
  7660. {
  7661. cur += 2;
  7662. _cur_lexeme = lex_less_or_equal;
  7663. }
  7664. else
  7665. {
  7666. cur += 1;
  7667. _cur_lexeme = lex_less;
  7668. }
  7669. break;
  7670. case '!':
  7671. if (*(cur+1) == '=')
  7672. {
  7673. cur += 2;
  7674. _cur_lexeme = lex_not_equal;
  7675. }
  7676. else
  7677. {
  7678. _cur_lexeme = lex_none;
  7679. }
  7680. break;
  7681. case '=':
  7682. cur += 1;
  7683. _cur_lexeme = lex_equal;
  7684. break;
  7685. case '+':
  7686. cur += 1;
  7687. _cur_lexeme = lex_plus;
  7688. break;
  7689. case '-':
  7690. cur += 1;
  7691. _cur_lexeme = lex_minus;
  7692. break;
  7693. case '*':
  7694. cur += 1;
  7695. _cur_lexeme = lex_multiply;
  7696. break;
  7697. case '|':
  7698. cur += 1;
  7699. _cur_lexeme = lex_union;
  7700. break;
  7701. case '$':
  7702. cur += 1;
  7703. if (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_start_symbol))
  7704. {
  7705. _cur_lexeme_contents.begin = cur;
  7706. while (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
  7707. if (cur[0] == ':' && PUGI_IMPL_IS_CHARTYPEX(cur[1], ctx_symbol)) // qname
  7708. {
  7709. cur++; // :
  7710. while (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
  7711. }
  7712. _cur_lexeme_contents.end = cur;
  7713. _cur_lexeme = lex_var_ref;
  7714. }
  7715. else
  7716. {
  7717. _cur_lexeme = lex_none;
  7718. }
  7719. break;
  7720. case '(':
  7721. cur += 1;
  7722. _cur_lexeme = lex_open_brace;
  7723. break;
  7724. case ')':
  7725. cur += 1;
  7726. _cur_lexeme = lex_close_brace;
  7727. break;
  7728. case '[':
  7729. cur += 1;
  7730. _cur_lexeme = lex_open_square_brace;
  7731. break;
  7732. case ']':
  7733. cur += 1;
  7734. _cur_lexeme = lex_close_square_brace;
  7735. break;
  7736. case ',':
  7737. cur += 1;
  7738. _cur_lexeme = lex_comma;
  7739. break;
  7740. case '/':
  7741. if (*(cur+1) == '/')
  7742. {
  7743. cur += 2;
  7744. _cur_lexeme = lex_double_slash;
  7745. }
  7746. else
  7747. {
  7748. cur += 1;
  7749. _cur_lexeme = lex_slash;
  7750. }
  7751. break;
  7752. case '.':
  7753. if (*(cur+1) == '.')
  7754. {
  7755. cur += 2;
  7756. _cur_lexeme = lex_double_dot;
  7757. }
  7758. else if (PUGI_IMPL_IS_CHARTYPEX(*(cur+1), ctx_digit))
  7759. {
  7760. _cur_lexeme_contents.begin = cur; // .
  7761. ++cur;
  7762. while (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_digit)) cur++;
  7763. _cur_lexeme_contents.end = cur;
  7764. _cur_lexeme = lex_number;
  7765. }
  7766. else
  7767. {
  7768. cur += 1;
  7769. _cur_lexeme = lex_dot;
  7770. }
  7771. break;
  7772. case '@':
  7773. cur += 1;
  7774. _cur_lexeme = lex_axis_attribute;
  7775. break;
  7776. case '"':
  7777. case '\'':
  7778. {
  7779. char_t terminator = *cur;
  7780. ++cur;
  7781. _cur_lexeme_contents.begin = cur;
  7782. while (*cur && *cur != terminator) cur++;
  7783. _cur_lexeme_contents.end = cur;
  7784. if (!*cur)
  7785. _cur_lexeme = lex_none;
  7786. else
  7787. {
  7788. cur += 1;
  7789. _cur_lexeme = lex_quoted_string;
  7790. }
  7791. break;
  7792. }
  7793. case ':':
  7794. if (*(cur+1) == ':')
  7795. {
  7796. cur += 2;
  7797. _cur_lexeme = lex_double_colon;
  7798. }
  7799. else
  7800. {
  7801. _cur_lexeme = lex_none;
  7802. }
  7803. break;
  7804. default:
  7805. if (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_digit))
  7806. {
  7807. _cur_lexeme_contents.begin = cur;
  7808. while (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_digit)) cur++;
  7809. if (*cur == '.')
  7810. {
  7811. cur++;
  7812. while (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_digit)) cur++;
  7813. }
  7814. _cur_lexeme_contents.end = cur;
  7815. _cur_lexeme = lex_number;
  7816. }
  7817. else if (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_start_symbol))
  7818. {
  7819. _cur_lexeme_contents.begin = cur;
  7820. while (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
  7821. if (cur[0] == ':')
  7822. {
  7823. if (cur[1] == '*') // namespace test ncname:*
  7824. {
  7825. cur += 2; // :*
  7826. }
  7827. else if (PUGI_IMPL_IS_CHARTYPEX(cur[1], ctx_symbol)) // namespace test qname
  7828. {
  7829. cur++; // :
  7830. while (PUGI_IMPL_IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
  7831. }
  7832. }
  7833. _cur_lexeme_contents.end = cur;
  7834. _cur_lexeme = lex_string;
  7835. }
  7836. else
  7837. {
  7838. _cur_lexeme = lex_none;
  7839. }
  7840. }
  7841. _cur = cur;
  7842. }
  7843. lexeme_t current() const
  7844. {
  7845. return _cur_lexeme;
  7846. }
  7847. const char_t* current_pos() const
  7848. {
  7849. return _cur_lexeme_pos;
  7850. }
  7851. const xpath_lexer_string& contents() const
  7852. {
  7853. assert(_cur_lexeme == lex_var_ref || _cur_lexeme == lex_number || _cur_lexeme == lex_string || _cur_lexeme == lex_quoted_string);
  7854. return _cur_lexeme_contents;
  7855. }
  7856. };
  7857. enum ast_type_t
  7858. {
  7859. ast_unknown,
  7860. ast_op_or, // left or right
  7861. ast_op_and, // left and right
  7862. ast_op_equal, // left = right
  7863. ast_op_not_equal, // left != right
  7864. ast_op_less, // left < right
  7865. ast_op_greater, // left > right
  7866. ast_op_less_or_equal, // left <= right
  7867. ast_op_greater_or_equal, // left >= right
  7868. ast_op_add, // left + right
  7869. ast_op_subtract, // left - right
  7870. ast_op_multiply, // left * right
  7871. ast_op_divide, // left / right
  7872. ast_op_mod, // left % right
  7873. ast_op_negate, // left - right
  7874. ast_op_union, // left | right
  7875. ast_predicate, // apply predicate to set; next points to next predicate
  7876. ast_filter, // select * from left where right
  7877. ast_string_constant, // string constant
  7878. ast_number_constant, // number constant
  7879. ast_variable, // variable
  7880. ast_func_last, // last()
  7881. ast_func_position, // position()
  7882. ast_func_count, // count(left)
  7883. ast_func_id, // id(left)
  7884. ast_func_local_name_0, // local-name()
  7885. ast_func_local_name_1, // local-name(left)
  7886. ast_func_namespace_uri_0, // namespace-uri()
  7887. ast_func_namespace_uri_1, // namespace-uri(left)
  7888. ast_func_name_0, // name()
  7889. ast_func_name_1, // name(left)
  7890. ast_func_string_0, // string()
  7891. ast_func_string_1, // string(left)
  7892. ast_func_concat, // concat(left, right, siblings)
  7893. ast_func_starts_with, // starts_with(left, right)
  7894. ast_func_contains, // contains(left, right)
  7895. ast_func_substring_before, // substring-before(left, right)
  7896. ast_func_substring_after, // substring-after(left, right)
  7897. ast_func_substring_2, // substring(left, right)
  7898. ast_func_substring_3, // substring(left, right, third)
  7899. ast_func_string_length_0, // string-length()
  7900. ast_func_string_length_1, // string-length(left)
  7901. ast_func_normalize_space_0, // normalize-space()
  7902. ast_func_normalize_space_1, // normalize-space(left)
  7903. ast_func_translate, // translate(left, right, third)
  7904. ast_func_boolean, // boolean(left)
  7905. ast_func_not, // not(left)
  7906. ast_func_true, // true()
  7907. ast_func_false, // false()
  7908. ast_func_lang, // lang(left)
  7909. ast_func_number_0, // number()
  7910. ast_func_number_1, // number(left)
  7911. ast_func_sum, // sum(left)
  7912. ast_func_floor, // floor(left)
  7913. ast_func_ceiling, // ceiling(left)
  7914. ast_func_round, // round(left)
  7915. ast_step, // process set left with step
  7916. ast_step_root, // select root node
  7917. ast_opt_translate_table, // translate(left, right, third) where right/third are constants
  7918. ast_opt_compare_attribute // @name = 'string'
  7919. };
  7920. enum axis_t
  7921. {
  7922. axis_ancestor,
  7923. axis_ancestor_or_self,
  7924. axis_attribute,
  7925. axis_child,
  7926. axis_descendant,
  7927. axis_descendant_or_self,
  7928. axis_following,
  7929. axis_following_sibling,
  7930. axis_namespace,
  7931. axis_parent,
  7932. axis_preceding,
  7933. axis_preceding_sibling,
  7934. axis_self
  7935. };
  7936. enum nodetest_t
  7937. {
  7938. nodetest_none,
  7939. nodetest_name,
  7940. nodetest_type_node,
  7941. nodetest_type_comment,
  7942. nodetest_type_pi,
  7943. nodetest_type_text,
  7944. nodetest_pi,
  7945. nodetest_all,
  7946. nodetest_all_in_namespace
  7947. };
  7948. enum predicate_t
  7949. {
  7950. predicate_default,
  7951. predicate_posinv,
  7952. predicate_constant,
  7953. predicate_constant_one
  7954. };
  7955. enum nodeset_eval_t
  7956. {
  7957. nodeset_eval_all,
  7958. nodeset_eval_any,
  7959. nodeset_eval_first
  7960. };
  7961. template <axis_t N> struct axis_to_type
  7962. {
  7963. static const axis_t axis;
  7964. };
  7965. template <axis_t N> const axis_t axis_to_type<N>::axis = N;
  7966. class xpath_ast_node
  7967. {
  7968. private:
  7969. // node type
  7970. char _type;
  7971. char _rettype;
  7972. // for ast_step
  7973. char _axis;
  7974. // for ast_step/ast_predicate/ast_filter
  7975. char _test;
  7976. // tree node structure
  7977. xpath_ast_node* _left;
  7978. xpath_ast_node* _right;
  7979. xpath_ast_node* _next;
  7980. union
  7981. {
  7982. // value for ast_string_constant
  7983. const char_t* string;
  7984. // value for ast_number_constant
  7985. double number;
  7986. // variable for ast_variable
  7987. xpath_variable* variable;
  7988. // node test for ast_step (node name/namespace/node type/pi target)
  7989. const char_t* nodetest;
  7990. // table for ast_opt_translate_table
  7991. const unsigned char* table;
  7992. } _data;
  7993. xpath_ast_node(const xpath_ast_node&);
  7994. xpath_ast_node& operator=(const xpath_ast_node&);
  7995. template <class Comp> static bool compare_eq(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp)
  7996. {
  7997. xpath_value_type lt = lhs->rettype(), rt = rhs->rettype();
  7998. if (lt != xpath_type_node_set && rt != xpath_type_node_set)
  7999. {
  8000. if (lt == xpath_type_boolean || rt == xpath_type_boolean)
  8001. return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack));
  8002. else if (lt == xpath_type_number || rt == xpath_type_number)
  8003. return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack));
  8004. else if (lt == xpath_type_string || rt == xpath_type_string)
  8005. {
  8006. xpath_allocator_capture cr(stack.result);
  8007. xpath_string ls = lhs->eval_string(c, stack);
  8008. xpath_string rs = rhs->eval_string(c, stack);
  8009. return comp(ls, rs);
  8010. }
  8011. }
  8012. else if (lt == xpath_type_node_set && rt == xpath_type_node_set)
  8013. {
  8014. xpath_allocator_capture cr(stack.result);
  8015. xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all);
  8016. xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
  8017. for (const xpath_node* li = ls.begin(); li != ls.end(); ++li)
  8018. for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
  8019. {
  8020. xpath_allocator_capture cri(stack.result);
  8021. if (comp(string_value(*li, stack.result), string_value(*ri, stack.result)))
  8022. return true;
  8023. }
  8024. return false;
  8025. }
  8026. else
  8027. {
  8028. if (lt == xpath_type_node_set)
  8029. {
  8030. swap(lhs, rhs);
  8031. swap(lt, rt);
  8032. }
  8033. if (lt == xpath_type_boolean)
  8034. return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack));
  8035. else if (lt == xpath_type_number)
  8036. {
  8037. xpath_allocator_capture cr(stack.result);
  8038. double l = lhs->eval_number(c, stack);
  8039. xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
  8040. for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
  8041. {
  8042. xpath_allocator_capture cri(stack.result);
  8043. if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str())))
  8044. return true;
  8045. }
  8046. return false;
  8047. }
  8048. else if (lt == xpath_type_string)
  8049. {
  8050. xpath_allocator_capture cr(stack.result);
  8051. xpath_string l = lhs->eval_string(c, stack);
  8052. xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
  8053. for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
  8054. {
  8055. xpath_allocator_capture cri(stack.result);
  8056. if (comp(l, string_value(*ri, stack.result)))
  8057. return true;
  8058. }
  8059. return false;
  8060. }
  8061. }
  8062. assert(false && "Wrong types"); // unreachable
  8063. return false;
  8064. }
  8065. static bool eval_once(xpath_node_set::type_t type, nodeset_eval_t eval)
  8066. {
  8067. return type == xpath_node_set::type_sorted ? eval != nodeset_eval_all : eval == nodeset_eval_any;
  8068. }
  8069. template <class Comp> static bool compare_rel(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp)
  8070. {
  8071. xpath_value_type lt = lhs->rettype(), rt = rhs->rettype();
  8072. if (lt != xpath_type_node_set && rt != xpath_type_node_set)
  8073. return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack));
  8074. else if (lt == xpath_type_node_set && rt == xpath_type_node_set)
  8075. {
  8076. xpath_allocator_capture cr(stack.result);
  8077. xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all);
  8078. xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
  8079. for (const xpath_node* li = ls.begin(); li != ls.end(); ++li)
  8080. {
  8081. xpath_allocator_capture cri(stack.result);
  8082. double l = convert_string_to_number(string_value(*li, stack.result).c_str());
  8083. for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
  8084. {
  8085. xpath_allocator_capture crii(stack.result);
  8086. if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str())))
  8087. return true;
  8088. }
  8089. }
  8090. return false;
  8091. }
  8092. else if (lt != xpath_type_node_set && rt == xpath_type_node_set)
  8093. {
  8094. xpath_allocator_capture cr(stack.result);
  8095. double l = lhs->eval_number(c, stack);
  8096. xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
  8097. for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
  8098. {
  8099. xpath_allocator_capture cri(stack.result);
  8100. if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str())))
  8101. return true;
  8102. }
  8103. return false;
  8104. }
  8105. else if (lt == xpath_type_node_set && rt != xpath_type_node_set)
  8106. {
  8107. xpath_allocator_capture cr(stack.result);
  8108. xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all);
  8109. double r = rhs->eval_number(c, stack);
  8110. for (const xpath_node* li = ls.begin(); li != ls.end(); ++li)
  8111. {
  8112. xpath_allocator_capture cri(stack.result);
  8113. if (comp(convert_string_to_number(string_value(*li, stack.result).c_str()), r))
  8114. return true;
  8115. }
  8116. return false;
  8117. }
  8118. else
  8119. {
  8120. assert(false && "Wrong types"); // unreachable
  8121. return false;
  8122. }
  8123. }
  8124. static void apply_predicate_boolean(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack, bool once)
  8125. {
  8126. assert(ns.size() >= first);
  8127. assert(expr->rettype() != xpath_type_number);
  8128. size_t i = 1;
  8129. size_t size = ns.size() - first;
  8130. xpath_node* last = ns.begin() + first;
  8131. // remove_if... or well, sort of
  8132. for (xpath_node* it = last; it != ns.end(); ++it, ++i)
  8133. {
  8134. xpath_context c(*it, i, size);
  8135. if (expr->eval_boolean(c, stack))
  8136. {
  8137. *last++ = *it;
  8138. if (once) break;
  8139. }
  8140. }
  8141. ns.truncate(last);
  8142. }
  8143. static void apply_predicate_number(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack, bool once)
  8144. {
  8145. assert(ns.size() >= first);
  8146. assert(expr->rettype() == xpath_type_number);
  8147. size_t i = 1;
  8148. size_t size = ns.size() - first;
  8149. xpath_node* last = ns.begin() + first;
  8150. // remove_if... or well, sort of
  8151. for (xpath_node* it = last; it != ns.end(); ++it, ++i)
  8152. {
  8153. xpath_context c(*it, i, size);
  8154. if (expr->eval_number(c, stack) == static_cast<double>(i))
  8155. {
  8156. *last++ = *it;
  8157. if (once) break;
  8158. }
  8159. }
  8160. ns.truncate(last);
  8161. }
  8162. static void apply_predicate_number_const(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack)
  8163. {
  8164. assert(ns.size() >= first);
  8165. assert(expr->rettype() == xpath_type_number);
  8166. size_t size = ns.size() - first;
  8167. xpath_node* last = ns.begin() + first;
  8168. xpath_node cn;
  8169. xpath_context c(cn, 1, size);
  8170. double er = expr->eval_number(c, stack);
  8171. if (er >= 1.0 && er <= static_cast<double>(size))
  8172. {
  8173. size_t eri = static_cast<size_t>(er);
  8174. if (er == static_cast<double>(eri))
  8175. {
  8176. xpath_node r = last[eri - 1];
  8177. *last++ = r;
  8178. }
  8179. }
  8180. ns.truncate(last);
  8181. }
  8182. void apply_predicate(xpath_node_set_raw& ns, size_t first, const xpath_stack& stack, bool once)
  8183. {
  8184. if (ns.size() == first) return;
  8185. assert(_type == ast_filter || _type == ast_predicate);
  8186. if (_test == predicate_constant || _test == predicate_constant_one)
  8187. apply_predicate_number_const(ns, first, _right, stack);
  8188. else if (_right->rettype() == xpath_type_number)
  8189. apply_predicate_number(ns, first, _right, stack, once);
  8190. else
  8191. apply_predicate_boolean(ns, first, _right, stack, once);
  8192. }
  8193. void apply_predicates(xpath_node_set_raw& ns, size_t first, const xpath_stack& stack, nodeset_eval_t eval)
  8194. {
  8195. if (ns.size() == first) return;
  8196. bool last_once = eval_once(ns.type(), eval);
  8197. for (xpath_ast_node* pred = _right; pred; pred = pred->_next)
  8198. pred->apply_predicate(ns, first, stack, !pred->_next && last_once);
  8199. }
  8200. bool step_push(xpath_node_set_raw& ns, xml_attribute_struct* a, xml_node_struct* parent, xpath_allocator* alloc)
  8201. {
  8202. assert(a);
  8203. const char_t* name = a->name ? a->name + 0 : PUGIXML_TEXT("");
  8204. switch (_test)
  8205. {
  8206. case nodetest_name:
  8207. if (strequal(name, _data.nodetest) && is_xpath_attribute(name))
  8208. {
  8209. ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc);
  8210. return true;
  8211. }
  8212. break;
  8213. case nodetest_type_node:
  8214. case nodetest_all:
  8215. if (is_xpath_attribute(name))
  8216. {
  8217. ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc);
  8218. return true;
  8219. }
  8220. break;
  8221. case nodetest_all_in_namespace:
  8222. if (starts_with(name, _data.nodetest) && is_xpath_attribute(name))
  8223. {
  8224. ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc);
  8225. return true;
  8226. }
  8227. break;
  8228. default:
  8229. ;
  8230. }
  8231. return false;
  8232. }
  8233. bool step_push(xpath_node_set_raw& ns, xml_node_struct* n, xpath_allocator* alloc)
  8234. {
  8235. assert(n);
  8236. xml_node_type type = PUGI_IMPL_NODETYPE(n);
  8237. switch (_test)
  8238. {
  8239. case nodetest_name:
  8240. if (type == node_element && n->name && strequal(n->name, _data.nodetest))
  8241. {
  8242. ns.push_back(xml_node(n), alloc);
  8243. return true;
  8244. }
  8245. break;
  8246. case nodetest_type_node:
  8247. ns.push_back(xml_node(n), alloc);
  8248. return true;
  8249. case nodetest_type_comment:
  8250. if (type == node_comment)
  8251. {
  8252. ns.push_back(xml_node(n), alloc);
  8253. return true;
  8254. }
  8255. break;
  8256. case nodetest_type_text:
  8257. if (type == node_pcdata || type == node_cdata)
  8258. {
  8259. ns.push_back(xml_node(n), alloc);
  8260. return true;
  8261. }
  8262. break;
  8263. case nodetest_type_pi:
  8264. if (type == node_pi)
  8265. {
  8266. ns.push_back(xml_node(n), alloc);
  8267. return true;
  8268. }
  8269. break;
  8270. case nodetest_pi:
  8271. if (type == node_pi && n->name && strequal(n->name, _data.nodetest))
  8272. {
  8273. ns.push_back(xml_node(n), alloc);
  8274. return true;
  8275. }
  8276. break;
  8277. case nodetest_all:
  8278. if (type == node_element)
  8279. {
  8280. ns.push_back(xml_node(n), alloc);
  8281. return true;
  8282. }
  8283. break;
  8284. case nodetest_all_in_namespace:
  8285. if (type == node_element && n->name && starts_with(n->name, _data.nodetest))
  8286. {
  8287. ns.push_back(xml_node(n), alloc);
  8288. return true;
  8289. }
  8290. break;
  8291. default:
  8292. assert(false && "Unknown axis"); // unreachable
  8293. }
  8294. return false;
  8295. }
  8296. template <class T> void step_fill(xpath_node_set_raw& ns, xml_node_struct* n, xpath_allocator* alloc, bool once, T)
  8297. {
  8298. const axis_t axis = T::axis;
  8299. switch (axis)
  8300. {
  8301. case axis_attribute:
  8302. {
  8303. for (xml_attribute_struct* a = n->first_attribute; a; a = a->next_attribute)
  8304. if (step_push(ns, a, n, alloc) & once)
  8305. return;
  8306. break;
  8307. }
  8308. case axis_child:
  8309. {
  8310. for (xml_node_struct* c = n->first_child; c; c = c->next_sibling)
  8311. if (step_push(ns, c, alloc) & once)
  8312. return;
  8313. break;
  8314. }
  8315. case axis_descendant:
  8316. case axis_descendant_or_self:
  8317. {
  8318. if (axis == axis_descendant_or_self)
  8319. if (step_push(ns, n, alloc) & once)
  8320. return;
  8321. xml_node_struct* cur = n->first_child;
  8322. while (cur)
  8323. {
  8324. if (step_push(ns, cur, alloc) & once)
  8325. return;
  8326. if (cur->first_child)
  8327. cur = cur->first_child;
  8328. else
  8329. {
  8330. while (!cur->next_sibling)
  8331. {
  8332. cur = cur->parent;
  8333. if (cur == n) return;
  8334. }
  8335. cur = cur->next_sibling;
  8336. }
  8337. }
  8338. break;
  8339. }
  8340. case axis_following_sibling:
  8341. {
  8342. for (xml_node_struct* c = n->next_sibling; c; c = c->next_sibling)
  8343. if (step_push(ns, c, alloc) & once)
  8344. return;
  8345. break;
  8346. }
  8347. case axis_preceding_sibling:
  8348. {
  8349. for (xml_node_struct* c = n->prev_sibling_c; c->next_sibling; c = c->prev_sibling_c)
  8350. if (step_push(ns, c, alloc) & once)
  8351. return;
  8352. break;
  8353. }
  8354. case axis_following:
  8355. {
  8356. xml_node_struct* cur = n;
  8357. // exit from this node so that we don't include descendants
  8358. while (!cur->next_sibling)
  8359. {
  8360. cur = cur->parent;
  8361. if (!cur) return;
  8362. }
  8363. cur = cur->next_sibling;
  8364. while (cur)
  8365. {
  8366. if (step_push(ns, cur, alloc) & once)
  8367. return;
  8368. if (cur->first_child)
  8369. cur = cur->first_child;
  8370. else
  8371. {
  8372. while (!cur->next_sibling)
  8373. {
  8374. cur = cur->parent;
  8375. if (!cur) return;
  8376. }
  8377. cur = cur->next_sibling;
  8378. }
  8379. }
  8380. break;
  8381. }
  8382. case axis_preceding:
  8383. {
  8384. xml_node_struct* cur = n;
  8385. // exit from this node so that we don't include descendants
  8386. while (!cur->prev_sibling_c->next_sibling)
  8387. {
  8388. cur = cur->parent;
  8389. if (!cur) return;
  8390. }
  8391. cur = cur->prev_sibling_c;
  8392. while (cur)
  8393. {
  8394. if (cur->first_child)
  8395. cur = cur->first_child->prev_sibling_c;
  8396. else
  8397. {
  8398. // leaf node, can't be ancestor
  8399. if (step_push(ns, cur, alloc) & once)
  8400. return;
  8401. while (!cur->prev_sibling_c->next_sibling)
  8402. {
  8403. cur = cur->parent;
  8404. if (!cur) return;
  8405. if (!node_is_ancestor(cur, n))
  8406. if (step_push(ns, cur, alloc) & once)
  8407. return;
  8408. }
  8409. cur = cur->prev_sibling_c;
  8410. }
  8411. }
  8412. break;
  8413. }
  8414. case axis_ancestor:
  8415. case axis_ancestor_or_self:
  8416. {
  8417. if (axis == axis_ancestor_or_self)
  8418. if (step_push(ns, n, alloc) & once)
  8419. return;
  8420. xml_node_struct* cur = n->parent;
  8421. while (cur)
  8422. {
  8423. if (step_push(ns, cur, alloc) & once)
  8424. return;
  8425. cur = cur->parent;
  8426. }
  8427. break;
  8428. }
  8429. case axis_self:
  8430. {
  8431. step_push(ns, n, alloc);
  8432. break;
  8433. }
  8434. case axis_parent:
  8435. {
  8436. if (n->parent)
  8437. step_push(ns, n->parent, alloc);
  8438. break;
  8439. }
  8440. default:
  8441. assert(false && "Unimplemented axis"); // unreachable
  8442. }
  8443. }
  8444. template <class T> void step_fill(xpath_node_set_raw& ns, xml_attribute_struct* a, xml_node_struct* p, xpath_allocator* alloc, bool once, T v)
  8445. {
  8446. const axis_t axis = T::axis;
  8447. switch (axis)
  8448. {
  8449. case axis_ancestor:
  8450. case axis_ancestor_or_self:
  8451. {
  8452. if (axis == axis_ancestor_or_self && _test == nodetest_type_node) // reject attributes based on principal node type test
  8453. if (step_push(ns, a, p, alloc) & once)
  8454. return;
  8455. xml_node_struct* cur = p;
  8456. while (cur)
  8457. {
  8458. if (step_push(ns, cur, alloc) & once)
  8459. return;
  8460. cur = cur->parent;
  8461. }
  8462. break;
  8463. }
  8464. case axis_descendant_or_self:
  8465. case axis_self:
  8466. {
  8467. if (_test == nodetest_type_node) // reject attributes based on principal node type test
  8468. step_push(ns, a, p, alloc);
  8469. break;
  8470. }
  8471. case axis_following:
  8472. {
  8473. xml_node_struct* cur = p;
  8474. while (cur)
  8475. {
  8476. if (cur->first_child)
  8477. cur = cur->first_child;
  8478. else
  8479. {
  8480. while (!cur->next_sibling)
  8481. {
  8482. cur = cur->parent;
  8483. if (!cur) return;
  8484. }
  8485. cur = cur->next_sibling;
  8486. }
  8487. if (step_push(ns, cur, alloc) & once)
  8488. return;
  8489. }
  8490. break;
  8491. }
  8492. case axis_parent:
  8493. {
  8494. step_push(ns, p, alloc);
  8495. break;
  8496. }
  8497. case axis_preceding:
  8498. {
  8499. // preceding:: axis does not include attribute nodes and attribute ancestors (they are the same as parent's ancestors), so we can reuse node preceding
  8500. step_fill(ns, p, alloc, once, v);
  8501. break;
  8502. }
  8503. default:
  8504. assert(false && "Unimplemented axis"); // unreachable
  8505. }
  8506. }
  8507. template <class T> void step_fill(xpath_node_set_raw& ns, const xpath_node& xn, xpath_allocator* alloc, bool once, T v)
  8508. {
  8509. const axis_t axis = T::axis;
  8510. const bool axis_has_attributes = (axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_descendant_or_self || axis == axis_following || axis == axis_parent || axis == axis_preceding || axis == axis_self);
  8511. if (xn.node())
  8512. step_fill(ns, xn.node().internal_object(), alloc, once, v);
  8513. else if (axis_has_attributes && xn.attribute() && xn.parent())
  8514. step_fill(ns, xn.attribute().internal_object(), xn.parent().internal_object(), alloc, once, v);
  8515. }
  8516. template <class T> xpath_node_set_raw step_do(const xpath_context& c, const xpath_stack& stack, nodeset_eval_t eval, T v)
  8517. {
  8518. const axis_t axis = T::axis;
  8519. const bool axis_reverse = (axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_preceding || axis == axis_preceding_sibling);
  8520. const xpath_node_set::type_t axis_type = axis_reverse ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted;
  8521. bool once =
  8522. (axis == axis_attribute && _test == nodetest_name) ||
  8523. (!_right && eval_once(axis_type, eval)) ||
  8524. // coverity[mixed_enums]
  8525. (_right && !_right->_next && _right->_test == predicate_constant_one);
  8526. xpath_node_set_raw ns;
  8527. ns.set_type(axis_type);
  8528. if (_left)
  8529. {
  8530. xpath_node_set_raw s = _left->eval_node_set(c, stack, nodeset_eval_all);
  8531. // self axis preserves the original order
  8532. if (axis == axis_self) ns.set_type(s.type());
  8533. for (const xpath_node* it = s.begin(); it != s.end(); ++it)
  8534. {
  8535. size_t size = ns.size();
  8536. // in general, all axes generate elements in a particular order, but there is no order guarantee if axis is applied to two nodes
  8537. if (axis != axis_self && size != 0) ns.set_type(xpath_node_set::type_unsorted);
  8538. step_fill(ns, *it, stack.result, once, v);
  8539. if (_right) apply_predicates(ns, size, stack, eval);
  8540. }
  8541. }
  8542. else
  8543. {
  8544. step_fill(ns, c.n, stack.result, once, v);
  8545. if (_right) apply_predicates(ns, 0, stack, eval);
  8546. }
  8547. // child, attribute and self axes always generate unique set of nodes
  8548. // for other axis, if the set stayed sorted, it stayed unique because the traversal algorithms do not visit the same node twice
  8549. if (axis != axis_child && axis != axis_attribute && axis != axis_self && ns.type() == xpath_node_set::type_unsorted)
  8550. ns.remove_duplicates(stack.temp);
  8551. return ns;
  8552. }
  8553. public:
  8554. xpath_ast_node(ast_type_t type, xpath_value_type rettype_, const char_t* value):
  8555. _type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(NULL), _right(NULL), _next(NULL)
  8556. {
  8557. assert(type == ast_string_constant);
  8558. _data.string = value;
  8559. }
  8560. xpath_ast_node(ast_type_t type, xpath_value_type rettype_, double value):
  8561. _type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(NULL), _right(NULL), _next(NULL)
  8562. {
  8563. assert(type == ast_number_constant);
  8564. _data.number = value;
  8565. }
  8566. xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_variable* value):
  8567. _type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(NULL), _right(NULL), _next(NULL)
  8568. {
  8569. assert(type == ast_variable);
  8570. _data.variable = value;
  8571. }
  8572. xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_ast_node* left = NULL, xpath_ast_node* right = NULL):
  8573. _type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(left), _right(right), _next(NULL)
  8574. {
  8575. }
  8576. xpath_ast_node(ast_type_t type, xpath_ast_node* left, axis_t axis, nodetest_t test, const char_t* contents):
  8577. _type(static_cast<char>(type)), _rettype(xpath_type_node_set), _axis(static_cast<char>(axis)), _test(static_cast<char>(test)), _left(left), _right(NULL), _next(NULL)
  8578. {
  8579. assert(type == ast_step);
  8580. _data.nodetest = contents;
  8581. }
  8582. xpath_ast_node(ast_type_t type, xpath_ast_node* left, xpath_ast_node* right, predicate_t test):
  8583. _type(static_cast<char>(type)), _rettype(xpath_type_node_set), _axis(0), _test(static_cast<char>(test)), _left(left), _right(right), _next(NULL)
  8584. {
  8585. assert(type == ast_filter || type == ast_predicate);
  8586. }
  8587. void set_next(xpath_ast_node* value)
  8588. {
  8589. _next = value;
  8590. }
  8591. void set_right(xpath_ast_node* value)
  8592. {
  8593. _right = value;
  8594. }
  8595. bool eval_boolean(const xpath_context& c, const xpath_stack& stack)
  8596. {
  8597. switch (_type)
  8598. {
  8599. case ast_op_or:
  8600. return _left->eval_boolean(c, stack) || _right->eval_boolean(c, stack);
  8601. case ast_op_and:
  8602. return _left->eval_boolean(c, stack) && _right->eval_boolean(c, stack);
  8603. case ast_op_equal:
  8604. return compare_eq(_left, _right, c, stack, equal_to());
  8605. case ast_op_not_equal:
  8606. return compare_eq(_left, _right, c, stack, not_equal_to());
  8607. case ast_op_less:
  8608. return compare_rel(_left, _right, c, stack, less());
  8609. case ast_op_greater:
  8610. return compare_rel(_right, _left, c, stack, less());
  8611. case ast_op_less_or_equal:
  8612. return compare_rel(_left, _right, c, stack, less_equal());
  8613. case ast_op_greater_or_equal:
  8614. return compare_rel(_right, _left, c, stack, less_equal());
  8615. case ast_func_starts_with:
  8616. {
  8617. xpath_allocator_capture cr(stack.result);
  8618. xpath_string lr = _left->eval_string(c, stack);
  8619. xpath_string rr = _right->eval_string(c, stack);
  8620. return starts_with(lr.c_str(), rr.c_str());
  8621. }
  8622. case ast_func_contains:
  8623. {
  8624. xpath_allocator_capture cr(stack.result);
  8625. xpath_string lr = _left->eval_string(c, stack);
  8626. xpath_string rr = _right->eval_string(c, stack);
  8627. return find_substring(lr.c_str(), rr.c_str()) != NULL;
  8628. }
  8629. case ast_func_boolean:
  8630. return _left->eval_boolean(c, stack);
  8631. case ast_func_not:
  8632. return !_left->eval_boolean(c, stack);
  8633. case ast_func_true:
  8634. return true;
  8635. case ast_func_false:
  8636. return false;
  8637. case ast_func_lang:
  8638. {
  8639. if (c.n.attribute()) return false;
  8640. xpath_allocator_capture cr(stack.result);
  8641. xpath_string lang = _left->eval_string(c, stack);
  8642. for (xml_node n = c.n.node(); n; n = n.parent())
  8643. {
  8644. xml_attribute a = n.attribute(PUGIXML_TEXT("xml:lang"));
  8645. if (a)
  8646. {
  8647. const char_t* value = a.value();
  8648. // strnicmp / strncasecmp is not portable
  8649. for (const char_t* lit = lang.c_str(); *lit; ++lit)
  8650. {
  8651. if (tolower_ascii(*lit) != tolower_ascii(*value)) return false;
  8652. ++value;
  8653. }
  8654. return *value == 0 || *value == '-';
  8655. }
  8656. }
  8657. return false;
  8658. }
  8659. case ast_opt_compare_attribute:
  8660. {
  8661. const char_t* value = (_right->_type == ast_string_constant) ? _right->_data.string : _right->_data.variable->get_string();
  8662. xml_attribute attr = c.n.node().attribute(_left->_data.nodetest);
  8663. return attr && strequal(attr.value(), value) && is_xpath_attribute(attr.name());
  8664. }
  8665. case ast_variable:
  8666. {
  8667. assert(_rettype == _data.variable->type());
  8668. if (_rettype == xpath_type_boolean)
  8669. return _data.variable->get_boolean();
  8670. // variable needs to be converted to the correct type, this is handled by the fallthrough block below
  8671. break;
  8672. }
  8673. default:
  8674. ;
  8675. }
  8676. // none of the ast types that return the value directly matched, we need to perform type conversion
  8677. switch (_rettype)
  8678. {
  8679. case xpath_type_number:
  8680. return convert_number_to_boolean(eval_number(c, stack));
  8681. case xpath_type_string:
  8682. {
  8683. xpath_allocator_capture cr(stack.result);
  8684. return !eval_string(c, stack).empty();
  8685. }
  8686. case xpath_type_node_set:
  8687. {
  8688. xpath_allocator_capture cr(stack.result);
  8689. return !eval_node_set(c, stack, nodeset_eval_any).empty();
  8690. }
  8691. default:
  8692. assert(false && "Wrong expression for return type boolean"); // unreachable
  8693. return false;
  8694. }
  8695. }
  8696. double eval_number(const xpath_context& c, const xpath_stack& stack)
  8697. {
  8698. switch (_type)
  8699. {
  8700. case ast_op_add:
  8701. return _left->eval_number(c, stack) + _right->eval_number(c, stack);
  8702. case ast_op_subtract:
  8703. return _left->eval_number(c, stack) - _right->eval_number(c, stack);
  8704. case ast_op_multiply:
  8705. return _left->eval_number(c, stack) * _right->eval_number(c, stack);
  8706. case ast_op_divide:
  8707. return _left->eval_number(c, stack) / _right->eval_number(c, stack);
  8708. case ast_op_mod:
  8709. return fmod(_left->eval_number(c, stack), _right->eval_number(c, stack));
  8710. case ast_op_negate:
  8711. return -_left->eval_number(c, stack);
  8712. case ast_number_constant:
  8713. return _data.number;
  8714. case ast_func_last:
  8715. return static_cast<double>(c.size);
  8716. case ast_func_position:
  8717. return static_cast<double>(c.position);
  8718. case ast_func_count:
  8719. {
  8720. xpath_allocator_capture cr(stack.result);
  8721. return static_cast<double>(_left->eval_node_set(c, stack, nodeset_eval_all).size());
  8722. }
  8723. case ast_func_string_length_0:
  8724. {
  8725. xpath_allocator_capture cr(stack.result);
  8726. return static_cast<double>(string_value(c.n, stack.result).length());
  8727. }
  8728. case ast_func_string_length_1:
  8729. {
  8730. xpath_allocator_capture cr(stack.result);
  8731. return static_cast<double>(_left->eval_string(c, stack).length());
  8732. }
  8733. case ast_func_number_0:
  8734. {
  8735. xpath_allocator_capture cr(stack.result);
  8736. return convert_string_to_number(string_value(c.n, stack.result).c_str());
  8737. }
  8738. case ast_func_number_1:
  8739. return _left->eval_number(c, stack);
  8740. case ast_func_sum:
  8741. {
  8742. xpath_allocator_capture cr(stack.result);
  8743. double r = 0;
  8744. xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_all);
  8745. for (const xpath_node* it = ns.begin(); it != ns.end(); ++it)
  8746. {
  8747. xpath_allocator_capture cri(stack.result);
  8748. r += convert_string_to_number(string_value(*it, stack.result).c_str());
  8749. }
  8750. return r;
  8751. }
  8752. case ast_func_floor:
  8753. {
  8754. double r = _left->eval_number(c, stack);
  8755. return r == r ? floor(r) : r;
  8756. }
  8757. case ast_func_ceiling:
  8758. {
  8759. double r = _left->eval_number(c, stack);
  8760. return r == r ? ceil(r) : r;
  8761. }
  8762. case ast_func_round:
  8763. return round_nearest_nzero(_left->eval_number(c, stack));
  8764. case ast_variable:
  8765. {
  8766. assert(_rettype == _data.variable->type());
  8767. if (_rettype == xpath_type_number)
  8768. return _data.variable->get_number();
  8769. // variable needs to be converted to the correct type, this is handled by the fallthrough block below
  8770. break;
  8771. }
  8772. default:
  8773. ;
  8774. }
  8775. // none of the ast types that return the value directly matched, we need to perform type conversion
  8776. switch (_rettype)
  8777. {
  8778. case xpath_type_boolean:
  8779. return eval_boolean(c, stack) ? 1 : 0;
  8780. case xpath_type_string:
  8781. case xpath_type_node_set: // implicit conversion to string
  8782. {
  8783. xpath_allocator_capture cr(stack.result);
  8784. return convert_string_to_number(eval_string(c, stack).c_str());
  8785. }
  8786. default:
  8787. assert(false && "Wrong expression for return type number"); // unreachable
  8788. return 0;
  8789. }
  8790. }
  8791. xpath_string eval_string_concat(const xpath_context& c, const xpath_stack& stack)
  8792. {
  8793. assert(_type == ast_func_concat);
  8794. xpath_allocator_capture ct(stack.temp);
  8795. // count the string number
  8796. size_t count = 1;
  8797. for (xpath_ast_node* nc = _right; nc; nc = nc->_next) count++;
  8798. // allocate a buffer for temporary string objects
  8799. xpath_string* buffer = static_cast<xpath_string*>(stack.temp->allocate(count * sizeof(xpath_string)));
  8800. if (!buffer) return xpath_string();
  8801. // evaluate all strings to temporary stack
  8802. xpath_stack swapped_stack = {stack.temp, stack.result};
  8803. buffer[0] = _left->eval_string(c, swapped_stack);
  8804. size_t pos = 1;
  8805. for (xpath_ast_node* n = _right; n; n = n->_next, ++pos) buffer[pos] = n->eval_string(c, swapped_stack);
  8806. assert(pos == count);
  8807. // get total length
  8808. size_t length = 0;
  8809. for (size_t i = 0; i < count; ++i) length += buffer[i].length();
  8810. // create final string
  8811. char_t* result = static_cast<char_t*>(stack.result->allocate((length + 1) * sizeof(char_t)));
  8812. if (!result) return xpath_string();
  8813. char_t* ri = result;
  8814. for (size_t j = 0; j < count; ++j)
  8815. for (const char_t* bi = buffer[j].c_str(); *bi; ++bi)
  8816. *ri++ = *bi;
  8817. *ri = 0;
  8818. return xpath_string::from_heap_preallocated(result, ri);
  8819. }
  8820. xpath_string eval_string(const xpath_context& c, const xpath_stack& stack)
  8821. {
  8822. switch (_type)
  8823. {
  8824. case ast_string_constant:
  8825. return xpath_string::from_const(_data.string);
  8826. case ast_func_local_name_0:
  8827. {
  8828. xpath_node na = c.n;
  8829. return xpath_string::from_const(local_name(na));
  8830. }
  8831. case ast_func_local_name_1:
  8832. {
  8833. xpath_allocator_capture cr(stack.result);
  8834. xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first);
  8835. xpath_node na = ns.first();
  8836. return xpath_string::from_const(local_name(na));
  8837. }
  8838. case ast_func_name_0:
  8839. {
  8840. xpath_node na = c.n;
  8841. return xpath_string::from_const(qualified_name(na));
  8842. }
  8843. case ast_func_name_1:
  8844. {
  8845. xpath_allocator_capture cr(stack.result);
  8846. xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first);
  8847. xpath_node na = ns.first();
  8848. return xpath_string::from_const(qualified_name(na));
  8849. }
  8850. case ast_func_namespace_uri_0:
  8851. {
  8852. xpath_node na = c.n;
  8853. return xpath_string::from_const(namespace_uri(na));
  8854. }
  8855. case ast_func_namespace_uri_1:
  8856. {
  8857. xpath_allocator_capture cr(stack.result);
  8858. xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first);
  8859. xpath_node na = ns.first();
  8860. return xpath_string::from_const(namespace_uri(na));
  8861. }
  8862. case ast_func_string_0:
  8863. return string_value(c.n, stack.result);
  8864. case ast_func_string_1:
  8865. return _left->eval_string(c, stack);
  8866. case ast_func_concat:
  8867. return eval_string_concat(c, stack);
  8868. case ast_func_substring_before:
  8869. {
  8870. xpath_allocator_capture cr(stack.temp);
  8871. xpath_stack swapped_stack = {stack.temp, stack.result};
  8872. xpath_string s = _left->eval_string(c, swapped_stack);
  8873. xpath_string p = _right->eval_string(c, swapped_stack);
  8874. const char_t* pos = find_substring(s.c_str(), p.c_str());
  8875. return pos ? xpath_string::from_heap(s.c_str(), pos, stack.result) : xpath_string();
  8876. }
  8877. case ast_func_substring_after:
  8878. {
  8879. xpath_allocator_capture cr(stack.temp);
  8880. xpath_stack swapped_stack = {stack.temp, stack.result};
  8881. xpath_string s = _left->eval_string(c, swapped_stack);
  8882. xpath_string p = _right->eval_string(c, swapped_stack);
  8883. const char_t* pos = find_substring(s.c_str(), p.c_str());
  8884. if (!pos) return xpath_string();
  8885. const char_t* rbegin = pos + p.length();
  8886. const char_t* rend = s.c_str() + s.length();
  8887. return s.uses_heap() ? xpath_string::from_heap(rbegin, rend, stack.result) : xpath_string::from_const(rbegin);
  8888. }
  8889. case ast_func_substring_2:
  8890. {
  8891. xpath_allocator_capture cr(stack.temp);
  8892. xpath_stack swapped_stack = {stack.temp, stack.result};
  8893. xpath_string s = _left->eval_string(c, swapped_stack);
  8894. size_t s_length = s.length();
  8895. double first = round_nearest(_right->eval_number(c, stack));
  8896. if (is_nan(first)) return xpath_string(); // NaN
  8897. else if (first >= static_cast<double>(s_length + 1)) return xpath_string();
  8898. size_t pos = first < 1 ? 1 : static_cast<size_t>(first);
  8899. assert(1 <= pos && pos <= s_length + 1);
  8900. const char_t* rbegin = s.c_str() + (pos - 1);
  8901. const char_t* rend = s.c_str() + s.length();
  8902. return s.uses_heap() ? xpath_string::from_heap(rbegin, rend, stack.result) : xpath_string::from_const(rbegin);
  8903. }
  8904. case ast_func_substring_3:
  8905. {
  8906. xpath_allocator_capture cr(stack.temp);
  8907. xpath_stack swapped_stack = {stack.temp, stack.result};
  8908. xpath_string s = _left->eval_string(c, swapped_stack);
  8909. size_t s_length = s.length();
  8910. double first = round_nearest(_right->eval_number(c, stack));
  8911. double last = first + round_nearest(_right->_next->eval_number(c, stack));
  8912. if (is_nan(first) || is_nan(last)) return xpath_string();
  8913. else if (first >= static_cast<double>(s_length + 1)) return xpath_string();
  8914. else if (first >= last) return xpath_string();
  8915. else if (last < 1) return xpath_string();
  8916. size_t pos = first < 1 ? 1 : static_cast<size_t>(first);
  8917. size_t end = last >= static_cast<double>(s_length + 1) ? s_length + 1 : static_cast<size_t>(last);
  8918. assert(1 <= pos && pos <= end && end <= s_length + 1);
  8919. const char_t* rbegin = s.c_str() + (pos - 1);
  8920. const char_t* rend = s.c_str() + (end - 1);
  8921. return (end == s_length + 1 && !s.uses_heap()) ? xpath_string::from_const(rbegin) : xpath_string::from_heap(rbegin, rend, stack.result);
  8922. }
  8923. case ast_func_normalize_space_0:
  8924. {
  8925. xpath_string s = string_value(c.n, stack.result);
  8926. char_t* begin = s.data(stack.result);
  8927. if (!begin) return xpath_string();
  8928. char_t* end = normalize_space(begin);
  8929. return xpath_string::from_heap_preallocated(begin, end);
  8930. }
  8931. case ast_func_normalize_space_1:
  8932. {
  8933. xpath_string s = _left->eval_string(c, stack);
  8934. char_t* begin = s.data(stack.result);
  8935. if (!begin) return xpath_string();
  8936. char_t* end = normalize_space(begin);
  8937. return xpath_string::from_heap_preallocated(begin, end);
  8938. }
  8939. case ast_func_translate:
  8940. {
  8941. xpath_allocator_capture cr(stack.temp);
  8942. xpath_stack swapped_stack = {stack.temp, stack.result};
  8943. xpath_string s = _left->eval_string(c, stack);
  8944. xpath_string from = _right->eval_string(c, swapped_stack);
  8945. xpath_string to = _right->_next->eval_string(c, swapped_stack);
  8946. char_t* begin = s.data(stack.result);
  8947. if (!begin) return xpath_string();
  8948. char_t* end = translate(begin, from.c_str(), to.c_str(), to.length());
  8949. return xpath_string::from_heap_preallocated(begin, end);
  8950. }
  8951. case ast_opt_translate_table:
  8952. {
  8953. xpath_string s = _left->eval_string(c, stack);
  8954. char_t* begin = s.data(stack.result);
  8955. if (!begin) return xpath_string();
  8956. char_t* end = translate_table(begin, _data.table);
  8957. return xpath_string::from_heap_preallocated(begin, end);
  8958. }
  8959. case ast_variable:
  8960. {
  8961. assert(_rettype == _data.variable->type());
  8962. if (_rettype == xpath_type_string)
  8963. return xpath_string::from_const(_data.variable->get_string());
  8964. // variable needs to be converted to the correct type, this is handled by the fallthrough block below
  8965. break;
  8966. }
  8967. default:
  8968. ;
  8969. }
  8970. // none of the ast types that return the value directly matched, we need to perform type conversion
  8971. switch (_rettype)
  8972. {
  8973. case xpath_type_boolean:
  8974. return xpath_string::from_const(eval_boolean(c, stack) ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false"));
  8975. case xpath_type_number:
  8976. return convert_number_to_string(eval_number(c, stack), stack.result);
  8977. case xpath_type_node_set:
  8978. {
  8979. xpath_allocator_capture cr(stack.temp);
  8980. xpath_stack swapped_stack = {stack.temp, stack.result};
  8981. xpath_node_set_raw ns = eval_node_set(c, swapped_stack, nodeset_eval_first);
  8982. return ns.empty() ? xpath_string() : string_value(ns.first(), stack.result);
  8983. }
  8984. default:
  8985. assert(false && "Wrong expression for return type string"); // unreachable
  8986. return xpath_string();
  8987. }
  8988. }
  8989. xpath_node_set_raw eval_node_set(const xpath_context& c, const xpath_stack& stack, nodeset_eval_t eval)
  8990. {
  8991. switch (_type)
  8992. {
  8993. case ast_op_union:
  8994. {
  8995. xpath_allocator_capture cr(stack.temp);
  8996. xpath_stack swapped_stack = {stack.temp, stack.result};
  8997. xpath_node_set_raw ls = _left->eval_node_set(c, stack, eval);
  8998. xpath_node_set_raw rs = _right->eval_node_set(c, swapped_stack, eval);
  8999. // we can optimize merging two sorted sets, but this is a very rare operation, so don't bother
  9000. ls.set_type(xpath_node_set::type_unsorted);
  9001. ls.append(rs.begin(), rs.end(), stack.result);
  9002. ls.remove_duplicates(stack.temp);
  9003. return ls;
  9004. }
  9005. case ast_filter:
  9006. {
  9007. xpath_node_set_raw set = _left->eval_node_set(c, stack, _test == predicate_constant_one ? nodeset_eval_first : nodeset_eval_all);
  9008. // either expression is a number or it contains position() call; sort by document order
  9009. if (_test != predicate_posinv) set.sort_do();
  9010. bool once = eval_once(set.type(), eval);
  9011. apply_predicate(set, 0, stack, once);
  9012. return set;
  9013. }
  9014. case ast_func_id:
  9015. return xpath_node_set_raw();
  9016. case ast_step:
  9017. {
  9018. switch (_axis)
  9019. {
  9020. case axis_ancestor:
  9021. return step_do(c, stack, eval, axis_to_type<axis_ancestor>());
  9022. case axis_ancestor_or_self:
  9023. return step_do(c, stack, eval, axis_to_type<axis_ancestor_or_self>());
  9024. case axis_attribute:
  9025. return step_do(c, stack, eval, axis_to_type<axis_attribute>());
  9026. case axis_child:
  9027. return step_do(c, stack, eval, axis_to_type<axis_child>());
  9028. case axis_descendant:
  9029. return step_do(c, stack, eval, axis_to_type<axis_descendant>());
  9030. case axis_descendant_or_self:
  9031. return step_do(c, stack, eval, axis_to_type<axis_descendant_or_self>());
  9032. case axis_following:
  9033. return step_do(c, stack, eval, axis_to_type<axis_following>());
  9034. case axis_following_sibling:
  9035. return step_do(c, stack, eval, axis_to_type<axis_following_sibling>());
  9036. case axis_namespace:
  9037. // namespaced axis is not supported
  9038. return xpath_node_set_raw();
  9039. case axis_parent:
  9040. return step_do(c, stack, eval, axis_to_type<axis_parent>());
  9041. case axis_preceding:
  9042. return step_do(c, stack, eval, axis_to_type<axis_preceding>());
  9043. case axis_preceding_sibling:
  9044. return step_do(c, stack, eval, axis_to_type<axis_preceding_sibling>());
  9045. case axis_self:
  9046. return step_do(c, stack, eval, axis_to_type<axis_self>());
  9047. default:
  9048. assert(false && "Unknown axis"); // unreachable
  9049. return xpath_node_set_raw();
  9050. }
  9051. }
  9052. case ast_step_root:
  9053. {
  9054. assert(!_right); // root step can't have any predicates
  9055. xpath_node_set_raw ns;
  9056. ns.set_type(xpath_node_set::type_sorted);
  9057. if (c.n.node()) ns.push_back(c.n.node().root(), stack.result);
  9058. else if (c.n.attribute()) ns.push_back(c.n.parent().root(), stack.result);
  9059. return ns;
  9060. }
  9061. case ast_variable:
  9062. {
  9063. assert(_rettype == _data.variable->type());
  9064. if (_rettype == xpath_type_node_set)
  9065. {
  9066. const xpath_node_set& s = _data.variable->get_node_set();
  9067. xpath_node_set_raw ns;
  9068. ns.set_type(s.type());
  9069. ns.append(s.begin(), s.end(), stack.result);
  9070. return ns;
  9071. }
  9072. // variable needs to be converted to the correct type, this is handled by the fallthrough block below
  9073. break;
  9074. }
  9075. default:
  9076. ;
  9077. }
  9078. // none of the ast types that return the value directly matched, but conversions to node set are invalid
  9079. assert(false && "Wrong expression for return type node set"); // unreachable
  9080. return xpath_node_set_raw();
  9081. }
  9082. void optimize(xpath_allocator* alloc)
  9083. {
  9084. if (_left)
  9085. _left->optimize(alloc);
  9086. if (_right)
  9087. _right->optimize(alloc);
  9088. if (_next)
  9089. _next->optimize(alloc);
  9090. // coverity[var_deref_model]
  9091. optimize_self(alloc);
  9092. }
  9093. void optimize_self(xpath_allocator* alloc)
  9094. {
  9095. // Rewrite [position()=expr] with [expr]
  9096. // Note that this step has to go before classification to recognize [position()=1]
  9097. if ((_type == ast_filter || _type == ast_predicate) &&
  9098. _right && // workaround for clang static analyzer (_right is never null for ast_filter/ast_predicate)
  9099. _right->_type == ast_op_equal && _right->_left->_type == ast_func_position && _right->_right->_rettype == xpath_type_number)
  9100. {
  9101. _right = _right->_right;
  9102. }
  9103. // Classify filter/predicate ops to perform various optimizations during evaluation
  9104. if ((_type == ast_filter || _type == ast_predicate) && _right) // workaround for clang static analyzer (_right is never null for ast_filter/ast_predicate)
  9105. {
  9106. assert(_test == predicate_default);
  9107. if (_right->_type == ast_number_constant && _right->_data.number == 1.0)
  9108. _test = predicate_constant_one;
  9109. else if (_right->_rettype == xpath_type_number && (_right->_type == ast_number_constant || _right->_type == ast_variable || _right->_type == ast_func_last))
  9110. _test = predicate_constant;
  9111. else if (_right->_rettype != xpath_type_number && _right->is_posinv_expr())
  9112. _test = predicate_posinv;
  9113. }
  9114. // Rewrite descendant-or-self::node()/child::foo with descendant::foo
  9115. // The former is a full form of //foo, the latter is much faster since it executes the node test immediately
  9116. // Do a similar kind of rewrite for self/descendant/descendant-or-self axes
  9117. // Note that we only rewrite positionally invariant steps (//foo[1] != /descendant::foo[1])
  9118. if (_type == ast_step && (_axis == axis_child || _axis == axis_self || _axis == axis_descendant || _axis == axis_descendant_or_self) &&
  9119. _left && _left->_type == ast_step && _left->_axis == axis_descendant_or_self && _left->_test == nodetest_type_node && !_left->_right &&
  9120. is_posinv_step())
  9121. {
  9122. if (_axis == axis_child || _axis == axis_descendant)
  9123. _axis = axis_descendant;
  9124. else
  9125. _axis = axis_descendant_or_self;
  9126. _left = _left->_left;
  9127. }
  9128. // Use optimized lookup table implementation for translate() with constant arguments
  9129. if (_type == ast_func_translate &&
  9130. _right && // workaround for clang static analyzer (_right is never null for ast_func_translate)
  9131. _right->_type == ast_string_constant && _right->_next->_type == ast_string_constant)
  9132. {
  9133. unsigned char* table = translate_table_generate(alloc, _right->_data.string, _right->_next->_data.string);
  9134. if (table)
  9135. {
  9136. _type = ast_opt_translate_table;
  9137. _data.table = table;
  9138. }
  9139. }
  9140. // Use optimized path for @attr = 'value' or @attr = $value
  9141. if (_type == ast_op_equal &&
  9142. _left && _right && // workaround for clang static analyzer and Coverity (_left and _right are never null for ast_op_equal)
  9143. // coverity[mixed_enums]
  9144. _left->_type == ast_step && _left->_axis == axis_attribute && _left->_test == nodetest_name && !_left->_left && !_left->_right &&
  9145. (_right->_type == ast_string_constant || (_right->_type == ast_variable && _right->_rettype == xpath_type_string)))
  9146. {
  9147. _type = ast_opt_compare_attribute;
  9148. }
  9149. }
  9150. bool is_posinv_expr() const
  9151. {
  9152. switch (_type)
  9153. {
  9154. case ast_func_position:
  9155. case ast_func_last:
  9156. return false;
  9157. case ast_string_constant:
  9158. case ast_number_constant:
  9159. case ast_variable:
  9160. return true;
  9161. case ast_step:
  9162. case ast_step_root:
  9163. return true;
  9164. case ast_predicate:
  9165. case ast_filter:
  9166. return true;
  9167. default:
  9168. if (_left && !_left->is_posinv_expr()) return false;
  9169. for (xpath_ast_node* n = _right; n; n = n->_next)
  9170. if (!n->is_posinv_expr()) return false;
  9171. return true;
  9172. }
  9173. }
  9174. bool is_posinv_step() const
  9175. {
  9176. assert(_type == ast_step);
  9177. for (xpath_ast_node* n = _right; n; n = n->_next)
  9178. {
  9179. assert(n->_type == ast_predicate);
  9180. if (n->_test != predicate_posinv)
  9181. return false;
  9182. }
  9183. return true;
  9184. }
  9185. xpath_value_type rettype() const
  9186. {
  9187. return static_cast<xpath_value_type>(_rettype);
  9188. }
  9189. };
  9190. static const size_t xpath_ast_depth_limit =
  9191. #ifdef PUGIXML_XPATH_DEPTH_LIMIT
  9192. PUGIXML_XPATH_DEPTH_LIMIT
  9193. #else
  9194. 1024
  9195. #endif
  9196. ;
  9197. struct xpath_parser
  9198. {
  9199. xpath_allocator* _alloc;
  9200. xpath_lexer _lexer;
  9201. const char_t* _query;
  9202. xpath_variable_set* _variables;
  9203. xpath_parse_result* _result;
  9204. char_t _scratch[32];
  9205. size_t _depth;
  9206. xpath_ast_node* error(const char* message)
  9207. {
  9208. _result->error = message;
  9209. _result->offset = _lexer.current_pos() - _query;
  9210. return NULL;
  9211. }
  9212. xpath_ast_node* error_oom()
  9213. {
  9214. assert(_alloc->_error);
  9215. *_alloc->_error = true;
  9216. return NULL;
  9217. }
  9218. xpath_ast_node* error_rec()
  9219. {
  9220. return error("Exceeded maximum allowed query depth");
  9221. }
  9222. void* alloc_node()
  9223. {
  9224. return _alloc->allocate(sizeof(xpath_ast_node));
  9225. }
  9226. xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, const char_t* value)
  9227. {
  9228. void* memory = alloc_node();
  9229. return memory ? new (memory) xpath_ast_node(type, rettype, value) : NULL;
  9230. }
  9231. xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, double value)
  9232. {
  9233. void* memory = alloc_node();
  9234. return memory ? new (memory) xpath_ast_node(type, rettype, value) : NULL;
  9235. }
  9236. xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, xpath_variable* value)
  9237. {
  9238. void* memory = alloc_node();
  9239. return memory ? new (memory) xpath_ast_node(type, rettype, value) : NULL;
  9240. }
  9241. xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, xpath_ast_node* left = NULL, xpath_ast_node* right = NULL)
  9242. {
  9243. void* memory = alloc_node();
  9244. return memory ? new (memory) xpath_ast_node(type, rettype, left, right) : NULL;
  9245. }
  9246. xpath_ast_node* alloc_node(ast_type_t type, xpath_ast_node* left, axis_t axis, nodetest_t test, const char_t* contents)
  9247. {
  9248. void* memory = alloc_node();
  9249. return memory ? new (memory) xpath_ast_node(type, left, axis, test, contents) : NULL;
  9250. }
  9251. xpath_ast_node* alloc_node(ast_type_t type, xpath_ast_node* left, xpath_ast_node* right, predicate_t test)
  9252. {
  9253. void* memory = alloc_node();
  9254. return memory ? new (memory) xpath_ast_node(type, left, right, test) : NULL;
  9255. }
  9256. const char_t* alloc_string(const xpath_lexer_string& value)
  9257. {
  9258. if (!value.begin)
  9259. return PUGIXML_TEXT("");
  9260. size_t length = static_cast<size_t>(value.end - value.begin);
  9261. char_t* c = static_cast<char_t*>(_alloc->allocate((length + 1) * sizeof(char_t)));
  9262. if (!c) return NULL;
  9263. memcpy(c, value.begin, length * sizeof(char_t));
  9264. c[length] = 0;
  9265. return c;
  9266. }
  9267. xpath_ast_node* parse_function(const xpath_lexer_string& name, size_t argc, xpath_ast_node* args[2])
  9268. {
  9269. switch (name.begin[0])
  9270. {
  9271. case 'b':
  9272. if (name == PUGIXML_TEXT("boolean") && argc == 1)
  9273. return alloc_node(ast_func_boolean, xpath_type_boolean, args[0]);
  9274. break;
  9275. case 'c':
  9276. if (name == PUGIXML_TEXT("count") && argc == 1)
  9277. {
  9278. if (args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set");
  9279. return alloc_node(ast_func_count, xpath_type_number, args[0]);
  9280. }
  9281. else if (name == PUGIXML_TEXT("contains") && argc == 2)
  9282. return alloc_node(ast_func_contains, xpath_type_boolean, args[0], args[1]);
  9283. else if (name == PUGIXML_TEXT("concat") && argc >= 2)
  9284. return alloc_node(ast_func_concat, xpath_type_string, args[0], args[1]);
  9285. else if (name == PUGIXML_TEXT("ceiling") && argc == 1)
  9286. return alloc_node(ast_func_ceiling, xpath_type_number, args[0]);
  9287. break;
  9288. case 'f':
  9289. if (name == PUGIXML_TEXT("false") && argc == 0)
  9290. return alloc_node(ast_func_false, xpath_type_boolean);
  9291. else if (name == PUGIXML_TEXT("floor") && argc == 1)
  9292. return alloc_node(ast_func_floor, xpath_type_number, args[0]);
  9293. break;
  9294. case 'i':
  9295. if (name == PUGIXML_TEXT("id") && argc == 1)
  9296. return alloc_node(ast_func_id, xpath_type_node_set, args[0]);
  9297. break;
  9298. case 'l':
  9299. if (name == PUGIXML_TEXT("last") && argc == 0)
  9300. return alloc_node(ast_func_last, xpath_type_number);
  9301. else if (name == PUGIXML_TEXT("lang") && argc == 1)
  9302. return alloc_node(ast_func_lang, xpath_type_boolean, args[0]);
  9303. else if (name == PUGIXML_TEXT("local-name") && argc <= 1)
  9304. {
  9305. if (argc == 1 && args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set");
  9306. return alloc_node(argc == 0 ? ast_func_local_name_0 : ast_func_local_name_1, xpath_type_string, args[0]);
  9307. }
  9308. break;
  9309. case 'n':
  9310. if (name == PUGIXML_TEXT("name") && argc <= 1)
  9311. {
  9312. if (argc == 1 && args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set");
  9313. return alloc_node(argc == 0 ? ast_func_name_0 : ast_func_name_1, xpath_type_string, args[0]);
  9314. }
  9315. else if (name == PUGIXML_TEXT("namespace-uri") && argc <= 1)
  9316. {
  9317. if (argc == 1 && args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set");
  9318. return alloc_node(argc == 0 ? ast_func_namespace_uri_0 : ast_func_namespace_uri_1, xpath_type_string, args[0]);
  9319. }
  9320. else if (name == PUGIXML_TEXT("normalize-space") && argc <= 1)
  9321. return alloc_node(argc == 0 ? ast_func_normalize_space_0 : ast_func_normalize_space_1, xpath_type_string, args[0], args[1]);
  9322. else if (name == PUGIXML_TEXT("not") && argc == 1)
  9323. return alloc_node(ast_func_not, xpath_type_boolean, args[0]);
  9324. else if (name == PUGIXML_TEXT("number") && argc <= 1)
  9325. return alloc_node(argc == 0 ? ast_func_number_0 : ast_func_number_1, xpath_type_number, args[0]);
  9326. break;
  9327. case 'p':
  9328. if (name == PUGIXML_TEXT("position") && argc == 0)
  9329. return alloc_node(ast_func_position, xpath_type_number);
  9330. break;
  9331. case 'r':
  9332. if (name == PUGIXML_TEXT("round") && argc == 1)
  9333. return alloc_node(ast_func_round, xpath_type_number, args[0]);
  9334. break;
  9335. case 's':
  9336. if (name == PUGIXML_TEXT("string") && argc <= 1)
  9337. return alloc_node(argc == 0 ? ast_func_string_0 : ast_func_string_1, xpath_type_string, args[0]);
  9338. else if (name == PUGIXML_TEXT("string-length") && argc <= 1)
  9339. return alloc_node(argc == 0 ? ast_func_string_length_0 : ast_func_string_length_1, xpath_type_number, args[0]);
  9340. else if (name == PUGIXML_TEXT("starts-with") && argc == 2)
  9341. return alloc_node(ast_func_starts_with, xpath_type_boolean, args[0], args[1]);
  9342. else if (name == PUGIXML_TEXT("substring-before") && argc == 2)
  9343. return alloc_node(ast_func_substring_before, xpath_type_string, args[0], args[1]);
  9344. else if (name == PUGIXML_TEXT("substring-after") && argc == 2)
  9345. return alloc_node(ast_func_substring_after, xpath_type_string, args[0], args[1]);
  9346. else if (name == PUGIXML_TEXT("substring") && (argc == 2 || argc == 3))
  9347. return alloc_node(argc == 2 ? ast_func_substring_2 : ast_func_substring_3, xpath_type_string, args[0], args[1]);
  9348. else if (name == PUGIXML_TEXT("sum") && argc == 1)
  9349. {
  9350. if (args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set");
  9351. return alloc_node(ast_func_sum, xpath_type_number, args[0]);
  9352. }
  9353. break;
  9354. case 't':
  9355. if (name == PUGIXML_TEXT("translate") && argc == 3)
  9356. return alloc_node(ast_func_translate, xpath_type_string, args[0], args[1]);
  9357. else if (name == PUGIXML_TEXT("true") && argc == 0)
  9358. return alloc_node(ast_func_true, xpath_type_boolean);
  9359. break;
  9360. default:
  9361. break;
  9362. }
  9363. return error("Unrecognized function or wrong parameter count");
  9364. }
  9365. axis_t parse_axis_name(const xpath_lexer_string& name, bool& specified)
  9366. {
  9367. specified = true;
  9368. switch (name.begin[0])
  9369. {
  9370. case 'a':
  9371. if (name == PUGIXML_TEXT("ancestor"))
  9372. return axis_ancestor;
  9373. else if (name == PUGIXML_TEXT("ancestor-or-self"))
  9374. return axis_ancestor_or_self;
  9375. else if (name == PUGIXML_TEXT("attribute"))
  9376. return axis_attribute;
  9377. break;
  9378. case 'c':
  9379. if (name == PUGIXML_TEXT("child"))
  9380. return axis_child;
  9381. break;
  9382. case 'd':
  9383. if (name == PUGIXML_TEXT("descendant"))
  9384. return axis_descendant;
  9385. else if (name == PUGIXML_TEXT("descendant-or-self"))
  9386. return axis_descendant_or_self;
  9387. break;
  9388. case 'f':
  9389. if (name == PUGIXML_TEXT("following"))
  9390. return axis_following;
  9391. else if (name == PUGIXML_TEXT("following-sibling"))
  9392. return axis_following_sibling;
  9393. break;
  9394. case 'n':
  9395. if (name == PUGIXML_TEXT("namespace"))
  9396. return axis_namespace;
  9397. break;
  9398. case 'p':
  9399. if (name == PUGIXML_TEXT("parent"))
  9400. return axis_parent;
  9401. else if (name == PUGIXML_TEXT("preceding"))
  9402. return axis_preceding;
  9403. else if (name == PUGIXML_TEXT("preceding-sibling"))
  9404. return axis_preceding_sibling;
  9405. break;
  9406. case 's':
  9407. if (name == PUGIXML_TEXT("self"))
  9408. return axis_self;
  9409. break;
  9410. default:
  9411. break;
  9412. }
  9413. specified = false;
  9414. return axis_child;
  9415. }
  9416. nodetest_t parse_node_test_type(const xpath_lexer_string& name)
  9417. {
  9418. switch (name.begin[0])
  9419. {
  9420. case 'c':
  9421. if (name == PUGIXML_TEXT("comment"))
  9422. return nodetest_type_comment;
  9423. break;
  9424. case 'n':
  9425. if (name == PUGIXML_TEXT("node"))
  9426. return nodetest_type_node;
  9427. break;
  9428. case 'p':
  9429. if (name == PUGIXML_TEXT("processing-instruction"))
  9430. return nodetest_type_pi;
  9431. break;
  9432. case 't':
  9433. if (name == PUGIXML_TEXT("text"))
  9434. return nodetest_type_text;
  9435. break;
  9436. default:
  9437. break;
  9438. }
  9439. return nodetest_none;
  9440. }
  9441. // PrimaryExpr ::= VariableReference | '(' Expr ')' | Literal | Number | FunctionCall
  9442. xpath_ast_node* parse_primary_expression()
  9443. {
  9444. switch (_lexer.current())
  9445. {
  9446. case lex_var_ref:
  9447. {
  9448. xpath_lexer_string name = _lexer.contents();
  9449. if (!_variables)
  9450. return error("Unknown variable: variable set is not provided");
  9451. xpath_variable* var = NULL;
  9452. if (!get_variable_scratch(_scratch, _variables, name.begin, name.end, &var))
  9453. return error_oom();
  9454. if (!var)
  9455. return error("Unknown variable: variable set does not contain the given name");
  9456. _lexer.next();
  9457. return alloc_node(ast_variable, var->type(), var);
  9458. }
  9459. case lex_open_brace:
  9460. {
  9461. _lexer.next();
  9462. xpath_ast_node* n = parse_expression();
  9463. if (!n) return NULL;
  9464. if (_lexer.current() != lex_close_brace)
  9465. return error("Expected ')' to match an opening '('");
  9466. _lexer.next();
  9467. return n;
  9468. }
  9469. case lex_quoted_string:
  9470. {
  9471. const char_t* value = alloc_string(_lexer.contents());
  9472. if (!value) return NULL;
  9473. _lexer.next();
  9474. return alloc_node(ast_string_constant, xpath_type_string, value);
  9475. }
  9476. case lex_number:
  9477. {
  9478. double value = 0;
  9479. if (!convert_string_to_number_scratch(_scratch, _lexer.contents().begin, _lexer.contents().end, &value))
  9480. return error_oom();
  9481. _lexer.next();
  9482. return alloc_node(ast_number_constant, xpath_type_number, value);
  9483. }
  9484. case lex_string:
  9485. {
  9486. xpath_ast_node* args[2] = {NULL};
  9487. size_t argc = 0;
  9488. xpath_lexer_string function = _lexer.contents();
  9489. _lexer.next();
  9490. xpath_ast_node* last_arg = NULL;
  9491. if (_lexer.current() != lex_open_brace)
  9492. return error("Unrecognized function call");
  9493. _lexer.next();
  9494. size_t old_depth = _depth;
  9495. while (_lexer.current() != lex_close_brace)
  9496. {
  9497. if (argc > 0)
  9498. {
  9499. if (_lexer.current() != lex_comma)
  9500. return error("No comma between function arguments");
  9501. _lexer.next();
  9502. }
  9503. if (++_depth > xpath_ast_depth_limit)
  9504. return error_rec();
  9505. xpath_ast_node* n = parse_expression();
  9506. if (!n) return NULL;
  9507. if (argc < 2) args[argc] = n;
  9508. else last_arg->set_next(n);
  9509. argc++;
  9510. last_arg = n;
  9511. }
  9512. _lexer.next();
  9513. _depth = old_depth;
  9514. return parse_function(function, argc, args);
  9515. }
  9516. default:
  9517. return error("Unrecognizable primary expression");
  9518. }
  9519. }
  9520. // FilterExpr ::= PrimaryExpr | FilterExpr Predicate
  9521. // Predicate ::= '[' PredicateExpr ']'
  9522. // PredicateExpr ::= Expr
  9523. xpath_ast_node* parse_filter_expression()
  9524. {
  9525. xpath_ast_node* n = parse_primary_expression();
  9526. if (!n) return NULL;
  9527. size_t old_depth = _depth;
  9528. while (_lexer.current() == lex_open_square_brace)
  9529. {
  9530. _lexer.next();
  9531. if (++_depth > xpath_ast_depth_limit)
  9532. return error_rec();
  9533. if (n->rettype() != xpath_type_node_set)
  9534. return error("Predicate has to be applied to node set");
  9535. xpath_ast_node* expr = parse_expression();
  9536. if (!expr) return NULL;
  9537. n = alloc_node(ast_filter, n, expr, predicate_default);
  9538. if (!n) return NULL;
  9539. if (_lexer.current() != lex_close_square_brace)
  9540. return error("Expected ']' to match an opening '['");
  9541. _lexer.next();
  9542. }
  9543. _depth = old_depth;
  9544. return n;
  9545. }
  9546. // Step ::= AxisSpecifier NodeTest Predicate* | AbbreviatedStep
  9547. // AxisSpecifier ::= AxisName '::' | '@'?
  9548. // NodeTest ::= NameTest | NodeType '(' ')' | 'processing-instruction' '(' Literal ')'
  9549. // NameTest ::= '*' | NCName ':' '*' | QName
  9550. // AbbreviatedStep ::= '.' | '..'
  9551. xpath_ast_node* parse_step(xpath_ast_node* set)
  9552. {
  9553. if (set && set->rettype() != xpath_type_node_set)
  9554. return error("Step has to be applied to node set");
  9555. bool axis_specified = false;
  9556. axis_t axis = axis_child; // implied child axis
  9557. if (_lexer.current() == lex_axis_attribute)
  9558. {
  9559. axis = axis_attribute;
  9560. axis_specified = true;
  9561. _lexer.next();
  9562. }
  9563. else if (_lexer.current() == lex_dot)
  9564. {
  9565. _lexer.next();
  9566. if (_lexer.current() == lex_open_square_brace)
  9567. return error("Predicates are not allowed after an abbreviated step");
  9568. return alloc_node(ast_step, set, axis_self, nodetest_type_node, NULL);
  9569. }
  9570. else if (_lexer.current() == lex_double_dot)
  9571. {
  9572. _lexer.next();
  9573. if (_lexer.current() == lex_open_square_brace)
  9574. return error("Predicates are not allowed after an abbreviated step");
  9575. return alloc_node(ast_step, set, axis_parent, nodetest_type_node, NULL);
  9576. }
  9577. nodetest_t nt_type = nodetest_none;
  9578. xpath_lexer_string nt_name;
  9579. if (_lexer.current() == lex_string)
  9580. {
  9581. // node name test
  9582. nt_name = _lexer.contents();
  9583. _lexer.next();
  9584. // was it an axis name?
  9585. if (_lexer.current() == lex_double_colon)
  9586. {
  9587. // parse axis name
  9588. if (axis_specified)
  9589. return error("Two axis specifiers in one step");
  9590. axis = parse_axis_name(nt_name, axis_specified);
  9591. if (!axis_specified)
  9592. return error("Unknown axis");
  9593. // read actual node test
  9594. _lexer.next();
  9595. if (_lexer.current() == lex_multiply)
  9596. {
  9597. nt_type = nodetest_all;
  9598. nt_name = xpath_lexer_string();
  9599. _lexer.next();
  9600. }
  9601. else if (_lexer.current() == lex_string)
  9602. {
  9603. nt_name = _lexer.contents();
  9604. _lexer.next();
  9605. }
  9606. else
  9607. {
  9608. return error("Unrecognized node test");
  9609. }
  9610. }
  9611. if (nt_type == nodetest_none)
  9612. {
  9613. // node type test or processing-instruction
  9614. if (_lexer.current() == lex_open_brace)
  9615. {
  9616. _lexer.next();
  9617. if (_lexer.current() == lex_close_brace)
  9618. {
  9619. _lexer.next();
  9620. nt_type = parse_node_test_type(nt_name);
  9621. if (nt_type == nodetest_none)
  9622. return error("Unrecognized node type");
  9623. nt_name = xpath_lexer_string();
  9624. }
  9625. else if (nt_name == PUGIXML_TEXT("processing-instruction"))
  9626. {
  9627. if (_lexer.current() != lex_quoted_string)
  9628. return error("Only literals are allowed as arguments to processing-instruction()");
  9629. nt_type = nodetest_pi;
  9630. nt_name = _lexer.contents();
  9631. _lexer.next();
  9632. if (_lexer.current() != lex_close_brace)
  9633. return error("Unmatched brace near processing-instruction()");
  9634. _lexer.next();
  9635. }
  9636. else
  9637. {
  9638. return error("Unmatched brace near node type test");
  9639. }
  9640. }
  9641. // QName or NCName:*
  9642. else
  9643. {
  9644. if (nt_name.end - nt_name.begin > 2 && nt_name.end[-2] == ':' && nt_name.end[-1] == '*') // NCName:*
  9645. {
  9646. nt_name.end--; // erase *
  9647. nt_type = nodetest_all_in_namespace;
  9648. }
  9649. else
  9650. {
  9651. nt_type = nodetest_name;
  9652. }
  9653. }
  9654. }
  9655. }
  9656. else if (_lexer.current() == lex_multiply)
  9657. {
  9658. nt_type = nodetest_all;
  9659. _lexer.next();
  9660. }
  9661. else
  9662. {
  9663. return error("Unrecognized node test");
  9664. }
  9665. const char_t* nt_name_copy = alloc_string(nt_name);
  9666. if (!nt_name_copy) return NULL;
  9667. xpath_ast_node* n = alloc_node(ast_step, set, axis, nt_type, nt_name_copy);
  9668. if (!n) return NULL;
  9669. size_t old_depth = _depth;
  9670. xpath_ast_node* last = NULL;
  9671. while (_lexer.current() == lex_open_square_brace)
  9672. {
  9673. _lexer.next();
  9674. if (++_depth > xpath_ast_depth_limit)
  9675. return error_rec();
  9676. xpath_ast_node* expr = parse_expression();
  9677. if (!expr) return NULL;
  9678. xpath_ast_node* pred = alloc_node(ast_predicate, NULL, expr, predicate_default);
  9679. if (!pred) return NULL;
  9680. if (_lexer.current() != lex_close_square_brace)
  9681. return error("Expected ']' to match an opening '['");
  9682. _lexer.next();
  9683. if (last) last->set_next(pred);
  9684. else n->set_right(pred);
  9685. last = pred;
  9686. }
  9687. _depth = old_depth;
  9688. return n;
  9689. }
  9690. // RelativeLocationPath ::= Step | RelativeLocationPath '/' Step | RelativeLocationPath '//' Step
  9691. xpath_ast_node* parse_relative_location_path(xpath_ast_node* set)
  9692. {
  9693. xpath_ast_node* n = parse_step(set);
  9694. if (!n) return NULL;
  9695. size_t old_depth = _depth;
  9696. while (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash)
  9697. {
  9698. lexeme_t l = _lexer.current();
  9699. _lexer.next();
  9700. if (l == lex_double_slash)
  9701. {
  9702. n = alloc_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, NULL);
  9703. if (!n) return NULL;
  9704. ++_depth;
  9705. }
  9706. if (++_depth > xpath_ast_depth_limit)
  9707. return error_rec();
  9708. n = parse_step(n);
  9709. if (!n) return NULL;
  9710. }
  9711. _depth = old_depth;
  9712. return n;
  9713. }
  9714. // LocationPath ::= RelativeLocationPath | AbsoluteLocationPath
  9715. // AbsoluteLocationPath ::= '/' RelativeLocationPath? | '//' RelativeLocationPath
  9716. xpath_ast_node* parse_location_path()
  9717. {
  9718. if (_lexer.current() == lex_slash)
  9719. {
  9720. _lexer.next();
  9721. xpath_ast_node* n = alloc_node(ast_step_root, xpath_type_node_set);
  9722. if (!n) return NULL;
  9723. // relative location path can start from axis_attribute, dot, double_dot, multiply and string lexemes; any other lexeme means standalone root path
  9724. lexeme_t l = _lexer.current();
  9725. if (l == lex_string || l == lex_axis_attribute || l == lex_dot || l == lex_double_dot || l == lex_multiply)
  9726. return parse_relative_location_path(n);
  9727. else
  9728. return n;
  9729. }
  9730. else if (_lexer.current() == lex_double_slash)
  9731. {
  9732. _lexer.next();
  9733. xpath_ast_node* n = alloc_node(ast_step_root, xpath_type_node_set);
  9734. if (!n) return NULL;
  9735. n = alloc_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, NULL);
  9736. if (!n) return NULL;
  9737. return parse_relative_location_path(n);
  9738. }
  9739. // else clause moved outside of if because of bogus warning 'control may reach end of non-void function being inlined' in gcc 4.0.1
  9740. return parse_relative_location_path(NULL);
  9741. }
  9742. // PathExpr ::= LocationPath
  9743. // | FilterExpr
  9744. // | FilterExpr '/' RelativeLocationPath
  9745. // | FilterExpr '//' RelativeLocationPath
  9746. // UnionExpr ::= PathExpr | UnionExpr '|' PathExpr
  9747. // UnaryExpr ::= UnionExpr | '-' UnaryExpr
  9748. xpath_ast_node* parse_path_or_unary_expression()
  9749. {
  9750. // Clarification.
  9751. // PathExpr begins with either LocationPath or FilterExpr.
  9752. // FilterExpr begins with PrimaryExpr
  9753. // PrimaryExpr begins with '$' in case of it being a variable reference,
  9754. // '(' in case of it being an expression, string literal, number constant or
  9755. // function call.
  9756. if (_lexer.current() == lex_var_ref || _lexer.current() == lex_open_brace ||
  9757. _lexer.current() == lex_quoted_string || _lexer.current() == lex_number ||
  9758. _lexer.current() == lex_string)
  9759. {
  9760. if (_lexer.current() == lex_string)
  9761. {
  9762. // This is either a function call, or not - if not, we shall proceed with location path
  9763. const char_t* state = _lexer.state();
  9764. while (PUGI_IMPL_IS_CHARTYPE(*state, ct_space)) ++state;
  9765. if (*state != '(')
  9766. return parse_location_path();
  9767. // This looks like a function call; however this still can be a node-test. Check it.
  9768. if (parse_node_test_type(_lexer.contents()) != nodetest_none)
  9769. return parse_location_path();
  9770. }
  9771. xpath_ast_node* n = parse_filter_expression();
  9772. if (!n) return NULL;
  9773. if (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash)
  9774. {
  9775. lexeme_t l = _lexer.current();
  9776. _lexer.next();
  9777. if (l == lex_double_slash)
  9778. {
  9779. if (n->rettype() != xpath_type_node_set)
  9780. return error("Step has to be applied to node set");
  9781. n = alloc_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, NULL);
  9782. if (!n) return NULL;
  9783. }
  9784. // select from location path
  9785. return parse_relative_location_path(n);
  9786. }
  9787. return n;
  9788. }
  9789. else if (_lexer.current() == lex_minus)
  9790. {
  9791. _lexer.next();
  9792. // precedence 7+ - only parses union expressions
  9793. xpath_ast_node* n = parse_expression(7);
  9794. if (!n) return NULL;
  9795. return alloc_node(ast_op_negate, xpath_type_number, n);
  9796. }
  9797. else
  9798. {
  9799. return parse_location_path();
  9800. }
  9801. }
  9802. struct binary_op_t
  9803. {
  9804. ast_type_t asttype;
  9805. xpath_value_type rettype;
  9806. int precedence;
  9807. binary_op_t(): asttype(ast_unknown), rettype(xpath_type_none), precedence(0)
  9808. {
  9809. }
  9810. binary_op_t(ast_type_t asttype_, xpath_value_type rettype_, int precedence_): asttype(asttype_), rettype(rettype_), precedence(precedence_)
  9811. {
  9812. }
  9813. static binary_op_t parse(xpath_lexer& lexer)
  9814. {
  9815. switch (lexer.current())
  9816. {
  9817. case lex_string:
  9818. if (lexer.contents() == PUGIXML_TEXT("or"))
  9819. return binary_op_t(ast_op_or, xpath_type_boolean, 1);
  9820. else if (lexer.contents() == PUGIXML_TEXT("and"))
  9821. return binary_op_t(ast_op_and, xpath_type_boolean, 2);
  9822. else if (lexer.contents() == PUGIXML_TEXT("div"))
  9823. return binary_op_t(ast_op_divide, xpath_type_number, 6);
  9824. else if (lexer.contents() == PUGIXML_TEXT("mod"))
  9825. return binary_op_t(ast_op_mod, xpath_type_number, 6);
  9826. else
  9827. return binary_op_t();
  9828. case lex_equal:
  9829. return binary_op_t(ast_op_equal, xpath_type_boolean, 3);
  9830. case lex_not_equal:
  9831. return binary_op_t(ast_op_not_equal, xpath_type_boolean, 3);
  9832. case lex_less:
  9833. return binary_op_t(ast_op_less, xpath_type_boolean, 4);
  9834. case lex_greater:
  9835. return binary_op_t(ast_op_greater, xpath_type_boolean, 4);
  9836. case lex_less_or_equal:
  9837. return binary_op_t(ast_op_less_or_equal, xpath_type_boolean, 4);
  9838. case lex_greater_or_equal:
  9839. return binary_op_t(ast_op_greater_or_equal, xpath_type_boolean, 4);
  9840. case lex_plus:
  9841. return binary_op_t(ast_op_add, xpath_type_number, 5);
  9842. case lex_minus:
  9843. return binary_op_t(ast_op_subtract, xpath_type_number, 5);
  9844. case lex_multiply:
  9845. return binary_op_t(ast_op_multiply, xpath_type_number, 6);
  9846. case lex_union:
  9847. return binary_op_t(ast_op_union, xpath_type_node_set, 7);
  9848. default:
  9849. return binary_op_t();
  9850. }
  9851. }
  9852. };
  9853. xpath_ast_node* parse_expression_rec(xpath_ast_node* lhs, int limit)
  9854. {
  9855. binary_op_t op = binary_op_t::parse(_lexer);
  9856. while (op.asttype != ast_unknown && op.precedence >= limit)
  9857. {
  9858. _lexer.next();
  9859. if (++_depth > xpath_ast_depth_limit)
  9860. return error_rec();
  9861. xpath_ast_node* rhs = parse_path_or_unary_expression();
  9862. if (!rhs) return NULL;
  9863. binary_op_t nextop = binary_op_t::parse(_lexer);
  9864. while (nextop.asttype != ast_unknown && nextop.precedence > op.precedence)
  9865. {
  9866. rhs = parse_expression_rec(rhs, nextop.precedence);
  9867. if (!rhs) return NULL;
  9868. nextop = binary_op_t::parse(_lexer);
  9869. }
  9870. if (op.asttype == ast_op_union && (lhs->rettype() != xpath_type_node_set || rhs->rettype() != xpath_type_node_set))
  9871. return error("Union operator has to be applied to node sets");
  9872. lhs = alloc_node(op.asttype, op.rettype, lhs, rhs);
  9873. if (!lhs) return NULL;
  9874. op = binary_op_t::parse(_lexer);
  9875. }
  9876. return lhs;
  9877. }
  9878. // Expr ::= OrExpr
  9879. // OrExpr ::= AndExpr | OrExpr 'or' AndExpr
  9880. // AndExpr ::= EqualityExpr | AndExpr 'and' EqualityExpr
  9881. // EqualityExpr ::= RelationalExpr
  9882. // | EqualityExpr '=' RelationalExpr
  9883. // | EqualityExpr '!=' RelationalExpr
  9884. // RelationalExpr ::= AdditiveExpr
  9885. // | RelationalExpr '<' AdditiveExpr
  9886. // | RelationalExpr '>' AdditiveExpr
  9887. // | RelationalExpr '<=' AdditiveExpr
  9888. // | RelationalExpr '>=' AdditiveExpr
  9889. // AdditiveExpr ::= MultiplicativeExpr
  9890. // | AdditiveExpr '+' MultiplicativeExpr
  9891. // | AdditiveExpr '-' MultiplicativeExpr
  9892. // MultiplicativeExpr ::= UnaryExpr
  9893. // | MultiplicativeExpr '*' UnaryExpr
  9894. // | MultiplicativeExpr 'div' UnaryExpr
  9895. // | MultiplicativeExpr 'mod' UnaryExpr
  9896. xpath_ast_node* parse_expression(int limit = 0)
  9897. {
  9898. size_t old_depth = _depth;
  9899. if (++_depth > xpath_ast_depth_limit)
  9900. return error_rec();
  9901. xpath_ast_node* n = parse_path_or_unary_expression();
  9902. if (!n) return NULL;
  9903. n = parse_expression_rec(n, limit);
  9904. _depth = old_depth;
  9905. return n;
  9906. }
  9907. xpath_parser(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result): _alloc(alloc), _lexer(query), _query(query), _variables(variables), _result(result), _depth(0)
  9908. {
  9909. }
  9910. xpath_ast_node* parse()
  9911. {
  9912. xpath_ast_node* n = parse_expression();
  9913. if (!n) return NULL;
  9914. assert(_depth == 0);
  9915. // check if there are unparsed tokens left
  9916. if (_lexer.current() != lex_eof)
  9917. return error("Incorrect query");
  9918. return n;
  9919. }
  9920. static xpath_ast_node* parse(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result)
  9921. {
  9922. xpath_parser parser(query, variables, alloc, result);
  9923. return parser.parse();
  9924. }
  9925. };
  9926. struct xpath_query_impl
  9927. {
  9928. static xpath_query_impl* create()
  9929. {
  9930. void* memory = xml_memory::allocate(sizeof(xpath_query_impl));
  9931. if (!memory) return NULL;
  9932. return new (memory) xpath_query_impl();
  9933. }
  9934. static void destroy(xpath_query_impl* impl)
  9935. {
  9936. // free all allocated pages
  9937. impl->alloc.release();
  9938. // free allocator memory (with the first page)
  9939. xml_memory::deallocate(impl);
  9940. }
  9941. xpath_query_impl(): root(NULL), alloc(&block, &oom), oom(false)
  9942. {
  9943. block.next = NULL;
  9944. block.capacity = sizeof(block.data);
  9945. }
  9946. xpath_ast_node* root;
  9947. xpath_allocator alloc;
  9948. xpath_memory_block block;
  9949. bool oom;
  9950. };
  9951. PUGI_IMPL_FN impl::xpath_ast_node* evaluate_node_set_prepare(xpath_query_impl* impl)
  9952. {
  9953. if (!impl) return NULL;
  9954. if (impl->root->rettype() != xpath_type_node_set)
  9955. {
  9956. #ifdef PUGIXML_NO_EXCEPTIONS
  9957. return 0;
  9958. #else
  9959. xpath_parse_result res;
  9960. res.error = "Expression does not evaluate to node set";
  9961. throw xpath_exception(res);
  9962. #endif
  9963. }
  9964. return impl->root;
  9965. }
  9966. PUGI_IMPL_NS_END
  9967. namespace pugi
  9968. {
  9969. #ifndef PUGIXML_NO_EXCEPTIONS
  9970. PUGI_IMPL_FN xpath_exception::xpath_exception(const xpath_parse_result& result_): _result(result_)
  9971. {
  9972. assert(_result.error);
  9973. }
  9974. PUGI_IMPL_FN const char* xpath_exception::what() const PUGIXML_NOEXCEPT
  9975. {
  9976. return _result.error;
  9977. }
  9978. PUGI_IMPL_FN const xpath_parse_result& xpath_exception::result() const
  9979. {
  9980. return _result;
  9981. }
  9982. #endif
  9983. PUGI_IMPL_FN xpath_node::xpath_node()
  9984. {
  9985. }
  9986. PUGI_IMPL_FN xpath_node::xpath_node(const xml_node& node_): _node(node_)
  9987. {
  9988. }
  9989. PUGI_IMPL_FN xpath_node::xpath_node(const xml_attribute& attribute_, const xml_node& parent_): _node(attribute_ ? parent_ : xml_node()), _attribute(attribute_)
  9990. {
  9991. }
  9992. PUGI_IMPL_FN xml_node xpath_node::node() const
  9993. {
  9994. return _attribute ? xml_node() : _node;
  9995. }
  9996. PUGI_IMPL_FN xml_attribute xpath_node::attribute() const
  9997. {
  9998. return _attribute;
  9999. }
  10000. PUGI_IMPL_FN xml_node xpath_node::parent() const
  10001. {
  10002. return _attribute ? _node : _node.parent();
  10003. }
  10004. PUGI_IMPL_FN static void unspecified_bool_xpath_node(xpath_node***)
  10005. {
  10006. }
  10007. PUGI_IMPL_FN xpath_node::operator xpath_node::unspecified_bool_type() const
  10008. {
  10009. return (_node || _attribute) ? unspecified_bool_xpath_node : NULL;
  10010. }
  10011. PUGI_IMPL_FN bool xpath_node::operator!() const
  10012. {
  10013. return !(_node || _attribute);
  10014. }
  10015. PUGI_IMPL_FN bool xpath_node::operator==(const xpath_node& n) const
  10016. {
  10017. return _node == n._node && _attribute == n._attribute;
  10018. }
  10019. PUGI_IMPL_FN bool xpath_node::operator!=(const xpath_node& n) const
  10020. {
  10021. return _node != n._node || _attribute != n._attribute;
  10022. }
  10023. #ifdef __BORLANDC__
  10024. PUGI_IMPL_FN bool operator&&(const xpath_node& lhs, bool rhs)
  10025. {
  10026. return (bool)lhs && rhs;
  10027. }
  10028. PUGI_IMPL_FN bool operator||(const xpath_node& lhs, bool rhs)
  10029. {
  10030. return (bool)lhs || rhs;
  10031. }
  10032. #endif
  10033. PUGI_IMPL_FN void xpath_node_set::_assign(const_iterator begin_, const_iterator end_, type_t type_)
  10034. {
  10035. assert(begin_ <= end_);
  10036. size_t size_ = static_cast<size_t>(end_ - begin_);
  10037. // use internal buffer for 0 or 1 elements, heap buffer otherwise
  10038. xpath_node* storage = (size_ <= 1) ? _storage : static_cast<xpath_node*>(impl::xml_memory::allocate(size_ * sizeof(xpath_node)));
  10039. if (!storage)
  10040. {
  10041. #ifdef PUGIXML_NO_EXCEPTIONS
  10042. return;
  10043. #else
  10044. throw std::bad_alloc();
  10045. #endif
  10046. }
  10047. // deallocate old buffer
  10048. if (_begin != _storage)
  10049. impl::xml_memory::deallocate(_begin);
  10050. // size check is necessary because for begin_ = end_ = nullptr, memcpy is UB
  10051. if (size_)
  10052. memcpy(storage, begin_, size_ * sizeof(xpath_node));
  10053. _begin = storage;
  10054. _end = storage + size_;
  10055. _type = type_;
  10056. }
  10057. #ifdef PUGIXML_HAS_MOVE
  10058. PUGI_IMPL_FN void xpath_node_set::_move(xpath_node_set& rhs) PUGIXML_NOEXCEPT
  10059. {
  10060. _type = rhs._type;
  10061. _storage[0] = rhs._storage[0];
  10062. _begin = (rhs._begin == rhs._storage) ? _storage : rhs._begin;
  10063. _end = _begin + (rhs._end - rhs._begin);
  10064. rhs._type = type_unsorted;
  10065. rhs._begin = rhs._storage;
  10066. rhs._end = rhs._storage;
  10067. }
  10068. #endif
  10069. PUGI_IMPL_FN xpath_node_set::xpath_node_set(): _type(type_unsorted), _begin(_storage), _end(_storage)
  10070. {
  10071. }
  10072. PUGI_IMPL_FN xpath_node_set::xpath_node_set(const_iterator begin_, const_iterator end_, type_t type_): _type(type_unsorted), _begin(_storage), _end(_storage)
  10073. {
  10074. _assign(begin_, end_, type_);
  10075. }
  10076. PUGI_IMPL_FN xpath_node_set::~xpath_node_set()
  10077. {
  10078. if (_begin != _storage)
  10079. impl::xml_memory::deallocate(_begin);
  10080. }
  10081. PUGI_IMPL_FN xpath_node_set::xpath_node_set(const xpath_node_set& ns): _type(type_unsorted), _begin(_storage), _end(_storage)
  10082. {
  10083. _assign(ns._begin, ns._end, ns._type);
  10084. }
  10085. PUGI_IMPL_FN xpath_node_set& xpath_node_set::operator=(const xpath_node_set& ns)
  10086. {
  10087. if (this == &ns) return *this;
  10088. _assign(ns._begin, ns._end, ns._type);
  10089. return *this;
  10090. }
  10091. #ifdef PUGIXML_HAS_MOVE
  10092. PUGI_IMPL_FN xpath_node_set::xpath_node_set(xpath_node_set&& rhs) PUGIXML_NOEXCEPT: _type(type_unsorted), _begin(_storage), _end(_storage)
  10093. {
  10094. _move(rhs);
  10095. }
  10096. PUGI_IMPL_FN xpath_node_set& xpath_node_set::operator=(xpath_node_set&& rhs) PUGIXML_NOEXCEPT
  10097. {
  10098. if (this == &rhs) return *this;
  10099. if (_begin != _storage)
  10100. impl::xml_memory::deallocate(_begin);
  10101. _move(rhs);
  10102. return *this;
  10103. }
  10104. #endif
  10105. PUGI_IMPL_FN xpath_node_set::type_t xpath_node_set::type() const
  10106. {
  10107. return _type;
  10108. }
  10109. PUGI_IMPL_FN size_t xpath_node_set::size() const
  10110. {
  10111. return _end - _begin;
  10112. }
  10113. PUGI_IMPL_FN bool xpath_node_set::empty() const
  10114. {
  10115. return _begin == _end;
  10116. }
  10117. PUGI_IMPL_FN const xpath_node& xpath_node_set::operator[](size_t index) const
  10118. {
  10119. assert(index < size());
  10120. return _begin[index];
  10121. }
  10122. PUGI_IMPL_FN xpath_node_set::const_iterator xpath_node_set::begin() const
  10123. {
  10124. return _begin;
  10125. }
  10126. PUGI_IMPL_FN xpath_node_set::const_iterator xpath_node_set::end() const
  10127. {
  10128. return _end;
  10129. }
  10130. PUGI_IMPL_FN void xpath_node_set::sort(bool reverse)
  10131. {
  10132. _type = impl::xpath_sort(_begin, _end, _type, reverse);
  10133. }
  10134. PUGI_IMPL_FN xpath_node xpath_node_set::first() const
  10135. {
  10136. return impl::xpath_first(_begin, _end, _type);
  10137. }
  10138. PUGI_IMPL_FN xpath_parse_result::xpath_parse_result(): error("Internal error"), offset(0)
  10139. {
  10140. }
  10141. PUGI_IMPL_FN xpath_parse_result::operator bool() const
  10142. {
  10143. return error == NULL;
  10144. }
  10145. PUGI_IMPL_FN const char* xpath_parse_result::description() const
  10146. {
  10147. return error ? error : "No error";
  10148. }
  10149. PUGI_IMPL_FN xpath_variable::xpath_variable(xpath_value_type type_): _type(type_), _next(NULL)
  10150. {
  10151. }
  10152. PUGI_IMPL_FN const char_t* xpath_variable::name() const
  10153. {
  10154. switch (_type)
  10155. {
  10156. case xpath_type_node_set:
  10157. return static_cast<const impl::xpath_variable_node_set*>(this)->name;
  10158. case xpath_type_number:
  10159. return static_cast<const impl::xpath_variable_number*>(this)->name;
  10160. case xpath_type_string:
  10161. return static_cast<const impl::xpath_variable_string*>(this)->name;
  10162. case xpath_type_boolean:
  10163. return static_cast<const impl::xpath_variable_boolean*>(this)->name;
  10164. default:
  10165. assert(false && "Invalid variable type"); // unreachable
  10166. return NULL;
  10167. }
  10168. }
  10169. PUGI_IMPL_FN xpath_value_type xpath_variable::type() const
  10170. {
  10171. return _type;
  10172. }
  10173. PUGI_IMPL_FN bool xpath_variable::get_boolean() const
  10174. {
  10175. return (_type == xpath_type_boolean) ? static_cast<const impl::xpath_variable_boolean*>(this)->value : false;
  10176. }
  10177. PUGI_IMPL_FN double xpath_variable::get_number() const
  10178. {
  10179. return (_type == xpath_type_number) ? static_cast<const impl::xpath_variable_number*>(this)->value : impl::gen_nan();
  10180. }
  10181. PUGI_IMPL_FN const char_t* xpath_variable::get_string() const
  10182. {
  10183. const char_t* value = (_type == xpath_type_string) ? static_cast<const impl::xpath_variable_string*>(this)->value : NULL;
  10184. return value ? value : PUGIXML_TEXT("");
  10185. }
  10186. PUGI_IMPL_FN const xpath_node_set& xpath_variable::get_node_set() const
  10187. {
  10188. if (_type == xpath_type_node_set)
  10189. return static_cast<const impl::xpath_variable_node_set*>(this)->value;
  10190. static const xpath_node_set dummy_node_set;
  10191. return dummy_node_set;
  10192. }
  10193. PUGI_IMPL_FN bool xpath_variable::set(bool value)
  10194. {
  10195. if (_type != xpath_type_boolean) return false;
  10196. static_cast<impl::xpath_variable_boolean*>(this)->value = value;
  10197. return true;
  10198. }
  10199. PUGI_IMPL_FN bool xpath_variable::set(double value)
  10200. {
  10201. if (_type != xpath_type_number) return false;
  10202. static_cast<impl::xpath_variable_number*>(this)->value = value;
  10203. return true;
  10204. }
  10205. PUGI_IMPL_FN bool xpath_variable::set(const char_t* value)
  10206. {
  10207. if (_type != xpath_type_string) return false;
  10208. impl::xpath_variable_string* var = static_cast<impl::xpath_variable_string*>(this);
  10209. // duplicate string
  10210. size_t size = (impl::strlength(value) + 1) * sizeof(char_t);
  10211. char_t* copy = static_cast<char_t*>(impl::xml_memory::allocate(size));
  10212. if (!copy) return false;
  10213. memcpy(copy, value, size);
  10214. // replace old string
  10215. if (var->value) impl::xml_memory::deallocate(var->value);
  10216. var->value = copy;
  10217. return true;
  10218. }
  10219. PUGI_IMPL_FN bool xpath_variable::set(const xpath_node_set& value)
  10220. {
  10221. if (_type != xpath_type_node_set) return false;
  10222. static_cast<impl::xpath_variable_node_set*>(this)->value = value;
  10223. return true;
  10224. }
  10225. PUGI_IMPL_FN xpath_variable_set::xpath_variable_set()
  10226. {
  10227. for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
  10228. _data[i] = NULL;
  10229. }
  10230. PUGI_IMPL_FN xpath_variable_set::~xpath_variable_set()
  10231. {
  10232. for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
  10233. _destroy(_data[i]);
  10234. }
  10235. PUGI_IMPL_FN xpath_variable_set::xpath_variable_set(const xpath_variable_set& rhs)
  10236. {
  10237. for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
  10238. _data[i] = NULL;
  10239. _assign(rhs);
  10240. }
  10241. PUGI_IMPL_FN xpath_variable_set& xpath_variable_set::operator=(const xpath_variable_set& rhs)
  10242. {
  10243. if (this == &rhs) return *this;
  10244. _assign(rhs);
  10245. return *this;
  10246. }
  10247. #ifdef PUGIXML_HAS_MOVE
  10248. PUGI_IMPL_FN xpath_variable_set::xpath_variable_set(xpath_variable_set&& rhs) PUGIXML_NOEXCEPT
  10249. {
  10250. for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
  10251. {
  10252. _data[i] = rhs._data[i];
  10253. rhs._data[i] = NULL;
  10254. }
  10255. }
  10256. PUGI_IMPL_FN xpath_variable_set& xpath_variable_set::operator=(xpath_variable_set&& rhs) PUGIXML_NOEXCEPT
  10257. {
  10258. for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
  10259. {
  10260. _destroy(_data[i]);
  10261. _data[i] = rhs._data[i];
  10262. rhs._data[i] = NULL;
  10263. }
  10264. return *this;
  10265. }
  10266. #endif
  10267. PUGI_IMPL_FN void xpath_variable_set::_assign(const xpath_variable_set& rhs)
  10268. {
  10269. xpath_variable_set temp;
  10270. for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
  10271. if (rhs._data[i] && !_clone(rhs._data[i], &temp._data[i]))
  10272. return;
  10273. _swap(temp);
  10274. }
  10275. PUGI_IMPL_FN void xpath_variable_set::_swap(xpath_variable_set& rhs)
  10276. {
  10277. for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
  10278. {
  10279. xpath_variable* chain = _data[i];
  10280. _data[i] = rhs._data[i];
  10281. rhs._data[i] = chain;
  10282. }
  10283. }
  10284. PUGI_IMPL_FN xpath_variable* xpath_variable_set::_find(const char_t* name) const
  10285. {
  10286. const size_t hash_size = sizeof(_data) / sizeof(_data[0]);
  10287. size_t hash = impl::hash_string(name) % hash_size;
  10288. // look for existing variable
  10289. for (xpath_variable* var = _data[hash]; var; var = var->_next)
  10290. {
  10291. const char_t* vn = var->name();
  10292. if (vn && impl::strequal(vn, name))
  10293. return var;
  10294. }
  10295. return NULL;
  10296. }
  10297. PUGI_IMPL_FN bool xpath_variable_set::_clone(xpath_variable* var, xpath_variable** out_result)
  10298. {
  10299. xpath_variable* last = NULL;
  10300. while (var)
  10301. {
  10302. // allocate storage for new variable
  10303. xpath_variable* nvar = impl::new_xpath_variable(var->_type, var->name());
  10304. if (!nvar) return false;
  10305. // link the variable to the result immediately to handle failures gracefully
  10306. if (last)
  10307. last->_next = nvar;
  10308. else
  10309. *out_result = nvar;
  10310. last = nvar;
  10311. // copy the value; this can fail due to out-of-memory conditions
  10312. if (!impl::copy_xpath_variable(nvar, var)) return false;
  10313. var = var->_next;
  10314. }
  10315. return true;
  10316. }
  10317. PUGI_IMPL_FN void xpath_variable_set::_destroy(xpath_variable* var)
  10318. {
  10319. while (var)
  10320. {
  10321. xpath_variable* next = var->_next;
  10322. impl::delete_xpath_variable(var->_type, var);
  10323. var = next;
  10324. }
  10325. }
  10326. PUGI_IMPL_FN xpath_variable* xpath_variable_set::add(const char_t* name, xpath_value_type type)
  10327. {
  10328. const size_t hash_size = sizeof(_data) / sizeof(_data[0]);
  10329. size_t hash = impl::hash_string(name) % hash_size;
  10330. // look for existing variable
  10331. for (xpath_variable* var = _data[hash]; var; var = var->_next)
  10332. {
  10333. const char_t* vn = var->name();
  10334. if (vn && impl::strequal(vn, name))
  10335. return var->type() == type ? var : NULL;
  10336. }
  10337. // add new variable
  10338. xpath_variable* result = impl::new_xpath_variable(type, name);
  10339. if (result)
  10340. {
  10341. result->_next = _data[hash];
  10342. _data[hash] = result;
  10343. }
  10344. return result;
  10345. }
  10346. PUGI_IMPL_FN bool xpath_variable_set::set(const char_t* name, bool value)
  10347. {
  10348. xpath_variable* var = add(name, xpath_type_boolean);
  10349. return var ? var->set(value) : false;
  10350. }
  10351. PUGI_IMPL_FN bool xpath_variable_set::set(const char_t* name, double value)
  10352. {
  10353. xpath_variable* var = add(name, xpath_type_number);
  10354. return var ? var->set(value) : false;
  10355. }
  10356. PUGI_IMPL_FN bool xpath_variable_set::set(const char_t* name, const char_t* value)
  10357. {
  10358. xpath_variable* var = add(name, xpath_type_string);
  10359. return var ? var->set(value) : false;
  10360. }
  10361. PUGI_IMPL_FN bool xpath_variable_set::set(const char_t* name, const xpath_node_set& value)
  10362. {
  10363. xpath_variable* var = add(name, xpath_type_node_set);
  10364. return var ? var->set(value) : false;
  10365. }
  10366. PUGI_IMPL_FN xpath_variable* xpath_variable_set::get(const char_t* name)
  10367. {
  10368. return _find(name);
  10369. }
  10370. PUGI_IMPL_FN const xpath_variable* xpath_variable_set::get(const char_t* name) const
  10371. {
  10372. return _find(name);
  10373. }
  10374. PUGI_IMPL_FN xpath_query::xpath_query(const char_t* query, xpath_variable_set* variables): _impl(NULL)
  10375. {
  10376. impl::xpath_query_impl* qimpl = impl::xpath_query_impl::create();
  10377. if (!qimpl)
  10378. {
  10379. #ifdef PUGIXML_NO_EXCEPTIONS
  10380. _result.error = "Out of memory";
  10381. #else
  10382. throw std::bad_alloc();
  10383. #endif
  10384. }
  10385. else
  10386. {
  10387. using impl::auto_deleter; // MSVC7 workaround
  10388. auto_deleter<impl::xpath_query_impl> impl(qimpl, impl::xpath_query_impl::destroy);
  10389. qimpl->root = impl::xpath_parser::parse(query, variables, &qimpl->alloc, &_result);
  10390. if (qimpl->root)
  10391. {
  10392. qimpl->root->optimize(&qimpl->alloc);
  10393. _impl = impl.release();
  10394. _result.error = NULL;
  10395. }
  10396. else
  10397. {
  10398. #ifdef PUGIXML_NO_EXCEPTIONS
  10399. if (qimpl->oom) _result.error = "Out of memory";
  10400. #else
  10401. if (qimpl->oom) throw std::bad_alloc();
  10402. throw xpath_exception(_result);
  10403. #endif
  10404. }
  10405. }
  10406. }
  10407. PUGI_IMPL_FN xpath_query::xpath_query(): _impl(NULL)
  10408. {
  10409. }
  10410. PUGI_IMPL_FN xpath_query::~xpath_query()
  10411. {
  10412. if (_impl)
  10413. impl::xpath_query_impl::destroy(static_cast<impl::xpath_query_impl*>(_impl));
  10414. }
  10415. #ifdef PUGIXML_HAS_MOVE
  10416. PUGI_IMPL_FN xpath_query::xpath_query(xpath_query&& rhs) PUGIXML_NOEXCEPT
  10417. {
  10418. _impl = rhs._impl;
  10419. _result = rhs._result;
  10420. rhs._impl = NULL;
  10421. rhs._result = xpath_parse_result();
  10422. }
  10423. PUGI_IMPL_FN xpath_query& xpath_query::operator=(xpath_query&& rhs) PUGIXML_NOEXCEPT
  10424. {
  10425. if (this == &rhs) return *this;
  10426. if (_impl)
  10427. impl::xpath_query_impl::destroy(static_cast<impl::xpath_query_impl*>(_impl));
  10428. _impl = rhs._impl;
  10429. _result = rhs._result;
  10430. rhs._impl = NULL;
  10431. rhs._result = xpath_parse_result();
  10432. return *this;
  10433. }
  10434. #endif
  10435. PUGI_IMPL_FN xpath_value_type xpath_query::return_type() const
  10436. {
  10437. if (!_impl) return xpath_type_none;
  10438. return static_cast<impl::xpath_query_impl*>(_impl)->root->rettype();
  10439. }
  10440. PUGI_IMPL_FN bool xpath_query::evaluate_boolean(const xpath_node& n) const
  10441. {
  10442. if (!_impl) return false;
  10443. impl::xpath_context c(n, 1, 1);
  10444. impl::xpath_stack_data sd;
  10445. bool r = static_cast<impl::xpath_query_impl*>(_impl)->root->eval_boolean(c, sd.stack);
  10446. if (sd.oom)
  10447. {
  10448. #ifdef PUGIXML_NO_EXCEPTIONS
  10449. return false;
  10450. #else
  10451. throw std::bad_alloc();
  10452. #endif
  10453. }
  10454. return r;
  10455. }
  10456. PUGI_IMPL_FN double xpath_query::evaluate_number(const xpath_node& n) const
  10457. {
  10458. if (!_impl) return impl::gen_nan();
  10459. impl::xpath_context c(n, 1, 1);
  10460. impl::xpath_stack_data sd;
  10461. double r = static_cast<impl::xpath_query_impl*>(_impl)->root->eval_number(c, sd.stack);
  10462. if (sd.oom)
  10463. {
  10464. #ifdef PUGIXML_NO_EXCEPTIONS
  10465. return impl::gen_nan();
  10466. #else
  10467. throw std::bad_alloc();
  10468. #endif
  10469. }
  10470. return r;
  10471. }
  10472. #ifndef PUGIXML_NO_STL
  10473. PUGI_IMPL_FN string_t xpath_query::evaluate_string(const xpath_node& n) const
  10474. {
  10475. if (!_impl) return string_t();
  10476. impl::xpath_context c(n, 1, 1);
  10477. impl::xpath_stack_data sd;
  10478. impl::xpath_string r = static_cast<impl::xpath_query_impl*>(_impl)->root->eval_string(c, sd.stack);
  10479. if (sd.oom)
  10480. {
  10481. #ifdef PUGIXML_NO_EXCEPTIONS
  10482. return string_t();
  10483. #else
  10484. throw std::bad_alloc();
  10485. #endif
  10486. }
  10487. return string_t(r.c_str(), r.length());
  10488. }
  10489. #endif
  10490. PUGI_IMPL_FN size_t xpath_query::evaluate_string(char_t* buffer, size_t capacity, const xpath_node& n) const
  10491. {
  10492. impl::xpath_context c(n, 1, 1);
  10493. impl::xpath_stack_data sd;
  10494. impl::xpath_string r = _impl ? static_cast<impl::xpath_query_impl*>(_impl)->root->eval_string(c, sd.stack) : impl::xpath_string();
  10495. if (sd.oom)
  10496. {
  10497. #ifdef PUGIXML_NO_EXCEPTIONS
  10498. r = impl::xpath_string();
  10499. #else
  10500. throw std::bad_alloc();
  10501. #endif
  10502. }
  10503. size_t full_size = r.length() + 1;
  10504. if (capacity > 0)
  10505. {
  10506. size_t size = (full_size < capacity) ? full_size : capacity;
  10507. assert(size > 0);
  10508. memcpy(buffer, r.c_str(), (size - 1) * sizeof(char_t));
  10509. buffer[size - 1] = 0;
  10510. }
  10511. return full_size;
  10512. }
  10513. PUGI_IMPL_FN xpath_node_set xpath_query::evaluate_node_set(const xpath_node& n) const
  10514. {
  10515. impl::xpath_ast_node* root = impl::evaluate_node_set_prepare(static_cast<impl::xpath_query_impl*>(_impl));
  10516. if (!root) return xpath_node_set();
  10517. impl::xpath_context c(n, 1, 1);
  10518. impl::xpath_stack_data sd;
  10519. impl::xpath_node_set_raw r = root->eval_node_set(c, sd.stack, impl::nodeset_eval_all);
  10520. if (sd.oom)
  10521. {
  10522. #ifdef PUGIXML_NO_EXCEPTIONS
  10523. return xpath_node_set();
  10524. #else
  10525. throw std::bad_alloc();
  10526. #endif
  10527. }
  10528. return xpath_node_set(r.begin(), r.end(), r.type());
  10529. }
  10530. PUGI_IMPL_FN xpath_node xpath_query::evaluate_node(const xpath_node& n) const
  10531. {
  10532. impl::xpath_ast_node* root = impl::evaluate_node_set_prepare(static_cast<impl::xpath_query_impl*>(_impl));
  10533. if (!root) return xpath_node();
  10534. impl::xpath_context c(n, 1, 1);
  10535. impl::xpath_stack_data sd;
  10536. impl::xpath_node_set_raw r = root->eval_node_set(c, sd.stack, impl::nodeset_eval_first);
  10537. if (sd.oom)
  10538. {
  10539. #ifdef PUGIXML_NO_EXCEPTIONS
  10540. return xpath_node();
  10541. #else
  10542. throw std::bad_alloc();
  10543. #endif
  10544. }
  10545. return r.first();
  10546. }
  10547. PUGI_IMPL_FN const xpath_parse_result& xpath_query::result() const
  10548. {
  10549. return _result;
  10550. }
  10551. PUGI_IMPL_FN static void unspecified_bool_xpath_query(xpath_query***)
  10552. {
  10553. }
  10554. PUGI_IMPL_FN xpath_query::operator xpath_query::unspecified_bool_type() const
  10555. {
  10556. return _impl ? unspecified_bool_xpath_query : NULL;
  10557. }
  10558. PUGI_IMPL_FN bool xpath_query::operator!() const
  10559. {
  10560. return !_impl;
  10561. }
  10562. PUGI_IMPL_FN xpath_node xml_node::select_node(const char_t* query, xpath_variable_set* variables) const
  10563. {
  10564. xpath_query q(query, variables);
  10565. return q.evaluate_node(*this);
  10566. }
  10567. PUGI_IMPL_FN xpath_node xml_node::select_node(const xpath_query& query) const
  10568. {
  10569. return query.evaluate_node(*this);
  10570. }
  10571. PUGI_IMPL_FN xpath_node_set xml_node::select_nodes(const char_t* query, xpath_variable_set* variables) const
  10572. {
  10573. xpath_query q(query, variables);
  10574. return q.evaluate_node_set(*this);
  10575. }
  10576. PUGI_IMPL_FN xpath_node_set xml_node::select_nodes(const xpath_query& query) const
  10577. {
  10578. return query.evaluate_node_set(*this);
  10579. }
  10580. PUGI_IMPL_FN xpath_node xml_node::select_single_node(const char_t* query, xpath_variable_set* variables) const
  10581. {
  10582. xpath_query q(query, variables);
  10583. return q.evaluate_node(*this);
  10584. }
  10585. PUGI_IMPL_FN xpath_node xml_node::select_single_node(const xpath_query& query) const
  10586. {
  10587. return query.evaluate_node(*this);
  10588. }
  10589. }
  10590. #endif
  10591. #ifdef __BORLANDC__
  10592. # pragma option pop
  10593. #endif
  10594. #if defined(_MSC_VER) && defined(__c2__)
  10595. # pragma clang diagnostic pop
  10596. #endif
  10597. #if defined(__clang__)
  10598. # pragma clang diagnostic pop
  10599. #endif
  10600. // Intel C++ does not properly keep warning state for function templates,
  10601. // so popping warning state at the end of translation unit leads to warnings in the middle.
  10602. #if defined(_MSC_VER) && !defined(__INTEL_COMPILER)
  10603. # pragma warning(pop)
  10604. #endif
  10605. // Undefine all local macros (makes sure we're not leaking macros in header-only mode)
  10606. #undef PUGI_IMPL_NO_INLINE
  10607. #undef PUGI_IMPL_UNLIKELY
  10608. #undef PUGI_IMPL_STATIC_ASSERT
  10609. #undef PUGI_IMPL_DMC_VOLATILE
  10610. #undef PUGI_IMPL_UNSIGNED_OVERFLOW
  10611. #undef PUGI_IMPL_MSVC_CRT_VERSION
  10612. #undef PUGI_IMPL_SNPRINTF
  10613. #undef PUGI_IMPL_NS_BEGIN
  10614. #undef PUGI_IMPL_NS_END
  10615. #undef PUGI_IMPL_FN
  10616. #undef PUGI_IMPL_FN_NO_INLINE
  10617. #undef PUGI_IMPL_GETHEADER_IMPL
  10618. #undef PUGI_IMPL_GETPAGE_IMPL
  10619. #undef PUGI_IMPL_GETPAGE
  10620. #undef PUGI_IMPL_NODETYPE
  10621. #undef PUGI_IMPL_IS_CHARTYPE_IMPL
  10622. #undef PUGI_IMPL_IS_CHARTYPE
  10623. #undef PUGI_IMPL_IS_CHARTYPEX
  10624. #undef PUGI_IMPL_ENDSWITH
  10625. #undef PUGI_IMPL_SKIPWS
  10626. #undef PUGI_IMPL_OPTSET
  10627. #undef PUGI_IMPL_PUSHNODE
  10628. #undef PUGI_IMPL_POPNODE
  10629. #undef PUGI_IMPL_SCANFOR
  10630. #undef PUGI_IMPL_SCANWHILE
  10631. #undef PUGI_IMPL_SCANWHILE_UNROLL
  10632. #undef PUGI_IMPL_ENDSEG
  10633. #undef PUGI_IMPL_THROW_ERROR
  10634. #undef PUGI_IMPL_CHECK_ERROR
  10635. #endif
  10636. /**
  10637. * Copyright (c) 2006-2026 Arseny Kapoulkine
  10638. *
  10639. * Permission is hereby granted, free of charge, to any person
  10640. * obtaining a copy of this software and associated documentation
  10641. * files (the "Software"), to deal in the Software without
  10642. * restriction, including without limitation the rights to use,
  10643. * copy, modify, merge, publish, distribute, sublicense, and/or sell
  10644. * copies of the Software, and to permit persons to whom the
  10645. * Software is furnished to do so, subject to the following
  10646. * conditions:
  10647. *
  10648. * The above copyright notice and this permission notice shall be
  10649. * included in all copies or substantial portions of the Software.
  10650. *
  10651. * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  10652. * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
  10653. * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  10654. * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
  10655. * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
  10656. * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  10657. * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
  10658. * OTHER DEALINGS IN THE SOFTWARE.
  10659. */