1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908290929102911291229132914291529162917291829192920292129222923292429252926292729282929293029312932293329342935293629372938293929402941294229432944294529462947294829492950295129522953295429552956295729582959296029612962296329642965296629672968296929702971297229732974297529762977297829792980298129822983298429852986298729882989299029912992299329942995299629972998299930003001300230033004300530063007300830093010301130123013301430153016301730183019302030213022302330243025302630273028302930303031303230333034303530363037303830393040304130423043304430453046304730483049305030513052305330543055305630573058305930603061306230633064306530663067306830693070307130723073307430753076307730783079308030813082308330843085308630873088308930903091309230933094309530963097309830993100310131023103310431053106310731083109311031113112311331143115311631173118311931203121312231233124312531263127312831293130313131323133313431353136313731383139314031413142314331443145314631473148314931503151315231533154315531563157315831593160316131623163316431653166316731683169317031713172317331743175317631773178317931803181318231833184318531863187318831893190319131923193319431953196319731983199320032013202320332043205320632073208320932103211321232133214321532163217321832193220322132223223322432253226322732283229323032313232323332343235323632373238323932403241324232433244324532463247324832493250325132523253325432553256325732583259326032613262326332643265326632673268326932703271327232733274327532763277327832793280328132823283328432853286328732883289329032913292329332943295329632973298329933003301330233033304330533063307330833093310331133123313331433153316331733183319332033213322332333243325332633273328332933303331333233333334333533363337333833393340334133423343334433453346334733483349335033513352335333543355335633573358335933603361336233633364336533663367336833693370337133723373337433753376337733783379338033813382338333843385338633873388338933903391339233933394339533963397339833993400340134023403340434053406340734083409341034113412341334143415341634173418341934203421342234233424342534263427342834293430343134323433343434353436343734383439344034413442344334443445344634473448344934503451345234533454345534563457345834593460346134623463346434653466346734683469347034713472347334743475347634773478347934803481348234833484348534863487348834893490349134923493349434953496349734983499350035013502350335043505350635073508350935103511351235133514351535163517351835193520352135223523352435253526352735283529353035313532353335343535353635373538353935403541354235433544354535463547354835493550355135523553355435553556355735583559356035613562356335643565356635673568356935703571357235733574357535763577357835793580358135823583358435853586358735883589359035913592359335943595359635973598359936003601360236033604360536063607360836093610361136123613361436153616361736183619362036213622362336243625362636273628362936303631363236333634363536363637363836393640364136423643364436453646364736483649365036513652365336543655365636573658365936603661366236633664366536663667366836693670367136723673367436753676367736783679368036813682368336843685368636873688368936903691369236933694369536963697369836993700370137023703370437053706370737083709371037113712371337143715371637173718371937203721372237233724372537263727372837293730373137323733373437353736373737383739374037413742374337443745374637473748374937503751375237533754375537563757375837593760376137623763376437653766376737683769377037713772377337743775377637773778377937803781378237833784378537863787378837893790379137923793379437953796379737983799380038013802380338043805380638073808380938103811381238133814381538163817381838193820382138223823382438253826382738283829383038313832383338343835383638373838383938403841384238433844384538463847384838493850385138523853385438553856385738583859386038613862386338643865386638673868386938703871387238733874387538763877387838793880388138823883388438853886388738883889389038913892389338943895389638973898389939003901390239033904390539063907390839093910391139123913391439153916391739183919392039213922392339243925392639273928392939303931393239333934393539363937393839393940394139423943394439453946394739483949395039513952395339543955395639573958395939603961396239633964396539663967396839693970397139723973397439753976397739783979398039813982398339843985398639873988398939903991399239933994399539963997399839994000400140024003400440054006400740084009401040114012401340144015401640174018401940204021402240234024402540264027402840294030403140324033403440354036403740384039404040414042404340444045404640474048404940504051405240534054405540564057405840594060406140624063406440654066406740684069407040714072407340744075407640774078407940804081408240834084408540864087408840894090409140924093409440954096409740984099410041014102410341044105410641074108410941104111411241134114411541164117411841194120412141224123412441254126412741284129413041314132413341344135413641374138413941404141414241434144414541464147414841494150415141524153415441554156415741584159416041614162416341644165416641674168416941704171417241734174417541764177417841794180418141824183418441854186418741884189419041914192419341944195419641974198419942004201420242034204420542064207420842094210421142124213421442154216421742184219422042214222422342244225422642274228422942304231423242334234423542364237423842394240424142424243424442454246424742484249425042514252425342544255425642574258425942604261426242634264426542664267426842694270427142724273427442754276427742784279428042814282428342844285428642874288428942904291429242934294429542964297429842994300430143024303430443054306430743084309431043114312431343144315431643174318431943204321432243234324432543264327432843294330433143324333433443354336433743384339434043414342434343444345434643474348434943504351435243534354435543564357435843594360436143624363436443654366436743684369437043714372437343744375437643774378437943804381438243834384438543864387438843894390439143924393439443954396439743984399440044014402440344044405440644074408440944104411441244134414441544164417441844194420442144224423442444254426442744284429443044314432443344344435443644374438443944404441444244434444444544464447444844494450445144524453445444554456445744584459446044614462446344644465446644674468446944704471447244734474447544764477447844794480448144824483448444854486448744884489449044914492449344944495449644974498449945004501450245034504450545064507450845094510451145124513451445154516451745184519452045214522452345244525452645274528452945304531453245334534453545364537453845394540454145424543454445454546454745484549455045514552455345544555455645574558455945604561456245634564456545664567456845694570457145724573457445754576457745784579458045814582458345844585458645874588458945904591459245934594459545964597459845994600460146024603460446054606460746084609461046114612461346144615461646174618461946204621462246234624462546264627462846294630463146324633463446354636463746384639464046414642464346444645464646474648464946504651465246534654465546564657465846594660466146624663466446654666466746684669467046714672467346744675467646774678467946804681468246834684468546864687468846894690469146924693469446954696469746984699470047014702470347044705470647074708470947104711471247134714471547164717471847194720472147224723472447254726472747284729473047314732473347344735473647374738473947404741474247434744474547464747474847494750475147524753475447554756475747584759476047614762476347644765476647674768476947704771477247734774477547764777477847794780478147824783478447854786478747884789 |
- <?php
- class HTMLPurifier_Lexer_PH5P extends HTMLPurifier_Lexer_DOMLex
- {
-
- public function tokenizeHTML($html, $config, $context)
- {
- $new_html = $this->normalize($html, $config, $context);
- $new_html = $this->wrapHTML($new_html, $config, $context, false );
- try {
- $parser = new HTML5($new_html);
- $doc = $parser->save();
- } catch (DOMException $e) {
-
- $lexer = new HTMLPurifier_Lexer_DirectLex();
- $context->register('PH5PError', $e);
- return $lexer->tokenizeHTML($html, $config, $context);
- }
- $tokens = array();
- $this->tokenizeDOM(
- $doc->getElementsByTagName('html')->item(0)->
- getElementsByTagName('body')->item(0)
- ,
- $tokens, $config
- );
- return $tokens;
- }
- }
- class HTML5
- {
- private $data;
- private $char;
- private $EOF;
- private $state;
- private $tree;
- private $token;
- private $content_model;
- private $escape = false;
- private $entities = array(
- 'AElig;',
- 'AElig',
- 'AMP;',
- 'AMP',
- 'Aacute;',
- 'Aacute',
- 'Acirc;',
- 'Acirc',
- 'Agrave;',
- 'Agrave',
- 'Alpha;',
- 'Aring;',
- 'Aring',
- 'Atilde;',
- 'Atilde',
- 'Auml;',
- 'Auml',
- 'Beta;',
- 'COPY;',
- 'COPY',
- 'Ccedil;',
- 'Ccedil',
- 'Chi;',
- 'Dagger;',
- 'Delta;',
- 'ETH;',
- 'ETH',
- 'Eacute;',
- 'Eacute',
- 'Ecirc;',
- 'Ecirc',
- 'Egrave;',
- 'Egrave',
- 'Epsilon;',
- 'Eta;',
- 'Euml;',
- 'Euml',
- 'GT;',
- 'GT',
- 'Gamma;',
- 'Iacute;',
- 'Iacute',
- 'Icirc;',
- 'Icirc',
- 'Igrave;',
- 'Igrave',
- 'Iota;',
- 'Iuml;',
- 'Iuml',
- 'Kappa;',
- 'LT;',
- 'LT',
- 'Lambda;',
- 'Mu;',
- 'Ntilde;',
- 'Ntilde',
- 'Nu;',
- 'OElig;',
- 'Oacute;',
- 'Oacute',
- 'Ocirc;',
- 'Ocirc',
- 'Ograve;',
- 'Ograve',
- 'Omega;',
- 'Omicron;',
- 'Oslash;',
- 'Oslash',
- 'Otilde;',
- 'Otilde',
- 'Ouml;',
- 'Ouml',
- 'Phi;',
- 'Pi;',
- 'Prime;',
- 'Psi;',
- 'QUOT;',
- 'QUOT',
- 'REG;',
- 'REG',
- 'Rho;',
- 'Scaron;',
- 'Sigma;',
- 'THORN;',
- 'THORN',
- 'TRADE;',
- 'Tau;',
- 'Theta;',
- 'Uacute;',
- 'Uacute',
- 'Ucirc;',
- 'Ucirc',
- 'Ugrave;',
- 'Ugrave',
- 'Upsilon;',
- 'Uuml;',
- 'Uuml',
- 'Xi;',
- 'Yacute;',
- 'Yacute',
- 'Yuml;',
- 'Zeta;',
- 'aacute;',
- 'aacute',
- 'acirc;',
- 'acirc',
- 'acute;',
- 'acute',
- 'aelig;',
- 'aelig',
- 'agrave;',
- 'agrave',
- 'alefsym;',
- 'alpha;',
- 'amp;',
- 'amp',
- 'and;',
- 'ang;',
- 'apos;',
- 'aring;',
- 'aring',
- 'asymp;',
- 'atilde;',
- 'atilde',
- 'auml;',
- 'auml',
- 'bdquo;',
- 'beta;',
- 'brvbar;',
- 'brvbar',
- 'bull;',
- 'cap;',
- 'ccedil;',
- 'ccedil',
- 'cedil;',
- 'cedil',
- 'cent;',
- 'cent',
- 'chi;',
- 'circ;',
- 'clubs;',
- 'cong;',
- 'copy;',
- 'copy',
- 'crarr;',
- 'cup;',
- 'curren;',
- 'curren',
- 'dArr;',
- 'dagger;',
- 'darr;',
- 'deg;',
- 'deg',
- 'delta;',
- 'diams;',
- 'divide;',
- 'divide',
- 'eacute;',
- 'eacute',
- 'ecirc;',
- 'ecirc',
- 'egrave;',
- 'egrave',
- 'empty;',
- 'emsp;',
- 'ensp;',
- 'epsilon;',
- 'equiv;',
- 'eta;',
- 'eth;',
- 'eth',
- 'euml;',
- 'euml',
- 'euro;',
- 'exist;',
- 'fnof;',
- 'forall;',
- 'frac12;',
- 'frac12',
- 'frac14;',
- 'frac14',
- 'frac34;',
- 'frac34',
- 'frasl;',
- 'gamma;',
- 'ge;',
- 'gt;',
- 'gt',
- 'hArr;',
- 'harr;',
- 'hearts;',
- 'hellip;',
- 'iacute;',
- 'iacute',
- 'icirc;',
- 'icirc',
- 'iexcl;',
- 'iexcl',
- 'igrave;',
- 'igrave',
- 'image;',
- 'infin;',
- 'int;',
- 'iota;',
- 'iquest;',
- 'iquest',
- 'isin;',
- 'iuml;',
- 'iuml',
- 'kappa;',
- 'lArr;',
- 'lambda;',
- 'lang;',
- 'laquo;',
- 'laquo',
- 'larr;',
- 'lceil;',
- 'ldquo;',
- 'le;',
- 'lfloor;',
- 'lowast;',
- 'loz;',
- 'lrm;',
- 'lsaquo;',
- 'lsquo;',
- 'lt;',
- 'lt',
- 'macr;',
- 'macr',
- 'mdash;',
- 'micro;',
- 'micro',
- 'middot;',
- 'middot',
- 'minus;',
- 'mu;',
- 'nabla;',
- 'nbsp;',
- 'nbsp',
- 'ndash;',
- 'ne;',
- 'ni;',
- 'not;',
- 'not',
- 'notin;',
- 'nsub;',
- 'ntilde;',
- 'ntilde',
- 'nu;',
- 'oacute;',
- 'oacute',
- 'ocirc;',
- 'ocirc',
- 'oelig;',
- 'ograve;',
- 'ograve',
- 'oline;',
- 'omega;',
- 'omicron;',
- 'oplus;',
- 'or;',
- 'ordf;',
- 'ordf',
- 'ordm;',
- 'ordm',
- 'oslash;',
- 'oslash',
- 'otilde;',
- 'otilde',
- 'otimes;',
- 'ouml;',
- 'ouml',
- 'para;',
- 'para',
- 'part;',
- 'permil;',
- 'perp;',
- 'phi;',
- 'pi;',
- 'piv;',
- 'plusmn;',
- 'plusmn',
- 'pound;',
- 'pound',
- 'prime;',
- 'prod;',
- 'prop;',
- 'psi;',
- 'quot;',
- 'quot',
- 'rArr;',
- 'radic;',
- 'rang;',
- 'raquo;',
- 'raquo',
- 'rarr;',
- 'rceil;',
- 'rdquo;',
- 'real;',
- 'reg;',
- 'reg',
- 'rfloor;',
- 'rho;',
- 'rlm;',
- 'rsaquo;',
- 'rsquo;',
- 'sbquo;',
- 'scaron;',
- 'sdot;',
- 'sect;',
- 'sect',
- 'shy;',
- 'shy',
- 'sigma;',
- 'sigmaf;',
- 'sim;',
- 'spades;',
- 'sub;',
- 'sube;',
- 'sum;',
- 'sup1;',
- 'sup1',
- 'sup2;',
- 'sup2',
- 'sup3;',
- 'sup3',
- 'sup;',
- 'supe;',
- 'szlig;',
- 'szlig',
- 'tau;',
- 'there4;',
- 'theta;',
- 'thetasym;',
- 'thinsp;',
- 'thorn;',
- 'thorn',
- 'tilde;',
- 'times;',
- 'times',
- 'trade;',
- 'uArr;',
- 'uacute;',
- 'uacute',
- 'uarr;',
- 'ucirc;',
- 'ucirc',
- 'ugrave;',
- 'ugrave',
- 'uml;',
- 'uml',
- 'upsih;',
- 'upsilon;',
- 'uuml;',
- 'uuml',
- 'weierp;',
- 'xi;',
- 'yacute;',
- 'yacute',
- 'yen;',
- 'yen',
- 'yuml;',
- 'yuml',
- 'zeta;',
- 'zwj;',
- 'zwnj;'
- );
- const PCDATA = 0;
- const RCDATA = 1;
- const CDATA = 2;
- const PLAINTEXT = 3;
- const DOCTYPE = 0;
- const STARTTAG = 1;
- const ENDTAG = 2;
- const COMMENT = 3;
- const CHARACTR = 4;
- const EOF = 5;
- public function __construct($data)
- {
- $this->data = $data;
- $this->char = -1;
- $this->EOF = strlen($data);
- $this->tree = new HTML5TreeConstructer;
- $this->content_model = self::PCDATA;
- $this->state = 'data';
- while ($this->state !== null) {
- $this->{$this->state . 'State'}();
- }
- }
- public function save()
- {
- return $this->tree->save();
- }
- private function char()
- {
- return ($this->char < $this->EOF)
- ? $this->data[$this->char]
- : false;
- }
- private function character($s, $l = 0)
- {
- if ($s + $l < $this->EOF) {
- if ($l === 0) {
- return $this->data[$s];
- } else {
- return substr($this->data, $s, $l);
- }
- }
- }
- private function characters($char_class, $start)
- {
- return preg_replace('#^([' . $char_class . ']+).*#s', '\\1', substr($this->data, $start));
- }
- private function dataState()
- {
-
- $this->char++;
- $char = $this->char();
- if ($char === '&' && ($this->content_model === self::PCDATA || $this->content_model === self::RCDATA)) {
-
- $this->state = 'entityData';
- } elseif ($char === '-') {
-
- if (($this->content_model === self::RCDATA || $this->content_model ===
- self::CDATA) && $this->escape === false &&
- $this->char >= 3 && $this->character($this->char - 4, 4) === '<!--'
- ) {
- $this->escape = true;
- }
-
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => $char
- )
- );
-
- } elseif ($char === '<' && ($this->content_model === self::PCDATA ||
- (($this->content_model === self::RCDATA ||
- $this->content_model === self::CDATA) && $this->escape === false))
- ) {
-
- $this->state = 'tagOpen';
-
- } elseif ($char === '>') {
-
- if (($this->content_model === self::RCDATA ||
- $this->content_model === self::CDATA) && $this->escape === true &&
- $this->character($this->char, 3) === '-->'
- ) {
- $this->escape = false;
- }
-
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => $char
- )
- );
- } elseif ($this->char === $this->EOF) {
-
- $this->EOF();
- } elseif ($this->content_model === self::PLAINTEXT) {
-
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => substr($this->data, $this->char)
- )
- );
- $this->EOF();
- } else {
-
- $len = strcspn($this->data, '<&', $this->char);
- $char = substr($this->data, $this->char, $len);
- $this->char += $len - 1;
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => $char
- )
- );
- $this->state = 'data';
- }
- }
- private function entityDataState()
- {
-
- $entity = $this->entity();
-
-
- $char = (!$entity) ? '&' : $entity;
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => $char
- )
- );
-
- $this->state = 'data';
- }
- private function tagOpenState()
- {
- switch ($this->content_model) {
- case self::RCDATA:
- case self::CDATA:
-
- if ($this->character($this->char + 1) === '/') {
- $this->char++;
- $this->state = 'closeTagOpen';
- } else {
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => '<'
- )
- );
- $this->state = 'data';
- }
- break;
- case self::PCDATA:
-
-
- $this->char++;
- $char = $this->char();
- if ($char === '!') {
-
- $this->state = 'markupDeclarationOpen';
- } elseif ($char === '/') {
-
- $this->state = 'closeTagOpen';
- } elseif (preg_match('/^[A-Za-z]$/', $char)) {
-
- $this->token = array(
- 'name' => strtolower($char),
- 'type' => self::STARTTAG,
- 'attr' => array()
- );
- $this->state = 'tagName';
- } elseif ($char === '>') {
-
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => '<>'
- )
- );
- $this->state = 'data';
- } elseif ($char === '?') {
-
- $this->state = 'bogusComment';
- } else {
-
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => '<'
- )
- );
- $this->char--;
- $this->state = 'data';
- }
- break;
- }
- }
- private function closeTagOpenState()
- {
- $next_node = strtolower($this->characters('A-Za-z', $this->char + 1));
- $the_same = count($this->tree->stack) > 0 && $next_node === end($this->tree->stack)->nodeName;
- if (($this->content_model === self::RCDATA || $this->content_model === self::CDATA) &&
- (!$the_same || ($the_same && (!preg_match(
- '/[\t\n\x0b\x0c >\/]/',
- $this->character($this->char + 1 + strlen($next_node))
- ) || $this->EOF === $this->char)))
- ) {
-
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => '</'
- )
- );
- $this->state = 'data';
- } else {
-
- $this->char++;
- $char = $this->char();
- if (preg_match('/^[A-Za-z]$/', $char)) {
-
- $this->token = array(
- 'name' => strtolower($char),
- 'type' => self::ENDTAG
- );
- $this->state = 'tagName';
- } elseif ($char === '>') {
-
- $this->state = 'data';
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken(
- array(
- 'type' => self::CHARACTR,
- 'data' => '</'
- )
- );
- $this->char--;
- $this->state = 'data';
- } else {
-
- $this->state = 'bogusComment';
- }
- }
- }
- private function tagNameState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- $this->state = 'beforeAttributeName';
- } elseif ($char === '>') {
-
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } elseif ($char === '/') {
-
- $this->state = 'beforeAttributeName';
- } else {
-
- $this->token['name'] .= strtolower($char);
- $this->state = 'tagName';
- }
- }
- private function beforeAttributeNameState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- $this->state = 'beforeAttributeName';
- } elseif ($char === '>') {
-
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif ($char === '/') {
-
- $this->state = 'beforeAttributeName';
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
-
- $this->token['attr'][] = array(
- 'name' => strtolower($char),
- 'value' => null
- );
- $this->state = 'attributeName';
- }
- }
- private function attributeNameState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- $this->state = 'afterAttributeName';
- } elseif ($char === '=') {
-
- $this->state = 'beforeAttributeValue';
- } elseif ($char === '>') {
-
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif ($char === '/' && $this->character($this->char + 1) !== '>') {
-
- $this->state = 'beforeAttributeName';
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
-
- $last = count($this->token['attr']) - 1;
- $this->token['attr'][$last]['name'] .= strtolower($char);
- $this->state = 'attributeName';
- }
- }
- private function afterAttributeNameState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- $this->state = 'afterAttributeName';
- } elseif ($char === '=') {
-
- $this->state = 'beforeAttributeValue';
- } elseif ($char === '>') {
-
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif ($char === '/' && $this->character($this->char + 1) !== '>') {
-
- $this->state = 'beforeAttributeName';
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
-
- $this->token['attr'][] = array(
- 'name' => strtolower($char),
- 'value' => null
- );
- $this->state = 'attributeName';
- }
- }
- private function beforeAttributeValueState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- $this->state = 'beforeAttributeValue';
- } elseif ($char === '"') {
-
- $this->state = 'attributeValueDoubleQuoted';
- } elseif ($char === '&') {
-
- $this->char--;
- $this->state = 'attributeValueUnquoted';
- } elseif ($char === '\'') {
-
- $this->state = 'attributeValueSingleQuoted';
- } elseif ($char === '>') {
-
- $this->emitToken($this->token);
- $this->state = 'data';
- } else {
-
- $last = count($this->token['attr']) - 1;
- $this->token['attr'][$last]['value'] .= $char;
- $this->state = 'attributeValueUnquoted';
- }
- }
- private function attributeValueDoubleQuotedState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if ($char === '"') {
-
- $this->state = 'beforeAttributeName';
- } elseif ($char === '&') {
-
- $this->entityInAttributeValueState('double');
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
-
- $last = count($this->token['attr']) - 1;
- $this->token['attr'][$last]['value'] .= $char;
- $this->state = 'attributeValueDoubleQuoted';
- }
- }
- private function attributeValueSingleQuotedState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if ($char === '\'') {
-
- $this->state = 'beforeAttributeName';
- } elseif ($char === '&') {
-
- $this->entityInAttributeValueState('single');
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
-
- $last = count($this->token['attr']) - 1;
- $this->token['attr'][$last]['value'] .= $char;
- $this->state = 'attributeValueSingleQuoted';
- }
- }
- private function attributeValueUnquotedState()
- {
-
- $this->char++;
- $char = $this->character($this->char);
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- $this->state = 'beforeAttributeName';
- } elseif ($char === '&') {
-
- $this->entityInAttributeValueState();
- } elseif ($char === '>') {
-
- $this->emitToken($this->token);
- $this->state = 'data';
- } else {
-
- $last = count($this->token['attr']) - 1;
- $this->token['attr'][$last]['value'] .= $char;
- $this->state = 'attributeValueUnquoted';
- }
- }
- private function entityInAttributeValueState()
- {
-
- $entity = $this->entity();
-
-
-
- $char = (!$entity)
- ? '&'
- : $entity;
- $last = count($this->token['attr']) - 1;
- $this->token['attr'][$last]['value'] .= $char;
- }
- private function bogusCommentState()
- {
-
- $data = $this->characters('^>', $this->char);
- $this->emitToken(
- array(
- 'data' => $data,
- 'type' => self::COMMENT
- )
- );
- $this->char += strlen($data);
-
- $this->state = 'data';
-
- if ($this->char === $this->EOF) {
- $this->char = $this->EOF - 1;
- }
- }
- private function markupDeclarationOpenState()
- {
-
- if ($this->character($this->char + 1, 2) === '--') {
- $this->char += 2;
- $this->state = 'comment';
- $this->token = array(
- 'data' => null,
- 'type' => self::COMMENT
- );
-
- } elseif (strtolower($this->character($this->char + 1, 7)) === 'doctype') {
- $this->char += 7;
- $this->state = 'doctype';
-
- } else {
- $this->char++;
- $this->state = 'bogusComment';
- }
- }
- private function commentState()
- {
-
- $this->char++;
- $char = $this->char();
-
- if ($char === '-') {
-
- $this->state = 'commentDash';
-
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
-
- } else {
-
- $this->token['data'] .= $char;
- }
- }
- private function commentDashState()
- {
-
- $this->char++;
- $char = $this->char();
-
- if ($char === '-') {
-
- $this->state = 'commentEnd';
-
- } elseif ($this->char === $this->EOF) {
-
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
-
- } else {
-
- $this->token['data'] .= '-' . $char;
- $this->state = 'comment';
- }
- }
- private function commentEndState()
- {
-
- $this->char++;
- $char = $this->char();
- if ($char === '>') {
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif ($char === '-') {
- $this->token['data'] .= '-';
- } elseif ($this->char === $this->EOF) {
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
- $this->token['data'] .= '--' . $char;
- $this->state = 'comment';
- }
- }
- private function doctypeState()
- {
-
- $this->char++;
- $char = $this->char();
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
- $this->state = 'beforeDoctypeName';
- } else {
- $this->char--;
- $this->state = 'beforeDoctypeName';
- }
- }
- private function beforeDoctypeNameState()
- {
-
- $this->char++;
- $char = $this->char();
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- } elseif (preg_match('/^[a-z]$/', $char)) {
- $this->token = array(
- 'name' => strtoupper($char),
- 'type' => self::DOCTYPE,
- 'error' => true
- );
- $this->state = 'doctypeName';
- } elseif ($char === '>') {
- $this->emitToken(
- array(
- 'name' => null,
- 'type' => self::DOCTYPE,
- 'error' => true
- )
- );
- $this->state = 'data';
- } elseif ($this->char === $this->EOF) {
- $this->emitToken(
- array(
- 'name' => null,
- 'type' => self::DOCTYPE,
- 'error' => true
- )
- );
- $this->char--;
- $this->state = 'data';
- } else {
- $this->token = array(
- 'name' => $char,
- 'type' => self::DOCTYPE,
- 'error' => true
- );
- $this->state = 'doctypeName';
- }
- }
- private function doctypeNameState()
- {
-
- $this->char++;
- $char = $this->char();
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
- $this->state = 'AfterDoctypeName';
- } elseif ($char === '>') {
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif (preg_match('/^[a-z]$/', $char)) {
- $this->token['name'] .= strtoupper($char);
- } elseif ($this->char === $this->EOF) {
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
- $this->token['name'] .= $char;
- }
- $this->token['error'] = ($this->token['name'] === 'HTML')
- ? false
- : true;
- }
- private function afterDoctypeNameState()
- {
-
- $this->char++;
- $char = $this->char();
- if (preg_match('/^[\t\n\x0b\x0c ]$/', $char)) {
-
- } elseif ($char === '>') {
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif ($this->char === $this->EOF) {
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
- $this->token['error'] = true;
- $this->state = 'bogusDoctype';
- }
- }
- private function bogusDoctypeState()
- {
-
- $this->char++;
- $char = $this->char();
- if ($char === '>') {
- $this->emitToken($this->token);
- $this->state = 'data';
- } elseif ($this->char === $this->EOF) {
- $this->emitToken($this->token);
- $this->char--;
- $this->state = 'data';
- } else {
-
- }
- }
- private function entity()
- {
- $start = $this->char;
-
-
-
-
- switch ($this->character($this->char + 1)) {
-
- case '#':
-
-
- switch ($this->character($this->char + 1)) {
-
-
- case 'x':
- case 'X':
-
-
-
-
-
-
- $char = 1;
- $char_class = '0-9A-Fa-f';
- break;
-
- default:
-
-
-
- $char = 0;
- $char_class = '0-9';
- break;
- }
-
-
- $this->char++;
- $e_name = $this->characters($char_class, $this->char + $char + 1);
- $entity = $this->character($start, $this->char);
- $cond = strlen($e_name) > 0;
-
- break;
-
- default:
-
-
-
- $e_name = $this->characters('0-9A-Za-z;', $this->char + 1);
- $len = strlen($e_name);
- for ($c = 1; $c <= $len; $c++) {
- $id = substr($e_name, 0, $c);
- $this->char++;
- if (in_array($id, $this->entities)) {
- if ($e_name[$c - 1] !== ';') {
- if ($c < $len && $e_name[$c] == ';') {
- $this->char++;
- }
- }
- $entity = $id;
- break;
- }
- }
- $cond = isset($entity);
-
- break;
- }
- if (!$cond) {
-
-
- $this->char = $start;
- return false;
- }
-
-
- return html_entity_decode('&' . rtrim($entity, ';') . ';', ENT_QUOTES, 'UTF-8');
- }
- private function emitToken($token)
- {
- $emit = $this->tree->emitToken($token);
- if (is_int($emit)) {
- $this->content_model = $emit;
- } elseif ($token['type'] === self::ENDTAG) {
- $this->content_model = self::PCDATA;
- }
- }
- private function EOF()
- {
- $this->state = null;
- $this->tree->emitToken(
- array(
- 'type' => self::EOF
- )
- );
- }
- }
- class HTML5TreeConstructer
- {
- public $stack = array();
- private $phase;
- private $mode;
- private $dom;
- private $foster_parent = null;
- private $a_formatting = array();
- private $head_pointer = null;
- private $form_pointer = null;
- private $scoping = array('button', 'caption', 'html', 'marquee', 'object', 'table', 'td', 'th');
- private $formatting = array(
- 'a',
- 'b',
- 'big',
- 'em',
- 'font',
- 'i',
- 'nobr',
- 's',
- 'small',
- 'strike',
- 'strong',
- 'tt',
- 'u'
- );
- private $special = array(
- 'address',
- 'area',
- 'base',
- 'basefont',
- 'bgsound',
- 'blockquote',
- 'body',
- 'br',
- 'center',
- 'col',
- 'colgroup',
- 'dd',
- 'dir',
- 'div',
- 'dl',
- 'dt',
- 'embed',
- 'fieldset',
- 'form',
- 'frame',
- 'frameset',
- 'h1',
- 'h2',
- 'h3',
- 'h4',
- 'h5',
- 'h6',
- 'head',
- 'hr',
- 'iframe',
- 'image',
- 'img',
- 'input',
- 'isindex',
- 'li',
- 'link',
- 'listing',
- 'menu',
- 'meta',
- 'noembed',
- 'noframes',
- 'noscript',
- 'ol',
- 'optgroup',
- 'option',
- 'p',
- 'param',
- 'plaintext',
- 'pre',
- 'script',
- 'select',
- 'spacer',
- 'style',
- 'tbody',
- 'textarea',
- 'tfoot',
- 'thead',
- 'title',
- 'tr',
- 'ul',
- 'wbr'
- );
-
- const INIT_PHASE = 0;
- const ROOT_PHASE = 1;
- const MAIN_PHASE = 2;
- const END_PHASE = 3;
-
- const BEFOR_HEAD = 0;
- const IN_HEAD = 1;
- const AFTER_HEAD = 2;
- const IN_BODY = 3;
- const IN_TABLE = 4;
- const IN_CAPTION = 5;
- const IN_CGROUP = 6;
- const IN_TBODY = 7;
- const IN_ROW = 8;
- const IN_CELL = 9;
- const IN_SELECT = 10;
- const AFTER_BODY = 11;
- const IN_FRAME = 12;
- const AFTR_FRAME = 13;
-
- const SPECIAL = 0;
- const SCOPING = 1;
- const FORMATTING = 2;
- const PHRASING = 3;
- const MARKER = 0;
- public function __construct()
- {
- $this->phase = self::INIT_PHASE;
- $this->mode = self::BEFOR_HEAD;
- $this->dom = new DOMDocument;
- $this->dom->encoding = 'UTF-8';
- $this->dom->preserveWhiteSpace = true;
- $this->dom->substituteEntities = true;
- $this->dom->strictErrorChecking = false;
- }
-
- public function emitToken($token)
- {
- switch ($this->phase) {
- case self::INIT_PHASE:
- return $this->initPhase($token);
- break;
- case self::ROOT_PHASE:
- return $this->rootElementPhase($token);
- break;
- case self::MAIN_PHASE:
- return $this->mainPhase($token);
- break;
- case self::END_PHASE :
- return $this->trailingEndPhase($token);
- break;
- }
- }
- private function initPhase($token)
- {
-
-
- if ((isset($token['error']) && $token['error']) ||
- $token['type'] === HTML5::COMMENT ||
- $token['type'] === HTML5::STARTTAG ||
- $token['type'] === HTML5::ENDTAG ||
- $token['type'] === HTML5::EOF ||
- ($token['type'] === HTML5::CHARACTR && isset($token['data']) &&
- !preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data']))
- ) {
-
- $this->phase = self::ROOT_PHASE;
- return $this->rootElementPhase($token);
-
- } elseif (isset($token['error']) && !$token['error']) {
-
- $doctype = new DOMDocumentType(null, null, 'HTML');
-
- $this->phase = self::ROOT_PHASE;
-
- } elseif (isset($token['data']) && preg_match(
- '/^[\t\n\x0b\x0c ]+$/',
- $token['data']
- )
- ) {
-
- $text = $this->dom->createTextNode($token['data']);
- $this->dom->appendChild($text);
- }
- }
- private function rootElementPhase($token)
- {
-
-
- if ($token['type'] === HTML5::DOCTYPE) {
-
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $comment = $this->dom->createComment($token['data']);
- $this->dom->appendChild($comment);
-
- } elseif ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $text = $this->dom->createTextNode($token['data']);
- $this->dom->appendChild($text);
-
- } elseif (($token['type'] === HTML5::CHARACTR &&
- !preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])) ||
- $token['type'] === HTML5::STARTTAG ||
- $token['type'] === HTML5::ENDTAG ||
- $token['type'] === HTML5::EOF
- ) {
-
- $html = $this->dom->createElement('html');
- $this->dom->appendChild($html);
- $this->stack[] = $html;
- $this->phase = self::MAIN_PHASE;
- return $this->mainPhase($token);
- }
- }
- private function mainPhase($token)
- {
-
-
- if ($token['type'] === HTML5::DOCTYPE) {
-
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'html') {
-
-
- foreach ($token['attr'] as $attr) {
- if (!$this->stack[0]->hasAttribute($attr['name'])) {
- $this->stack[0]->setAttribute($attr['name'], $attr['value']);
- }
- }
-
- } elseif ($token['type'] === HTML5::EOF) {
-
- $this->generateImpliedEndTags();
-
- } else {
-
- switch ($this->mode) {
- case self::BEFOR_HEAD:
- return $this->beforeHead($token);
- break;
- case self::IN_HEAD:
- return $this->inHead($token);
- break;
- case self::AFTER_HEAD:
- return $this->afterHead($token);
- break;
- case self::IN_BODY:
- return $this->inBody($token);
- break;
- case self::IN_TABLE:
- return $this->inTable($token);
- break;
- case self::IN_CAPTION:
- return $this->inCaption($token);
- break;
- case self::IN_CGROUP:
- return $this->inColumnGroup($token);
- break;
- case self::IN_TBODY:
- return $this->inTableBody($token);
- break;
- case self::IN_ROW:
- return $this->inRow($token);
- break;
- case self::IN_CELL:
- return $this->inCell($token);
- break;
- case self::IN_SELECT:
- return $this->inSelect($token);
- break;
- case self::AFTER_BODY:
- return $this->afterBody($token);
- break;
- case self::IN_FRAME:
- return $this->inFrameset($token);
- break;
- case self::AFTR_FRAME:
- return $this->afterFrameset($token);
- break;
- case self::END_PHASE:
- return $this->trailingEndPhase($token);
- break;
- }
- }
- }
- private function beforeHead($token)
- {
-
-
- if ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $this->insertText($token['data']);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $this->insertComment($token['data']);
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'head') {
-
- $element = $this->insertElement($token);
-
- $this->head_pointer = $element;
-
- $this->mode = self::IN_HEAD;
-
- } elseif ($token['type'] === HTML5::STARTTAG ||
- ($token['type'] === HTML5::ENDTAG && $token['name'] === 'html') ||
- ($token['type'] === HTML5::CHARACTR && !preg_match(
- '/^[\t\n\x0b\x0c ]$/',
- $token['data']
- ))
- ) {
-
- $this->beforeHead(
- array(
- 'name' => 'head',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
- return $this->inHead($token);
-
- } elseif ($token['type'] === HTML5::ENDTAG) {
-
- }
- }
- private function inHead($token)
- {
-
-
- if (($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])) || (
- $token['type'] === HTML5::CHARACTR && in_array(
- end($this->stack)->nodeName,
- array('title', 'style', 'script')
- ))
- ) {
-
- $this->insertText($token['data']);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $this->insertComment($token['data']);
- } elseif ($token['type'] === HTML5::ENDTAG &&
- in_array($token['name'], array('title', 'style', 'script'))
- ) {
- array_pop($this->stack);
- return HTML5::PCDATA;
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'title') {
-
- if ($this->head_pointer !== null) {
- $element = $this->insertElement($token, false);
- $this->head_pointer->appendChild($element);
- } else {
- $element = $this->insertElement($token);
- }
-
- return HTML5::RCDATA;
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'style') {
-
- if ($this->head_pointer !== null) {
- $element = $this->insertElement($token, false);
- $this->head_pointer->appendChild($element);
- } else {
- $this->insertElement($token);
- }
-
- return HTML5::CDATA;
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'script') {
-
- $element = $this->insertElement($token, false);
- $this->head_pointer->appendChild($element);
-
- return HTML5::CDATA;
-
- } elseif ($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array('base', 'link', 'meta')
- )
- ) {
-
- if ($this->head_pointer !== null) {
- $element = $this->insertElement($token, false);
- $this->head_pointer->appendChild($element);
- array_pop($this->stack);
- } else {
- $this->insertElement($token);
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG && $token['name'] === 'head') {
-
- if ($this->head_pointer->isSameNode(end($this->stack))) {
- array_pop($this->stack);
-
- } else {
-
- }
-
- $this->mode = self::AFTER_HEAD;
-
- } elseif (($token['type'] === HTML5::STARTTAG && $token['name'] === 'head') ||
- ($token['type'] === HTML5::ENDTAG && $token['name'] !== 'html')
- ) {
-
-
- } else {
-
- if ($this->head_pointer->isSameNode(end($this->stack))) {
- $this->inHead(
- array(
- 'name' => 'head',
- 'type' => HTML5::ENDTAG
- )
- );
-
- } else {
- $this->mode = self::AFTER_HEAD;
- }
-
- return $this->afterHead($token);
- }
- }
- private function afterHead($token)
- {
-
-
- if ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $this->insertText($token['data']);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $this->insertComment($token['data']);
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'body') {
-
- $this->insertElement($token);
-
- $this->mode = self::IN_BODY;
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'frameset') {
-
- $this->insertElement($token);
-
- $this->mode = self::IN_FRAME;
-
- } elseif ($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array('base', 'link', 'meta', 'script', 'style', 'title')
- )
- ) {
-
- $this->mode = self::IN_HEAD;
- return $this->inHead($token);
-
- } else {
-
- $this->afterHead(
- array(
- 'name' => 'body',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
- return $this->inBody($token);
- }
- }
- private function inBody($token)
- {
-
- switch ($token['type']) {
-
- case HTML5::CHARACTR:
-
- $this->reconstructActiveFormattingElements();
-
- $this->insertText($token['data']);
- break;
-
- case HTML5::COMMENT:
-
- $this->insertComment($token['data']);
- break;
- case HTML5::STARTTAG:
- switch ($token['name']) {
-
- case 'script':
- case 'style':
-
- return $this->inHead($token);
- break;
-
- case 'base':
- case 'link':
- case 'meta':
- case 'title':
-
- return $this->inHead($token);
- break;
-
- case 'body':
-
- if (count($this->stack) === 1 || $this->stack[1]->nodeName !== 'body') {
-
-
- } else {
- foreach ($token['attr'] as $attr) {
- if (!$this->stack[1]->hasAttribute($attr['name'])) {
- $this->stack[1]->setAttribute($attr['name'], $attr['value']);
- }
- }
- }
- break;
-
- case 'address':
- case 'blockquote':
- case 'center':
- case 'dir':
- case 'div':
- case 'dl':
- case 'fieldset':
- case 'listing':
- case 'menu':
- case 'ol':
- case 'p':
- case 'ul':
-
- if ($this->elementInScope('p')) {
- $this->emitToken(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $this->insertElement($token);
- break;
-
- case 'form':
-
- if ($this->form_pointer !== null) {
-
-
- } else {
-
- if ($this->elementInScope('p')) {
- $this->emitToken(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $element = $this->insertElement($token);
- $this->form_pointer = $element;
- }
- break;
-
- case 'li':
- case 'dd':
- case 'dt':
-
- if ($this->elementInScope('p')) {
- $this->emitToken(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
- }
- $stack_length = count($this->stack) - 1;
- for ($n = $stack_length; 0 <= $n; $n--) {
-
- $stop = false;
- $node = $this->stack[$n];
- $cat = $this->getElementCategory($node->tagName);
-
- if ($token['name'] === $node->tagName || ($token['name'] !== 'li'
- && ($node->tagName === 'dd' || $node->tagName === 'dt'))
- ) {
- for ($x = $stack_length; $x >= $n; $x--) {
- array_pop($this->stack);
- }
- break;
- }
-
- if ($cat !== self::FORMATTING && $cat !== self::PHRASING &&
- $node->tagName !== 'address' && $node->tagName !== 'div'
- ) {
- break;
- }
- }
-
- $this->insertElement($token);
- break;
-
- case 'plaintext':
-
- if ($this->elementInScope('p')) {
- $this->emitToken(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $this->insertElement($token);
- return HTML5::PLAINTEXT;
- break;
-
- case 'h1':
- case 'h2':
- case 'h3':
- case 'h4':
- case 'h5':
- case 'h6':
-
- if ($this->elementInScope('p')) {
- $this->emitToken(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- while ($this->elementInScope(array('h1', 'h2', 'h3', 'h4', 'h5', 'h6'))) {
- array_pop($this->stack);
- }
-
- $this->insertElement($token);
- break;
-
- case 'a':
-
- $leng = count($this->a_formatting);
- for ($n = $leng - 1; $n >= 0; $n--) {
- if ($this->a_formatting[$n] === self::MARKER) {
- break;
- } elseif ($this->a_formatting[$n]->nodeName === 'a') {
- $this->emitToken(
- array(
- 'name' => 'a',
- 'type' => HTML5::ENDTAG
- )
- );
- break;
- }
- }
-
- $this->reconstructActiveFormattingElements();
-
- $el = $this->insertElement($token);
-
- $this->a_formatting[] = $el;
- break;
-
- case 'b':
- case 'big':
- case 'em':
- case 'font':
- case 'i':
- case 'nobr':
- case 's':
- case 'small':
- case 'strike':
- case 'strong':
- case 'tt':
- case 'u':
-
- $this->reconstructActiveFormattingElements();
-
- $el = $this->insertElement($token);
-
- $this->a_formatting[] = $el;
- break;
-
- case 'button':
-
- if ($this->elementInScope('button')) {
- $this->inBody(
- array(
- 'name' => 'button',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $this->reconstructActiveFormattingElements();
-
- $this->insertElement($token);
-
- $this->a_formatting[] = self::MARKER;
- break;
-
- case 'marquee':
- case 'object':
-
- $this->reconstructActiveFormattingElements();
-
- $this->insertElement($token);
-
- $this->a_formatting[] = self::MARKER;
- break;
-
- case 'xmp':
-
- $this->reconstructActiveFormattingElements();
-
- $this->insertElement($token);
-
- return HTML5::CDATA;
- break;
-
- case 'table':
-
- if ($this->elementInScope('p')) {
- $this->emitToken(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $this->insertElement($token);
-
- $this->mode = self::IN_TABLE;
- break;
-
- case 'area':
- case 'basefont':
- case 'bgsound':
- case 'br':
- case 'embed':
- case 'img':
- case 'param':
- case 'spacer':
- case 'wbr':
-
- $this->reconstructActiveFormattingElements();
-
- $this->insertElement($token);
-
- array_pop($this->stack);
- break;
-
- case 'hr':
-
- if ($this->elementInScope('p')) {
- $this->emitToken(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $this->insertElement($token);
-
- array_pop($this->stack);
- break;
-
- case 'image':
-
- $token['name'] = 'img';
- return $this->inBody($token);
- break;
-
- case 'input':
-
- $this->reconstructActiveFormattingElements();
-
- $element = $this->insertElement($token, false);
-
- $this->form_pointer !== null
- ? $this->form_pointer->appendChild($element)
- : end($this->stack)->appendChild($element);
-
- array_pop($this->stack);
- break;
-
- case 'isindex':
-
-
-
- if ($this->form_pointer === null) {
-
- $this->inBody(
- array(
- 'name' => 'body',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
-
- $this->inBody(
- array(
- 'name' => 'hr',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
-
- $this->inBody(
- array(
- 'name' => 'p',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
-
- $this->inBody(
- array(
- 'name' => 'label',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
-
- $this->insertText(
- 'This is a searchable index. ' .
- 'Insert your search keywords here: '
- );
-
- $attr = $token['attr'];
- $attr[] = array('name' => 'name', 'value' => 'isindex');
- $this->inBody(
- array(
- 'name' => 'input',
- 'type' => HTML5::STARTTAG,
- 'attr' => $attr
- )
- );
-
- $this->insertText(
- 'This is a searchable index. ' .
- 'Insert your search keywords here: '
- );
-
- $this->inBody(
- array(
- 'name' => 'label',
- 'type' => HTML5::ENDTAG
- )
- );
-
- $this->inBody(
- array(
- 'name' => 'p',
- 'type' => HTML5::ENDTAG
- )
- );
-
- $this->inBody(
- array(
- 'name' => 'hr',
- 'type' => HTML5::ENDTAG
- )
- );
-
- $this->inBody(
- array(
- 'name' => 'form',
- 'type' => HTML5::ENDTAG
- )
- );
- }
- break;
-
- case 'textarea':
- $this->insertElement($token);
-
- return HTML5::RCDATA;
- break;
-
- case 'iframe':
- case 'noembed':
- case 'noframes':
- $this->insertElement($token);
-
- return HTML5::CDATA;
- break;
-
- case 'select':
-
- $this->reconstructActiveFormattingElements();
-
- $this->insertElement($token);
-
- $this->mode = self::IN_SELECT;
- break;
-
- case 'caption':
- case 'col':
- case 'colgroup':
- case 'frame':
- case 'frameset':
- case 'head':
- case 'option':
- case 'optgroup':
- case 'tbody':
- case 'td':
- case 'tfoot':
- case 'th':
- case 'thead':
- case 'tr':
-
- break;
-
- case 'event-source':
- case 'section':
- case 'nav':
- case 'article':
- case 'aside':
- case 'header':
- case 'footer':
- case 'datagrid':
- case 'command':
-
- break;
-
- default:
-
- $this->reconstructActiveFormattingElements();
- $this->insertElement($token, true, true);
- break;
- }
- break;
- case HTML5::ENDTAG:
- switch ($token['name']) {
-
- case 'body':
-
- if (count($this->stack) < 2 || $this->stack[1]->nodeName !== 'body') {
-
-
- } elseif (end($this->stack)->nodeName !== 'body') {
-
- }
-
- $this->mode = self::AFTER_BODY;
- break;
-
- case 'html':
-
- $this->inBody(
- array(
- 'name' => 'body',
- 'type' => HTML5::ENDTAG
- )
- );
- return $this->afterBody($token);
- break;
-
- case 'address':
- case 'blockquote':
- case 'center':
- case 'dir':
- case 'div':
- case 'dl':
- case 'fieldset':
- case 'listing':
- case 'menu':
- case 'ol':
- case 'pre':
- case 'ul':
-
- if ($this->elementInScope($token['name'])) {
- $this->generateImpliedEndTags();
-
-
-
- for ($n = count($this->stack) - 1; $n >= 0; $n--) {
- if ($this->stack[$n]->nodeName === $token['name']) {
- $n = -1;
- }
- array_pop($this->stack);
- }
- }
- break;
-
- case 'form':
-
- if ($this->elementInScope($token['name'])) {
- $this->generateImpliedEndTags();
- }
- if (end($this->stack)->nodeName !== $token['name']) {
-
-
- } else {
-
- array_pop($this->stack);
- }
-
- $this->form_pointer = null;
- break;
-
- case 'p':
-
- if ($this->elementInScope('p')) {
- $this->generateImpliedEndTags(array('p'));
-
-
-
- for ($n = count($this->stack) - 1; $n >= 0; $n--) {
- if ($this->elementInScope('p')) {
- array_pop($this->stack);
- } else {
- break;
- }
- }
- }
- break;
-
- case 'dd':
- case 'dt':
- case 'li':
-
- if ($this->elementInScope($token['name'])) {
- $this->generateImpliedEndTags(array($token['name']));
-
-
-
- for ($n = count($this->stack) - 1; $n >= 0; $n--) {
- if ($this->stack[$n]->nodeName === $token['name']) {
- $n = -1;
- }
- array_pop($this->stack);
- }
- }
- break;
-
- case 'h1':
- case 'h2':
- case 'h3':
- case 'h4':
- case 'h5':
- case 'h6':
- $elements = array('h1', 'h2', 'h3', 'h4', 'h5', 'h6');
-
- if ($this->elementInScope($elements)) {
- $this->generateImpliedEndTags();
-
-
-
- while ($this->elementInScope($elements)) {
- array_pop($this->stack);
- }
- }
- break;
-
- case 'a':
- case 'b':
- case 'big':
- case 'em':
- case 'font':
- case 'i':
- case 'nobr':
- case 's':
- case 'small':
- case 'strike':
- case 'strong':
- case 'tt':
- case 'u':
-
- while (true) {
- for ($a = count($this->a_formatting) - 1; $a >= 0; $a--) {
- if ($this->a_formatting[$a] === self::MARKER) {
- break;
- } elseif ($this->a_formatting[$a]->tagName === $token['name']) {
- $formatting_element = $this->a_formatting[$a];
- $in_stack = in_array($formatting_element, $this->stack, true);
- $fe_af_pos = $a;
- break;
- }
- }
-
- if (!isset($formatting_element) || ($in_stack &&
- !$this->elementInScope($token['name']))
- ) {
- break;
-
- } elseif (isset($formatting_element) && !$in_stack) {
- unset($this->a_formatting[$fe_af_pos]);
- $this->a_formatting = array_merge($this->a_formatting);
- break;
- }
-
- $fe_s_pos = array_search($formatting_element, $this->stack, true);
- $length = count($this->stack);
- for ($s = $fe_s_pos + 1; $s < $length; $s++) {
- $category = $this->getElementCategory($this->stack[$s]->nodeName);
- if ($category !== self::PHRASING && $category !== self::FORMATTING) {
- $furthest_block = $this->stack[$s];
- }
- }
-
- if (!isset($furthest_block)) {
- for ($n = $length - 1; $n >= $fe_s_pos; $n--) {
- array_pop($this->stack);
- }
- unset($this->a_formatting[$fe_af_pos]);
- $this->a_formatting = array_merge($this->a_formatting);
- break;
- }
-
- $common_ancestor = $this->stack[$fe_s_pos - 1];
-
- if ($furthest_block->parentNode !== null) {
- $furthest_block->parentNode->removeChild($furthest_block);
- }
-
- $bookmark = $fe_af_pos;
-
- $node = $furthest_block;
- $last_node = $furthest_block;
- while (true) {
- for ($n = array_search($node, $this->stack, true) - 1; $n >= 0; $n--) {
-
- $node = $this->stack[$n];
-
- if (!in_array($node, $this->a_formatting, true)) {
- unset($this->stack[$n]);
- $this->stack = array_merge($this->stack);
- } else {
- break;
- }
- }
-
- if ($node === $formatting_element) {
- break;
-
- } elseif ($last_node === $furthest_block) {
- $bookmark = array_search($node, $this->a_formatting, true) + 1;
- }
-
- if ($node->hasChildNodes()) {
- $clone = $node->cloneNode();
- $s_pos = array_search($node, $this->stack, true);
- $a_pos = array_search($node, $this->a_formatting, true);
- $this->stack[$s_pos] = $clone;
- $this->a_formatting[$a_pos] = $clone;
- $node = $clone;
- }
-
- if ($last_node->parentNode !== null) {
- $last_node->parentNode->removeChild($last_node);
- }
- $node->appendChild($last_node);
-
- $last_node = $node;
- }
-
- if ($last_node->parentNode !== null) {
- $last_node->parentNode->removeChild($last_node);
- }
- $common_ancestor->appendChild($last_node);
-
- $clone = $formatting_element->cloneNode();
-
- while ($furthest_block->hasChildNodes()) {
- $child = $furthest_block->firstChild;
- $furthest_block->removeChild($child);
- $clone->appendChild($child);
- }
-
- $furthest_block->appendChild($clone);
-
- $fe_af_pos = array_search($formatting_element, $this->a_formatting, true);
- unset($this->a_formatting[$fe_af_pos]);
- $this->a_formatting = array_merge($this->a_formatting);
- $af_part1 = array_slice($this->a_formatting, 0, $bookmark - 1);
- $af_part2 = array_slice($this->a_formatting, $bookmark, count($this->a_formatting));
- $this->a_formatting = array_merge($af_part1, array($clone), $af_part2);
-
- $fe_s_pos = array_search($formatting_element, $this->stack, true);
- $fb_s_pos = array_search($furthest_block, $this->stack, true);
- unset($this->stack[$fe_s_pos]);
- $s_part1 = array_slice($this->stack, 0, $fb_s_pos);
- $s_part2 = array_slice($this->stack, $fb_s_pos + 1, count($this->stack));
- $this->stack = array_merge($s_part1, array($clone), $s_part2);
-
- unset($formatting_element, $fe_af_pos, $fe_s_pos, $furthest_block);
- }
- break;
-
- case 'button':
- case 'marquee':
- case 'object':
-
- if ($this->elementInScope($token['name'])) {
- $this->generateImpliedEndTags();
-
-
-
- for ($n = count($this->stack) - 1; $n >= 0; $n--) {
- if ($this->stack[$n]->nodeName === $token['name']) {
- $n = -1;
- }
- array_pop($this->stack);
- }
- $marker = end(array_keys($this->a_formatting, self::MARKER, true));
- for ($n = count($this->a_formatting) - 1; $n > $marker; $n--) {
- array_pop($this->a_formatting);
- }
- }
- break;
-
- case 'area':
- case 'basefont':
- case 'bgsound':
- case 'br':
- case 'embed':
- case 'hr':
- case 'iframe':
- case 'image':
- case 'img':
- case 'input':
- case 'isindex':
- case 'noembed':
- case 'noframes':
- case 'param':
- case 'select':
- case 'spacer':
- case 'table':
- case 'textarea':
- case 'wbr':
-
- break;
-
- default:
- for ($n = count($this->stack) - 1; $n >= 0; $n--) {
-
- $node = end($this->stack);
-
- if ($token['name'] === $node->nodeName) {
-
- $this->generateImpliedEndTags();
-
-
-
- for ($x = count($this->stack) - $n; $x >= $n; $x--) {
- array_pop($this->stack);
- }
- } else {
- $category = $this->getElementCategory($node);
- if ($category !== self::SPECIAL && $category !== self::SCOPING) {
-
- return false;
- }
- }
- }
- break;
- }
- break;
- }
- }
- private function inTable($token)
- {
- $clear = array('html', 'table');
-
- if ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $text = $this->dom->createTextNode($token['data']);
- end($this->stack)->appendChild($text);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $comment = $this->dom->createComment($token['data']);
- end($this->stack)->appendChild($comment);
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- $token['name'] === 'caption'
- ) {
-
- $this->clearStackToTableContext($clear);
-
- $this->a_formatting[] = self::MARKER;
-
- $this->insertElement($token);
- $this->mode = self::IN_CAPTION;
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- $token['name'] === 'colgroup'
- ) {
-
- $this->clearStackToTableContext($clear);
-
- $this->insertElement($token);
- $this->mode = self::IN_CGROUP;
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- $token['name'] === 'col'
- ) {
- $this->inTable(
- array(
- 'name' => 'colgroup',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
- $this->inColumnGroup($token);
-
- } elseif ($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array('tbody', 'tfoot', 'thead')
- )
- ) {
-
- $this->clearStackToTableContext($clear);
-
- $this->insertElement($token);
- $this->mode = self::IN_TBODY;
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- in_array($token['name'], array('td', 'th', 'tr'))
- ) {
-
- $this->inTable(
- array(
- 'name' => 'tbody',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
- return $this->inTableBody($token);
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- $token['name'] === 'table'
- ) {
-
- $this->inTable(
- array(
- 'name' => 'table',
- 'type' => HTML5::ENDTAG
- )
- );
- return $this->mainPhase($token);
-
- } elseif ($token['type'] === HTML5::ENDTAG &&
- $token['name'] === 'table'
- ) {
-
- if (!$this->elementInScope($token['name'], true)) {
- return false;
-
- } else {
-
- $this->generateImpliedEndTags();
-
-
-
- while (true) {
- $current = end($this->stack)->nodeName;
- array_pop($this->stack);
- if ($current === 'table') {
- break;
- }
- }
-
- $this->resetInsertionMode();
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG && in_array(
- $token['name'],
- array(
- 'body',
- 'caption',
- 'col',
- 'colgroup',
- 'html',
- 'tbody',
- 'td',
- 'tfoot',
- 'th',
- 'thead',
- 'tr'
- )
- )
- ) {
-
-
- } else {
-
-
- if (in_array(
- end($this->stack)->nodeName,
- array('table', 'tbody', 'tfoot', 'thead', 'tr')
- )
- ) {
-
- for ($n = count($this->stack) - 1; $n >= 0; $n--) {
- if ($this->stack[$n]->nodeName === 'table') {
- $table = $this->stack[$n];
- break;
- }
- }
- if (isset($table) && $table->parentNode !== null) {
- $this->foster_parent = $table->parentNode;
- } elseif (!isset($table)) {
- $this->foster_parent = $this->stack[0];
- } elseif (isset($table) && ($table->parentNode === null ||
- $table->parentNode->nodeType !== XML_ELEMENT_NODE)
- ) {
- $this->foster_parent = $this->stack[$n - 1];
- }
- }
- $this->inBody($token);
- }
- }
- private function inCaption($token)
- {
-
- if ($token['type'] === HTML5::ENDTAG && $token['name'] === 'caption') {
-
- if (!$this->elementInScope($token['name'], true)) {
-
-
- } else {
-
- $this->generateImpliedEndTags();
-
-
-
- while (true) {
- $node = end($this->stack)->nodeName;
- array_pop($this->stack);
- if ($node === 'caption') {
- break;
- }
- }
-
- $this->clearTheActiveFormattingElementsUpToTheLastMarker();
-
- $this->mode = self::IN_TABLE;
- }
-
- } elseif (($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array(
- 'caption',
- 'col',
- 'colgroup',
- 'tbody',
- 'td',
- 'tfoot',
- 'th',
- 'thead',
- 'tr'
- )
- )) || ($token['type'] === HTML5::ENDTAG &&
- $token['name'] === 'table')
- ) {
-
- $this->inCaption(
- array(
- 'name' => 'caption',
- 'type' => HTML5::ENDTAG
- )
- );
- return $this->inTable($token);
-
- } elseif ($token['type'] === HTML5::ENDTAG && in_array(
- $token['name'],
- array(
- 'body',
- 'col',
- 'colgroup',
- 'html',
- 'tbody',
- 'tfoot',
- 'th',
- 'thead',
- 'tr'
- )
- )
- ) {
-
-
- } else {
-
- $this->inBody($token);
- }
- }
- private function inColumnGroup($token)
- {
-
- if ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $text = $this->dom->createTextNode($token['data']);
- end($this->stack)->appendChild($text);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $comment = $this->dom->createComment($token['data']);
- end($this->stack)->appendChild($comment);
-
- } elseif ($token['type'] === HTML5::STARTTAG && $token['name'] === 'col') {
-
- $this->insertElement($token);
- array_pop($this->stack);
-
- } elseif ($token['type'] === HTML5::ENDTAG &&
- $token['name'] === 'colgroup'
- ) {
-
- if (end($this->stack)->nodeName === 'html') {
-
-
- } else {
- array_pop($this->stack);
- $this->mode = self::IN_TABLE;
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG && $token['name'] === 'col') {
-
-
- } else {
-
- $this->inColumnGroup(
- array(
- 'name' => 'colgroup',
- 'type' => HTML5::ENDTAG
- )
- );
- return $this->inTable($token);
- }
- }
- private function inTableBody($token)
- {
- $clear = array('tbody', 'tfoot', 'thead', 'html');
-
- if ($token['type'] === HTML5::STARTTAG && $token['name'] === 'tr') {
-
- $this->clearStackToTableContext($clear);
-
- $this->insertElement($token);
- $this->mode = self::IN_ROW;
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- ($token['name'] === 'th' || $token['name'] === 'td')
- ) {
-
- $this->inTableBody(
- array(
- 'name' => 'tr',
- 'type' => HTML5::STARTTAG,
- 'attr' => array()
- )
- );
- return $this->inRow($token);
-
- } elseif ($token['type'] === HTML5::ENDTAG &&
- in_array($token['name'], array('tbody', 'tfoot', 'thead'))
- ) {
-
- if (!$this->elementInScope($token['name'], true)) {
-
-
- } else {
-
- $this->clearStackToTableContext($clear);
-
- array_pop($this->stack);
- $this->mode = self::IN_TABLE;
- }
-
- } elseif (($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array('caption', 'col', 'colgroup', 'tbody', 'tfoor', 'thead')
- )) ||
- ($token['type'] === HTML5::STARTTAG && $token['name'] === 'table')
- ) {
-
- if (!$this->elementInScope(array('tbody', 'thead', 'tfoot'), true)) {
-
-
- } else {
-
- $this->clearStackToTableContext($clear);
-
- $this->inTableBody(
- array(
- 'name' => end($this->stack)->nodeName,
- 'type' => HTML5::ENDTAG
- )
- );
- return $this->mainPhase($token);
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG && in_array(
- $token['name'],
- array('body', 'caption', 'col', 'colgroup', 'html', 'td', 'th', 'tr')
- )
- ) {
-
-
- } else {
-
- $this->inTable($token);
- }
- }
- private function inRow($token)
- {
- $clear = array('tr', 'html');
-
- if ($token['type'] === HTML5::STARTTAG &&
- ($token['name'] === 'th' || $token['name'] === 'td')
- ) {
-
- $this->clearStackToTableContext($clear);
-
- $this->insertElement($token);
- $this->mode = self::IN_CELL;
-
- $this->a_formatting[] = self::MARKER;
-
- } elseif ($token['type'] === HTML5::ENDTAG && $token['name'] === 'tr') {
-
- if (!$this->elementInScope($token['name'], true)) {
-
-
- } else {
-
- $this->clearStackToTableContext($clear);
-
- array_pop($this->stack);
- $this->mode = self::IN_TBODY;
- }
-
- } elseif ($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array('caption', 'col', 'colgroup', 'tbody', 'tfoot', 'thead', 'tr')
- )
- ) {
-
- $this->inRow(
- array(
- 'name' => 'tr',
- 'type' => HTML5::ENDTAG
- )
- );
- return $this->inCell($token);
-
- } elseif ($token['type'] === HTML5::ENDTAG &&
- in_array($token['name'], array('tbody', 'tfoot', 'thead'))
- ) {
-
- if (!$this->elementInScope($token['name'], true)) {
-
-
- } else {
-
- $this->inRow(
- array(
- 'name' => 'tr',
- 'type' => HTML5::ENDTAG
- )
- );
- return $this->inCell($token);
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG && in_array(
- $token['name'],
- array('body', 'caption', 'col', 'colgroup', 'html', 'td', 'th', 'tr')
- )
- ) {
-
-
- } else {
-
- $this->inTable($token);
- }
- }
- private function inCell($token)
- {
-
- if ($token['type'] === HTML5::ENDTAG &&
- ($token['name'] === 'td' || $token['name'] === 'th')
- ) {
-
- if (!$this->elementInScope($token['name'], true)) {
-
-
- } else {
-
- $this->generateImpliedEndTags(array($token['name']));
-
-
-
- while (true) {
- $node = end($this->stack)->nodeName;
- array_pop($this->stack);
- if ($node === $token['name']) {
- break;
- }
- }
-
- $this->clearTheActiveFormattingElementsUpToTheLastMarker();
-
- $this->mode = self::IN_ROW;
- }
-
- } elseif ($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array(
- 'caption',
- 'col',
- 'colgroup',
- 'tbody',
- 'td',
- 'tfoot',
- 'th',
- 'thead',
- 'tr'
- )
- )
- ) {
-
- if (!$this->elementInScope(array('td', 'th'), true)) {
-
-
- } else {
- $this->closeCell();
- return $this->inRow($token);
- }
-
- } elseif ($token['type'] === HTML5::STARTTAG && in_array(
- $token['name'],
- array(
- 'caption',
- 'col',
- 'colgroup',
- 'tbody',
- 'td',
- 'tfoot',
- 'th',
- 'thead',
- 'tr'
- )
- )
- ) {
-
- if (!$this->elementInScope(array('td', 'th'), true)) {
-
-
- } else {
- $this->closeCell();
- return $this->inRow($token);
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG && in_array(
- $token['name'],
- array('body', 'caption', 'col', 'colgroup', 'html')
- )
- ) {
-
-
- } elseif ($token['type'] === HTML5::ENDTAG && in_array(
- $token['name'],
- array('table', 'tbody', 'tfoot', 'thead', 'tr')
- )
- ) {
-
- if (!$this->elementInScope($token['name'], true)) {
-
-
- } else {
- $this->closeCell();
- return $this->inRow($token);
- }
-
- } else {
-
- $this->inBody($token);
- }
- }
- private function inSelect($token)
- {
-
-
- if ($token['type'] === HTML5::CHARACTR) {
-
- $this->insertText($token['data']);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $this->insertComment($token['data']);
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- $token['name'] === 'option'
- ) {
-
- if (end($this->stack)->nodeName === 'option') {
- $this->inSelect(
- array(
- 'name' => 'option',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $this->insertElement($token);
-
- } elseif ($token['type'] === HTML5::STARTTAG &&
- $token['name'] === 'optgroup'
- ) {
-
- if (end($this->stack)->nodeName === 'option') {
- $this->inSelect(
- array(
- 'name' => 'option',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- if (end($this->stack)->nodeName === 'optgroup') {
- $this->inSelect(
- array(
- 'name' => 'optgroup',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- $this->insertElement($token);
-
- } elseif ($token['type'] === HTML5::ENDTAG &&
- $token['name'] === 'optgroup'
- ) {
-
- $elements_in_stack = count($this->stack);
- if ($this->stack[$elements_in_stack - 1]->nodeName === 'option' &&
- $this->stack[$elements_in_stack - 2]->nodeName === 'optgroup'
- ) {
- $this->inSelect(
- array(
- 'name' => 'option',
- 'type' => HTML5::ENDTAG
- )
- );
- }
-
- if ($this->stack[$elements_in_stack - 1] === 'optgroup') {
- array_pop($this->stack);
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG &&
- $token['name'] === 'option'
- ) {
-
- if (end($this->stack)->nodeName === 'option') {
- array_pop($this->stack);
- }
-
- } elseif ($token['type'] === HTML5::ENDTAG &&
- $token['name'] === 'select'
- ) {
-
- if (!$this->elementInScope($token['name'], true)) {
-
-
- } else {
-
- while (true) {
- $current = end($this->stack)->nodeName;
- array_pop($this->stack);
- if ($current === 'select') {
- break;
- }
- }
-
- $this->resetInsertionMode();
- }
-
- } elseif ($token['name'] === 'select' &&
- $token['type'] === HTML5::STARTTAG
- ) {
-
- $this->inSelect(
- array(
- 'name' => 'select',
- 'type' => HTML5::ENDTAG
- )
- );
-
- } elseif (in_array(
- $token['name'],
- array(
- 'caption',
- 'table',
- 'tbody',
- 'tfoot',
- 'thead',
- 'tr',
- 'td',
- 'th'
- )
- ) && $token['type'] === HTML5::ENDTAG
- ) {
-
-
-
- if ($this->elementInScope($token['name'], true)) {
- $this->inSelect(
- array(
- 'name' => 'select',
- 'type' => HTML5::ENDTAG
- )
- );
- $this->mainPhase($token);
- }
-
- } else {
-
- }
- }
- private function afterBody($token)
- {
-
-
- if ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $this->inBody($token);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $comment = $this->dom->createComment($token['data']);
- $this->stack[0]->appendChild($comment);
-
- } elseif ($token['type'] === HTML5::ENDTAG && $token['name'] === 'html') {
-
-
- $this->phase = self::END_PHASE;
-
- } else {
-
- $this->mode = self::IN_BODY;
- return $this->inBody($token);
- }
- }
- private function inFrameset($token)
- {
-
-
- if ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $this->insertText($token['data']);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $this->insertComment($token['data']);
-
- } elseif ($token['name'] === 'frameset' &&
- $token['type'] === HTML5::STARTTAG
- ) {
- $this->insertElement($token);
-
- } elseif ($token['name'] === 'frameset' &&
- $token['type'] === HTML5::ENDTAG
- ) {
-
- if (end($this->stack)->nodeName === 'html') {
-
- } else {
-
- array_pop($this->stack);
-
- $this->mode = self::AFTR_FRAME;
- }
-
- } elseif ($token['name'] === 'frame' &&
- $token['type'] === HTML5::STARTTAG
- ) {
-
- $this->insertElement($token);
-
- array_pop($this->stack);
-
- } elseif ($token['name'] === 'noframes' &&
- $token['type'] === HTML5::STARTTAG
- ) {
-
- $this->inBody($token);
-
- } else {
-
- }
- }
- private function afterFrameset($token)
- {
-
-
- if ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $this->insertText($token['data']);
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $this->insertComment($token['data']);
-
- } elseif ($token['name'] === 'html' &&
- $token['type'] === HTML5::ENDTAG
- ) {
-
- $this->phase = self::END_PHASE;
-
- } elseif ($token['name'] === 'noframes' &&
- $token['type'] === HTML5::STARTTAG
- ) {
-
- $this->inBody($token);
-
- } else {
-
- }
- }
- private function trailingEndPhase($token)
- {
-
-
- if ($token['type'] === HTML5::DOCTYPE) {
-
-
- } elseif ($token['type'] === HTML5::COMMENT) {
-
- $comment = $this->dom->createComment($token['data']);
- $this->dom->appendChild($comment);
-
- } elseif ($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])
- ) {
-
- $this->mainPhase($token);
-
- } elseif (($token['type'] === HTML5::CHARACTR &&
- preg_match('/^[\t\n\x0b\x0c ]+$/', $token['data'])) ||
- $token['type'] === HTML5::STARTTAG || $token['type'] === HTML5::ENDTAG
- ) {
-
- $this->phase = self::MAIN_PHASE;
- return $this->mainPhase($token);
-
- } elseif ($token['type'] === HTML5::EOF) {
-
- }
- }
- private function insertElement($token, $append = true, $check = false)
- {
-
- if ($check) {
-
-
- $token['name'] = preg_replace('/[^a-z0-9-]/i', '', $token['name']);
-
- $token['name'] = ltrim($token['name'], '-0..9');
-
- if ($token['name'] === '') {
- $token['name'] = 'span';
- }
- }
- $el = $this->dom->createElement($token['name']);
- foreach ($token['attr'] as $attr) {
- if (!$el->hasAttribute($attr['name'])) {
- $el->setAttribute($attr['name'], $attr['value']);
- }
- }
- $this->appendToRealParent($el);
- $this->stack[] = $el;
- return $el;
- }
- private function insertText($data)
- {
- $text = $this->dom->createTextNode($data);
- $this->appendToRealParent($text);
- }
- private function insertComment($data)
- {
- $comment = $this->dom->createComment($data);
- $this->appendToRealParent($comment);
- }
- private function appendToRealParent($node)
- {
- if ($this->foster_parent === null) {
- end($this->stack)->appendChild($node);
- } elseif ($this->foster_parent !== null) {
-
- for ($n = count($this->stack) - 1; $n >= 0; $n--) {
- if ($this->stack[$n]->nodeName === 'table' &&
- $this->stack[$n]->parentNode !== null
- ) {
- $table = $this->stack[$n];
- break;
- }
- }
- if (isset($table) && $this->foster_parent->isSameNode($table->parentNode)) {
- $this->foster_parent->insertBefore($node, $table);
- } else {
- $this->foster_parent->appendChild($node);
- }
- $this->foster_parent = null;
- }
- }
- private function elementInScope($el, $table = false)
- {
- if (is_array($el)) {
- foreach ($el as $element) {
- if ($this->elementInScope($element, $table)) {
- return true;
- }
- }
- return false;
- }
- $leng = count($this->stack);
- for ($n = 0; $n < $leng; $n++) {
-
- $node = $this->stack[$leng - 1 - $n];
- if ($node->tagName === $el) {
-
- return true;
- } elseif ($node->tagName === 'table') {
-
- return false;
- } elseif ($table === true && in_array(
- $node->tagName,
- array(
- 'caption',
- 'td',
- 'th',
- 'button',
- 'marquee',
- 'object'
- )
- )
- ) {
-
- return false;
- } elseif ($node === $node->ownerDocument->documentElement) {
-
- return false;
- }
-
- }
- }
- private function reconstructActiveFormattingElements()
- {
-
- $formatting_elements = count($this->a_formatting);
- if ($formatting_elements === 0) {
- return false;
- }
-
- $entry = end($this->a_formatting);
-
- if ($entry === self::MARKER || in_array($entry, $this->stack, true)) {
- return false;
- }
- for ($a = $formatting_elements - 1; $a >= 0; true) {
-
- if ($a === 0) {
- $step_seven = false;
- break;
- }
-
- $a--;
- $entry = $this->a_formatting[$a];
-
- if ($entry === self::MARKER || in_array($entry, $this->stack, true)) {
- break;
- }
- }
- while (true) {
-
- if (isset($step_seven) && $step_seven === true) {
- $a++;
- $entry = $this->a_formatting[$a];
- }
-
- $clone = $entry->cloneNode();
-
- end($this->stack)->appendChild($clone);
- $this->stack[] = $clone;
-
- $this->a_formatting[$a] = $clone;
-
- if (end($this->a_formatting) !== $clone) {
- $step_seven = true;
- } else {
- break;
- }
- }
- }
- private function clearTheActiveFormattingElementsUpToTheLastMarker()
- {
-
- while (true) {
-
- $entry = end($this->a_formatting);
-
- array_pop($this->a_formatting);
-
- if ($entry === self::MARKER) {
- break;
- }
- }
- }
- private function generateImpliedEndTags($exclude = array())
- {
-
- $node = end($this->stack);
- $elements = array_diff(array('dd', 'dt', 'li', 'p', 'td', 'th', 'tr'), $exclude);
- while (in_array(end($this->stack)->nodeName, $elements)) {
- array_pop($this->stack);
- }
- }
- private function getElementCategory($node)
- {
- $name = $node->tagName;
- if (in_array($name, $this->special)) {
- return self::SPECIAL;
- } elseif (in_array($name, $this->scoping)) {
- return self::SCOPING;
- } elseif (in_array($name, $this->formatting)) {
- return self::FORMATTING;
- } else {
- return self::PHRASING;
- }
- }
- private function clearStackToTableContext($elements)
- {
-
- while (true) {
- $node = end($this->stack)->nodeName;
- if (in_array($node, $elements)) {
- break;
- } else {
- array_pop($this->stack);
- }
- }
- }
- private function resetInsertionMode()
- {
-
- $last = false;
- $leng = count($this->stack);
- for ($n = $leng - 1; $n >= 0; $n--) {
-
- $node = $this->stack[$n];
-
- if ($this->stack[0]->isSameNode($node)) {
- $last = true;
- }
-
- if ($node->nodeName === 'select') {
- $this->mode = self::IN_SELECT;
- break;
-
- } elseif ($node->nodeName === 'td' || $node->nodeName === 'th') {
- $this->mode = self::IN_CELL;
- break;
-
- } elseif ($node->nodeName === 'tr') {
- $this->mode = self::IN_ROW;
- break;
-
- } elseif (in_array($node->nodeName, array('tbody', 'thead', 'tfoot'))) {
- $this->mode = self::IN_TBODY;
- break;
-
- } elseif ($node->nodeName === 'caption') {
- $this->mode = self::IN_CAPTION;
- break;
-
- } elseif ($node->nodeName === 'colgroup') {
- $this->mode = self::IN_CGROUP;
- break;
-
- } elseif ($node->nodeName === 'table') {
- $this->mode = self::IN_TABLE;
- break;
-
- } elseif ($node->nodeName === 'head') {
- $this->mode = self::IN_BODY;
- break;
-
- } elseif ($node->nodeName === 'body') {
- $this->mode = self::IN_BODY;
- break;
-
- } elseif ($node->nodeName === 'frameset') {
- $this->mode = self::IN_FRAME;
- break;
-
- } elseif ($node->nodeName === 'html') {
- $this->mode = ($this->head_pointer === null)
- ? self::BEFOR_HEAD
- : self::AFTER_HEAD;
- break;
-
- } elseif ($last) {
- $this->mode = self::IN_BODY;
- break;
- }
- }
- }
- private function closeCell()
- {
-
- foreach (array('td', 'th') as $cell) {
- if ($this->elementInScope($cell, true)) {
- $this->inCell(
- array(
- 'name' => $cell,
- 'type' => HTML5::ENDTAG
- )
- );
- break;
- }
- }
- }
- public function save()
- {
- return $this->dom;
- }
- }
|