HTMLTokenizer.cpp 91 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522252325242525252625272528252925302531253225332534253525362537253825392540254125422543254425452546254725482549255025512552255325542555255625572558255925602561256225632564256525662567256825692570257125722573257425752576257725782579258025812582258325842585258625872588258925902591259225932594259525962597259825992600260126022603260426052606260726082609261026112612261326142615261626172618261926202621262226232624262526262627262826292630263126322633263426352636263726382639264026412642264326442645264626472648264926502651265226532654265526562657265826592660266126622663266426652666266726682669267026712672267326742675267626772678267926802681268226832684268526862687
  1. /*
  2. * Copyright (c) 2020, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/Debug.h>
  7. #include <AK/SourceLocation.h>
  8. #include <LibTextCodec/Decoder.h>
  9. #include <LibWeb/HTML/Parser/Entities.h>
  10. #include <LibWeb/HTML/Parser/HTMLToken.h>
  11. #include <LibWeb/HTML/Parser/HTMLTokenizer.h>
  12. #include <ctype.h>
  13. #include <string.h>
  14. namespace Web::HTML {
  15. #pragma GCC diagnostic ignored "-Wunused-label"
  16. #define CONSUME_NEXT_INPUT_CHARACTER \
  17. current_input_character = next_code_point();
  18. #define SWITCH_TO(new_state) \
  19. do { \
  20. will_switch_to(State::new_state); \
  21. m_state = State::new_state; \
  22. CONSUME_NEXT_INPUT_CHARACTER; \
  23. goto new_state; \
  24. } while (0)
  25. #define RECONSUME_IN(new_state) \
  26. do { \
  27. will_reconsume_in(State::new_state); \
  28. m_state = State::new_state; \
  29. goto new_state; \
  30. } while (0)
  31. #define SWITCH_TO_RETURN_STATE \
  32. do { \
  33. will_switch_to(m_return_state); \
  34. m_state = m_return_state; \
  35. goto _StartOfFunction; \
  36. } while (0)
  37. #define RECONSUME_IN_RETURN_STATE \
  38. do { \
  39. will_reconsume_in(m_return_state); \
  40. m_state = m_return_state; \
  41. if (current_input_character.has_value()) \
  42. restore_to(m_prev_utf8_iterator); \
  43. goto _StartOfFunction; \
  44. } while (0)
  45. #define SWITCH_TO_AND_EMIT_CURRENT_TOKEN(new_state) \
  46. do { \
  47. will_switch_to(State::new_state); \
  48. m_state = State::new_state; \
  49. will_emit(m_current_token); \
  50. m_queued_tokens.enqueue(m_current_token); \
  51. return m_queued_tokens.dequeue(); \
  52. } while (0)
  53. #define EMIT_CHARACTER_AND_RECONSUME_IN(code_point, new_state) \
  54. do { \
  55. m_queued_tokens.enqueue(HTMLToken::make_character(code_point)); \
  56. will_reconsume_in(State::new_state); \
  57. m_state = State::new_state; \
  58. goto new_state; \
  59. } while (0)
  60. #define FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE \
  61. do { \
  62. for (auto code_point : m_temporary_buffer) { \
  63. if (consumed_as_part_of_an_attribute()) { \
  64. m_current_token.m_tag.attributes.last().value_builder.append_code_point(code_point); \
  65. } else { \
  66. create_new_token(HTMLToken::Type::Character); \
  67. m_current_token.m_comment_or_character.data.append_code_point(code_point); \
  68. m_queued_tokens.enqueue(m_current_token); \
  69. } \
  70. } \
  71. } while (0)
  72. #define DONT_CONSUME_NEXT_INPUT_CHARACTER \
  73. do { \
  74. restore_to(m_prev_utf8_iterator); \
  75. } while (0)
  76. #define ON(code_point) \
  77. if (current_input_character.has_value() && current_input_character.value() == code_point)
  78. #define ON_EOF \
  79. if (!current_input_character.has_value())
  80. #define ON_ASCII_ALPHA \
  81. if (current_input_character.has_value() && isalpha(current_input_character.value()))
  82. #define ON_ASCII_ALPHANUMERIC \
  83. if (current_input_character.has_value() && isalnum(current_input_character.value()))
  84. #define ON_ASCII_UPPER_ALPHA \
  85. if (current_input_character.has_value() && current_input_character.value() >= 'A' && current_input_character.value() <= 'Z')
  86. #define ON_ASCII_LOWER_ALPHA \
  87. if (current_input_character.has_value() && current_input_character.value() >= 'a' && current_input_character.value() <= 'z')
  88. #define ON_ASCII_DIGIT \
  89. if (current_input_character.has_value() && isdigit(current_input_character.value()))
  90. #define ON_ASCII_HEX_DIGIT \
  91. if (current_input_character.has_value() && isxdigit(current_input_character.value()))
  92. #define ON_WHITESPACE \
  93. if (current_input_character.has_value() && strchr("\t\n\f ", current_input_character.value()))
  94. #define ANYTHING_ELSE if (1)
  95. #define EMIT_EOF \
  96. do { \
  97. if (m_has_emitted_eof) \
  98. return {}; \
  99. m_has_emitted_eof = true; \
  100. create_new_token(HTMLToken::Type::EndOfFile); \
  101. will_emit(m_current_token); \
  102. m_queued_tokens.enqueue(m_current_token); \
  103. return m_queued_tokens.dequeue(); \
  104. } while (0)
  105. #define EMIT_CURRENT_TOKEN \
  106. do { \
  107. will_emit(m_current_token); \
  108. m_queued_tokens.enqueue(m_current_token); \
  109. return m_queued_tokens.dequeue(); \
  110. } while (0)
  111. #define EMIT_CHARACTER(code_point) \
  112. do { \
  113. create_new_token(HTMLToken::Type::Character); \
  114. m_current_token.m_comment_or_character.data.append_code_point(code_point); \
  115. m_queued_tokens.enqueue(m_current_token); \
  116. return m_queued_tokens.dequeue(); \
  117. } while (0)
  118. #define EMIT_CURRENT_CHARACTER \
  119. EMIT_CHARACTER(current_input_character.value());
  120. #define SWITCH_TO_AND_EMIT_CHARACTER(code_point, new_state) \
  121. do { \
  122. will_switch_to(State::new_state); \
  123. m_state = State::new_state; \
  124. EMIT_CHARACTER(code_point); \
  125. } while (0)
  126. #define SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(new_state) \
  127. SWITCH_TO_AND_EMIT_CHARACTER(current_input_character.value(), new_state)
  128. #define BEGIN_STATE(state) \
  129. state: \
  130. case State::state: { \
  131. { \
  132. {
  133. #define END_STATE \
  134. VERIFY_NOT_REACHED(); \
  135. break; \
  136. } \
  137. } \
  138. }
  139. static inline void log_parse_error(const SourceLocation& location = SourceLocation::current())
  140. {
  141. dbgln_if(TOKENIZER_TRACE_DEBUG, "Parse error (tokenization) {}", location);
  142. }
  143. static inline bool is_surrogate(u32 code_point)
  144. {
  145. return (code_point & 0xfffff800) == 0xd800;
  146. }
  147. static inline bool is_noncharacter(u32 code_point)
  148. {
  149. return code_point >= 0xfdd0 && (code_point <= 0xfdef || (code_point & 0xfffe) == 0xfffe) && code_point <= 0x10ffff;
  150. }
  151. static inline bool is_c0_control(u32 code_point)
  152. {
  153. return code_point <= 0x1f;
  154. }
  155. static inline bool is_control(u32 code_point)
  156. {
  157. return is_c0_control(code_point) || (code_point >= 0x7f && code_point <= 0x9f);
  158. }
  159. Optional<u32> HTMLTokenizer::next_code_point()
  160. {
  161. if (m_utf8_iterator == m_utf8_view.end())
  162. return {};
  163. skip(1);
  164. dbgln_if(TOKENIZER_TRACE_DEBUG, "(Tokenizer) Next code_point: {}", (char)*m_prev_utf8_iterator);
  165. return *m_prev_utf8_iterator;
  166. }
  167. void HTMLTokenizer::skip(size_t count)
  168. {
  169. m_prev_utf8_iterator = m_utf8_iterator;
  170. m_source_positions.append(m_source_positions.last());
  171. for (size_t i = 0; i < count; ++i) {
  172. auto code_point = *m_utf8_iterator;
  173. if (code_point == '\n') {
  174. m_source_positions.last().column = 0;
  175. m_source_positions.last().line++;
  176. } else {
  177. m_source_positions.last().column++;
  178. }
  179. ++m_utf8_iterator;
  180. }
  181. }
  182. Optional<u32> HTMLTokenizer::peek_code_point(size_t offset) const
  183. {
  184. auto it = m_utf8_iterator;
  185. for (size_t i = 0; i < offset && it != m_utf8_view.end(); ++i)
  186. ++it;
  187. if (it == m_utf8_view.end())
  188. return {};
  189. return *it;
  190. }
  191. Optional<HTMLToken> HTMLTokenizer::next_token()
  192. {
  193. _StartOfFunction:
  194. if (!m_queued_tokens.is_empty())
  195. return m_queued_tokens.dequeue();
  196. for (;;) {
  197. auto current_input_character = next_code_point();
  198. switch (m_state) {
  199. BEGIN_STATE(Data)
  200. {
  201. ON('&')
  202. {
  203. m_return_state = State::Data;
  204. SWITCH_TO(CharacterReference);
  205. }
  206. ON('<')
  207. {
  208. SWITCH_TO(TagOpen);
  209. }
  210. ON(0)
  211. {
  212. log_parse_error();
  213. EMIT_CURRENT_CHARACTER;
  214. }
  215. ON_EOF
  216. {
  217. EMIT_EOF;
  218. }
  219. ANYTHING_ELSE
  220. {
  221. EMIT_CURRENT_CHARACTER;
  222. }
  223. }
  224. END_STATE
  225. BEGIN_STATE(TagOpen)
  226. {
  227. ON('!')
  228. {
  229. SWITCH_TO(MarkupDeclarationOpen);
  230. }
  231. ON('/')
  232. {
  233. SWITCH_TO(EndTagOpen);
  234. }
  235. ON_ASCII_ALPHA
  236. {
  237. create_new_token(HTMLToken::Type::StartTag);
  238. RECONSUME_IN(TagName);
  239. }
  240. ON('?')
  241. {
  242. log_parse_error();
  243. create_new_token(HTMLToken::Type::Comment);
  244. RECONSUME_IN(BogusComment);
  245. }
  246. ON_EOF
  247. {
  248. log_parse_error();
  249. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  250. EMIT_EOF;
  251. }
  252. ANYTHING_ELSE
  253. {
  254. log_parse_error();
  255. EMIT_CHARACTER_AND_RECONSUME_IN('<', Data);
  256. }
  257. }
  258. END_STATE
  259. BEGIN_STATE(TagName)
  260. {
  261. ON_WHITESPACE
  262. {
  263. m_current_token.m_end_position = nth_last_position(1);
  264. SWITCH_TO(BeforeAttributeName);
  265. }
  266. ON('/')
  267. {
  268. m_current_token.m_end_position = nth_last_position(1);
  269. SWITCH_TO(SelfClosingStartTag);
  270. }
  271. ON('>')
  272. {
  273. m_current_token.m_end_position = nth_last_position(1);
  274. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  275. }
  276. ON_ASCII_UPPER_ALPHA
  277. {
  278. m_current_token.m_tag.tag_name.append(tolower(current_input_character.value()));
  279. m_current_token.m_end_position = nth_last_position(0);
  280. continue;
  281. }
  282. ON(0)
  283. {
  284. log_parse_error();
  285. m_current_token.m_tag.tag_name.append_code_point(0xFFFD);
  286. m_current_token.m_end_position = nth_last_position(0);
  287. continue;
  288. }
  289. ON_EOF
  290. {
  291. log_parse_error();
  292. m_current_token.m_end_position = nth_last_position(1);
  293. EMIT_EOF;
  294. }
  295. ANYTHING_ELSE
  296. {
  297. m_current_token.m_tag.tag_name.append_code_point(current_input_character.value());
  298. m_current_token.m_end_position = nth_last_position(0);
  299. continue;
  300. }
  301. }
  302. END_STATE
  303. BEGIN_STATE(EndTagOpen)
  304. {
  305. ON_ASCII_ALPHA
  306. {
  307. create_new_token(HTMLToken::Type::EndTag);
  308. RECONSUME_IN(TagName);
  309. }
  310. ON('>')
  311. {
  312. log_parse_error();
  313. SWITCH_TO(Data);
  314. }
  315. ON_EOF
  316. {
  317. log_parse_error();
  318. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  319. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  320. EMIT_EOF;
  321. }
  322. ANYTHING_ELSE
  323. {
  324. log_parse_error();
  325. create_new_token(HTMLToken::Type::Comment);
  326. RECONSUME_IN(BogusComment);
  327. }
  328. }
  329. END_STATE
  330. BEGIN_STATE(MarkupDeclarationOpen)
  331. {
  332. DONT_CONSUME_NEXT_INPUT_CHARACTER;
  333. if (consume_next_if_match("--")) {
  334. create_new_token(HTMLToken::Type::Comment);
  335. SWITCH_TO(CommentStart);
  336. }
  337. if (consume_next_if_match("DOCTYPE", CaseSensitivity::CaseInsensitive)) {
  338. SWITCH_TO(DOCTYPE);
  339. }
  340. if (consume_next_if_match("[CDATA[")) {
  341. TODO();
  342. }
  343. ANYTHING_ELSE
  344. {
  345. log_parse_error();
  346. create_new_token(HTMLToken::Type::Comment);
  347. SWITCH_TO(BogusComment);
  348. }
  349. }
  350. END_STATE
  351. BEGIN_STATE(BogusComment)
  352. {
  353. ON('>')
  354. {
  355. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  356. }
  357. ON_EOF
  358. {
  359. m_queued_tokens.enqueue(m_current_token);
  360. EMIT_EOF;
  361. }
  362. ON(0)
  363. {
  364. log_parse_error();
  365. m_current_token.m_comment_or_character.data.append_code_point(0xFFFD);
  366. continue;
  367. }
  368. ANYTHING_ELSE
  369. {
  370. m_current_token.m_comment_or_character.data.append_code_point(current_input_character.value());
  371. continue;
  372. }
  373. }
  374. END_STATE
  375. BEGIN_STATE(DOCTYPE)
  376. {
  377. ON_WHITESPACE
  378. {
  379. SWITCH_TO(BeforeDOCTYPEName);
  380. }
  381. ON('>')
  382. {
  383. RECONSUME_IN(BeforeDOCTYPEName);
  384. }
  385. ON_EOF
  386. {
  387. log_parse_error();
  388. create_new_token(HTMLToken::Type::DOCTYPE);
  389. m_current_token.m_doctype.force_quirks = true;
  390. m_queued_tokens.enqueue(m_current_token);
  391. EMIT_EOF;
  392. }
  393. ANYTHING_ELSE
  394. {
  395. log_parse_error();
  396. RECONSUME_IN(BeforeDOCTYPEName);
  397. }
  398. }
  399. END_STATE
  400. BEGIN_STATE(BeforeDOCTYPEName)
  401. {
  402. ON_WHITESPACE
  403. {
  404. continue;
  405. }
  406. ON_ASCII_UPPER_ALPHA
  407. {
  408. create_new_token(HTMLToken::Type::DOCTYPE);
  409. m_current_token.m_doctype.name.append(tolower(current_input_character.value()));
  410. m_current_token.m_doctype.missing_name = false;
  411. SWITCH_TO(DOCTYPEName);
  412. }
  413. ON(0)
  414. {
  415. log_parse_error();
  416. create_new_token(HTMLToken::Type::DOCTYPE);
  417. m_current_token.m_doctype.name.append_code_point(0xFFFD);
  418. m_current_token.m_doctype.missing_name = false;
  419. SWITCH_TO(DOCTYPEName);
  420. }
  421. ON('>')
  422. {
  423. log_parse_error();
  424. create_new_token(HTMLToken::Type::DOCTYPE);
  425. m_current_token.m_doctype.force_quirks = true;
  426. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  427. }
  428. ON_EOF
  429. {
  430. log_parse_error();
  431. create_new_token(HTMLToken::Type::DOCTYPE);
  432. m_current_token.m_doctype.force_quirks = true;
  433. m_queued_tokens.enqueue(m_current_token);
  434. EMIT_EOF;
  435. }
  436. ANYTHING_ELSE
  437. {
  438. create_new_token(HTMLToken::Type::DOCTYPE);
  439. m_current_token.m_doctype.name.append_code_point(current_input_character.value());
  440. m_current_token.m_doctype.missing_name = false;
  441. SWITCH_TO(DOCTYPEName);
  442. }
  443. }
  444. END_STATE
  445. BEGIN_STATE(DOCTYPEName)
  446. {
  447. ON_WHITESPACE
  448. {
  449. SWITCH_TO(AfterDOCTYPEName);
  450. }
  451. ON('>')
  452. {
  453. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  454. }
  455. ON_ASCII_UPPER_ALPHA
  456. {
  457. m_current_token.m_doctype.name.append(tolower(current_input_character.value()));
  458. continue;
  459. }
  460. ON(0)
  461. {
  462. log_parse_error();
  463. m_current_token.m_doctype.name.append_code_point(0xFFFD);
  464. continue;
  465. }
  466. ON_EOF
  467. {
  468. log_parse_error();
  469. m_current_token.m_doctype.force_quirks = true;
  470. m_queued_tokens.enqueue(m_current_token);
  471. EMIT_EOF;
  472. }
  473. ANYTHING_ELSE
  474. {
  475. m_current_token.m_doctype.name.append_code_point(current_input_character.value());
  476. continue;
  477. }
  478. }
  479. END_STATE
  480. BEGIN_STATE(AfterDOCTYPEName)
  481. {
  482. ON_WHITESPACE
  483. {
  484. continue;
  485. }
  486. ON('>')
  487. {
  488. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  489. }
  490. ON_EOF
  491. {
  492. log_parse_error();
  493. m_current_token.m_doctype.force_quirks = true;
  494. m_queued_tokens.enqueue(m_current_token);
  495. EMIT_EOF;
  496. }
  497. ANYTHING_ELSE
  498. {
  499. if (toupper(current_input_character.value()) == 'P' && consume_next_if_match("UBLIC", CaseSensitivity::CaseInsensitive)) {
  500. SWITCH_TO(AfterDOCTYPEPublicKeyword);
  501. }
  502. if (toupper(current_input_character.value()) == 'S' && consume_next_if_match("YSTEM", CaseSensitivity::CaseInsensitive)) {
  503. SWITCH_TO(AfterDOCTYPESystemKeyword);
  504. }
  505. log_parse_error();
  506. m_current_token.m_doctype.force_quirks = true;
  507. RECONSUME_IN(BogusDOCTYPE);
  508. }
  509. }
  510. END_STATE
  511. BEGIN_STATE(AfterDOCTYPEPublicKeyword)
  512. {
  513. ON_WHITESPACE
  514. {
  515. SWITCH_TO(BeforeDOCTYPEPublicIdentifier);
  516. }
  517. ON('"')
  518. {
  519. log_parse_error();
  520. m_current_token.m_doctype.public_identifier.clear();
  521. m_current_token.m_doctype.missing_public_identifier = false;
  522. SWITCH_TO(DOCTYPEPublicIdentifierDoubleQuoted);
  523. }
  524. ON('\'')
  525. {
  526. log_parse_error();
  527. m_current_token.m_doctype.public_identifier.clear();
  528. m_current_token.m_doctype.missing_public_identifier = false;
  529. SWITCH_TO(DOCTYPEPublicIdentifierSingleQuoted);
  530. }
  531. ON('>')
  532. {
  533. log_parse_error();
  534. m_current_token.m_doctype.force_quirks = true;
  535. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  536. }
  537. ON_EOF
  538. {
  539. log_parse_error();
  540. m_current_token.m_doctype.force_quirks = true;
  541. m_queued_tokens.enqueue(m_current_token);
  542. EMIT_EOF;
  543. }
  544. ANYTHING_ELSE
  545. {
  546. log_parse_error();
  547. m_current_token.m_doctype.force_quirks = true;
  548. RECONSUME_IN(BogusDOCTYPE);
  549. }
  550. }
  551. END_STATE
  552. BEGIN_STATE(AfterDOCTYPESystemKeyword)
  553. {
  554. ON_WHITESPACE
  555. {
  556. SWITCH_TO(BeforeDOCTYPESystemIdentifier);
  557. }
  558. ON('"')
  559. {
  560. log_parse_error();
  561. m_current_token.m_doctype.system_identifier.clear();
  562. m_current_token.m_doctype.missing_system_identifier = false;
  563. SWITCH_TO(DOCTYPESystemIdentifierDoubleQuoted);
  564. }
  565. ON('\'')
  566. {
  567. log_parse_error();
  568. m_current_token.m_doctype.system_identifier.clear();
  569. m_current_token.m_doctype.missing_system_identifier = false;
  570. SWITCH_TO(DOCTYPESystemIdentifierSingleQuoted);
  571. }
  572. ON('>')
  573. {
  574. log_parse_error();
  575. m_current_token.m_doctype.force_quirks = true;
  576. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  577. }
  578. ON_EOF
  579. {
  580. log_parse_error();
  581. m_current_token.m_doctype.force_quirks = true;
  582. m_queued_tokens.enqueue(m_current_token);
  583. EMIT_EOF;
  584. }
  585. ANYTHING_ELSE
  586. {
  587. log_parse_error();
  588. m_current_token.m_doctype.force_quirks = true;
  589. RECONSUME_IN(BogusDOCTYPE);
  590. }
  591. }
  592. END_STATE
  593. BEGIN_STATE(BeforeDOCTYPEPublicIdentifier)
  594. {
  595. ON_WHITESPACE
  596. {
  597. continue;
  598. }
  599. ON('"')
  600. {
  601. m_current_token.m_doctype.public_identifier.clear();
  602. m_current_token.m_doctype.missing_public_identifier = false;
  603. SWITCH_TO(DOCTYPEPublicIdentifierDoubleQuoted);
  604. }
  605. ON('\'')
  606. {
  607. m_current_token.m_doctype.public_identifier.clear();
  608. m_current_token.m_doctype.missing_public_identifier = false;
  609. SWITCH_TO(DOCTYPEPublicIdentifierSingleQuoted);
  610. }
  611. ON('>')
  612. {
  613. log_parse_error();
  614. m_current_token.m_doctype.force_quirks = true;
  615. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  616. }
  617. ON_EOF
  618. {
  619. log_parse_error();
  620. m_current_token.m_doctype.force_quirks = true;
  621. m_queued_tokens.enqueue(m_current_token);
  622. EMIT_EOF;
  623. }
  624. ANYTHING_ELSE
  625. {
  626. log_parse_error();
  627. m_current_token.m_doctype.force_quirks = true;
  628. RECONSUME_IN(BogusDOCTYPE);
  629. }
  630. }
  631. END_STATE
  632. BEGIN_STATE(BeforeDOCTYPESystemIdentifier)
  633. {
  634. ON_WHITESPACE
  635. {
  636. continue;
  637. }
  638. ON('"')
  639. {
  640. m_current_token.m_doctype.system_identifier.clear();
  641. m_current_token.m_doctype.missing_system_identifier = false;
  642. SWITCH_TO(DOCTYPESystemIdentifierDoubleQuoted);
  643. }
  644. ON('\'')
  645. {
  646. m_current_token.m_doctype.system_identifier.clear();
  647. m_current_token.m_doctype.missing_system_identifier = false;
  648. SWITCH_TO(DOCTYPESystemIdentifierSingleQuoted);
  649. }
  650. ON('>')
  651. {
  652. log_parse_error();
  653. m_current_token.m_doctype.force_quirks = true;
  654. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  655. }
  656. ON_EOF
  657. {
  658. log_parse_error();
  659. m_current_token.m_doctype.force_quirks = true;
  660. m_queued_tokens.enqueue(m_current_token);
  661. EMIT_EOF;
  662. }
  663. ANYTHING_ELSE
  664. {
  665. log_parse_error();
  666. m_current_token.m_doctype.force_quirks = true;
  667. RECONSUME_IN(BogusDOCTYPE);
  668. }
  669. }
  670. END_STATE
  671. BEGIN_STATE(DOCTYPEPublicIdentifierDoubleQuoted)
  672. {
  673. ON('"')
  674. {
  675. SWITCH_TO(AfterDOCTYPEPublicIdentifier);
  676. }
  677. ON(0)
  678. {
  679. log_parse_error();
  680. m_current_token.m_doctype.public_identifier.append_code_point(0xFFFD);
  681. continue;
  682. }
  683. ON('>')
  684. {
  685. log_parse_error();
  686. m_current_token.m_doctype.force_quirks = true;
  687. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  688. }
  689. ON_EOF
  690. {
  691. log_parse_error();
  692. m_current_token.m_doctype.force_quirks = true;
  693. m_queued_tokens.enqueue(m_current_token);
  694. EMIT_EOF;
  695. }
  696. ANYTHING_ELSE
  697. {
  698. m_current_token.m_doctype.public_identifier.append_code_point(current_input_character.value());
  699. continue;
  700. }
  701. }
  702. END_STATE
  703. BEGIN_STATE(DOCTYPEPublicIdentifierSingleQuoted)
  704. {
  705. ON('\'')
  706. {
  707. SWITCH_TO(AfterDOCTYPEPublicIdentifier);
  708. }
  709. ON(0)
  710. {
  711. log_parse_error();
  712. m_current_token.m_doctype.public_identifier.append_code_point(0xFFFD);
  713. continue;
  714. }
  715. ON('>')
  716. {
  717. log_parse_error();
  718. m_current_token.m_doctype.force_quirks = true;
  719. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  720. }
  721. ON_EOF
  722. {
  723. log_parse_error();
  724. m_current_token.m_doctype.force_quirks = true;
  725. m_queued_tokens.enqueue(m_current_token);
  726. EMIT_EOF;
  727. }
  728. ANYTHING_ELSE
  729. {
  730. m_current_token.m_doctype.public_identifier.append_code_point(current_input_character.value());
  731. continue;
  732. }
  733. }
  734. END_STATE
  735. BEGIN_STATE(DOCTYPESystemIdentifierDoubleQuoted)
  736. {
  737. ON('"')
  738. {
  739. SWITCH_TO(AfterDOCTYPESystemIdentifier);
  740. }
  741. ON(0)
  742. {
  743. log_parse_error();
  744. m_current_token.m_doctype.system_identifier.append_code_point(0xFFFD);
  745. continue;
  746. }
  747. ON('>')
  748. {
  749. log_parse_error();
  750. m_current_token.m_doctype.force_quirks = true;
  751. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  752. }
  753. ON_EOF
  754. {
  755. log_parse_error();
  756. m_current_token.m_doctype.force_quirks = true;
  757. m_queued_tokens.enqueue(m_current_token);
  758. EMIT_EOF;
  759. }
  760. ANYTHING_ELSE
  761. {
  762. m_current_token.m_doctype.system_identifier.append_code_point(current_input_character.value());
  763. continue;
  764. }
  765. }
  766. END_STATE
  767. BEGIN_STATE(DOCTYPESystemIdentifierSingleQuoted)
  768. {
  769. ON('\'')
  770. {
  771. SWITCH_TO(AfterDOCTYPESystemIdentifier);
  772. }
  773. ON(0)
  774. {
  775. log_parse_error();
  776. m_current_token.m_doctype.system_identifier.append_code_point(0xFFFD);
  777. continue;
  778. }
  779. ON('>')
  780. {
  781. log_parse_error();
  782. m_current_token.m_doctype.force_quirks = true;
  783. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  784. }
  785. ON_EOF
  786. {
  787. log_parse_error();
  788. m_current_token.m_doctype.force_quirks = true;
  789. m_queued_tokens.enqueue(m_current_token);
  790. EMIT_EOF;
  791. }
  792. ANYTHING_ELSE
  793. {
  794. m_current_token.m_doctype.system_identifier.append_code_point(current_input_character.value());
  795. continue;
  796. }
  797. }
  798. END_STATE
  799. BEGIN_STATE(AfterDOCTYPEPublicIdentifier)
  800. {
  801. ON_WHITESPACE
  802. {
  803. SWITCH_TO(BetweenDOCTYPEPublicAndSystemIdentifiers);
  804. }
  805. ON('>')
  806. {
  807. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  808. }
  809. ON('"')
  810. {
  811. log_parse_error();
  812. m_current_token.m_doctype.system_identifier.clear();
  813. m_current_token.m_doctype.missing_system_identifier = false;
  814. SWITCH_TO(DOCTYPESystemIdentifierDoubleQuoted);
  815. }
  816. ON('\'')
  817. {
  818. log_parse_error();
  819. m_current_token.m_doctype.system_identifier.clear();
  820. m_current_token.m_doctype.missing_system_identifier = false;
  821. SWITCH_TO(DOCTYPESystemIdentifierSingleQuoted);
  822. }
  823. ON_EOF
  824. {
  825. log_parse_error();
  826. m_current_token.m_doctype.force_quirks = true;
  827. m_queued_tokens.enqueue(m_current_token);
  828. EMIT_EOF;
  829. }
  830. ANYTHING_ELSE
  831. {
  832. log_parse_error();
  833. m_current_token.m_doctype.force_quirks = true;
  834. RECONSUME_IN(BogusDOCTYPE);
  835. }
  836. }
  837. END_STATE
  838. BEGIN_STATE(BetweenDOCTYPEPublicAndSystemIdentifiers)
  839. {
  840. ON_WHITESPACE
  841. {
  842. continue;
  843. }
  844. ON('>')
  845. {
  846. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  847. }
  848. ON('"')
  849. {
  850. m_current_token.m_doctype.system_identifier.clear();
  851. m_current_token.m_doctype.missing_system_identifier = false;
  852. SWITCH_TO(DOCTYPESystemIdentifierDoubleQuoted);
  853. }
  854. ON('\'')
  855. {
  856. m_current_token.m_doctype.system_identifier.clear();
  857. m_current_token.m_doctype.missing_system_identifier = false;
  858. SWITCH_TO(DOCTYPESystemIdentifierSingleQuoted);
  859. }
  860. ON_EOF
  861. {
  862. log_parse_error();
  863. m_current_token.m_doctype.force_quirks = true;
  864. m_queued_tokens.enqueue(m_current_token);
  865. EMIT_EOF;
  866. }
  867. ANYTHING_ELSE
  868. {
  869. log_parse_error();
  870. m_current_token.m_doctype.force_quirks = true;
  871. RECONSUME_IN(BogusDOCTYPE);
  872. }
  873. }
  874. END_STATE
  875. BEGIN_STATE(AfterDOCTYPESystemIdentifier)
  876. {
  877. ON_WHITESPACE
  878. {
  879. continue;
  880. }
  881. ON('>')
  882. {
  883. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  884. }
  885. ON_EOF
  886. {
  887. log_parse_error();
  888. m_current_token.m_doctype.force_quirks = true;
  889. m_queued_tokens.enqueue(m_current_token);
  890. EMIT_EOF;
  891. }
  892. ANYTHING_ELSE
  893. {
  894. log_parse_error();
  895. RECONSUME_IN(BogusDOCTYPE);
  896. }
  897. }
  898. END_STATE
  899. BEGIN_STATE(BogusDOCTYPE)
  900. {
  901. ON('>')
  902. {
  903. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  904. }
  905. ON(0)
  906. {
  907. log_parse_error();
  908. continue;
  909. }
  910. ON_EOF
  911. {
  912. m_queued_tokens.enqueue(m_current_token);
  913. EMIT_EOF;
  914. }
  915. ANYTHING_ELSE
  916. {
  917. continue;
  918. }
  919. }
  920. END_STATE
  921. BEGIN_STATE(BeforeAttributeName)
  922. {
  923. ON_WHITESPACE
  924. {
  925. continue;
  926. }
  927. ON('/')
  928. {
  929. if (!m_current_token.m_tag.attributes.is_empty())
  930. m_current_token.m_tag.attributes.last().name_end_position = nth_last_position(1);
  931. RECONSUME_IN(AfterAttributeName);
  932. }
  933. ON('>')
  934. {
  935. RECONSUME_IN(AfterAttributeName);
  936. }
  937. ON_EOF
  938. {
  939. RECONSUME_IN(AfterAttributeName);
  940. }
  941. ON('=')
  942. {
  943. log_parse_error();
  944. auto new_attribute = HTMLToken::AttributeBuilder();
  945. new_attribute.name_start_position = nth_last_position(1);
  946. new_attribute.local_name_builder.append_code_point(current_input_character.value());
  947. m_current_token.m_tag.attributes.append(new_attribute);
  948. SWITCH_TO(AttributeName);
  949. }
  950. ANYTHING_ELSE
  951. {
  952. auto new_attribute = HTMLToken::AttributeBuilder();
  953. new_attribute.name_start_position = nth_last_position(1);
  954. m_current_token.m_tag.attributes.append(move(new_attribute));
  955. RECONSUME_IN(AttributeName);
  956. }
  957. }
  958. END_STATE
  959. BEGIN_STATE(SelfClosingStartTag)
  960. {
  961. ON('>')
  962. {
  963. m_current_token.m_tag.self_closing = true;
  964. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  965. }
  966. ON_EOF
  967. {
  968. log_parse_error();
  969. EMIT_EOF;
  970. }
  971. ANYTHING_ELSE
  972. {
  973. log_parse_error();
  974. RECONSUME_IN(BeforeAttributeName);
  975. }
  976. }
  977. END_STATE
  978. BEGIN_STATE(AttributeName)
  979. {
  980. ON_WHITESPACE
  981. {
  982. RECONSUME_IN(AfterAttributeName);
  983. }
  984. ON('/')
  985. {
  986. RECONSUME_IN(AfterAttributeName);
  987. }
  988. ON('>')
  989. {
  990. RECONSUME_IN(AfterAttributeName);
  991. }
  992. ON_EOF
  993. {
  994. RECONSUME_IN(AfterAttributeName);
  995. }
  996. ON('=')
  997. {
  998. SWITCH_TO(BeforeAttributeValue);
  999. }
  1000. ON_ASCII_UPPER_ALPHA
  1001. {
  1002. m_current_token.m_tag.attributes.last().local_name_builder.append_code_point(tolower(current_input_character.value()));
  1003. continue;
  1004. }
  1005. ON(0)
  1006. {
  1007. log_parse_error();
  1008. m_current_token.m_tag.attributes.last().local_name_builder.append_code_point(0xFFFD);
  1009. continue;
  1010. }
  1011. ON('"')
  1012. {
  1013. log_parse_error();
  1014. goto AnythingElseAttributeName;
  1015. }
  1016. ON('\'')
  1017. {
  1018. log_parse_error();
  1019. goto AnythingElseAttributeName;
  1020. }
  1021. ON('<')
  1022. {
  1023. log_parse_error();
  1024. goto AnythingElseAttributeName;
  1025. }
  1026. ANYTHING_ELSE
  1027. {
  1028. AnythingElseAttributeName:
  1029. m_current_token.m_tag.attributes.last().local_name_builder.append_code_point(current_input_character.value());
  1030. continue;
  1031. }
  1032. }
  1033. END_STATE
  1034. BEGIN_STATE(AfterAttributeName)
  1035. {
  1036. ON_WHITESPACE
  1037. {
  1038. continue;
  1039. }
  1040. ON('/')
  1041. {
  1042. SWITCH_TO(SelfClosingStartTag);
  1043. }
  1044. ON('=')
  1045. {
  1046. m_current_token.m_tag.attributes.last().name_end_position = nth_last_position(1);
  1047. SWITCH_TO(BeforeAttributeValue);
  1048. }
  1049. ON('>')
  1050. {
  1051. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1052. }
  1053. ON_EOF
  1054. {
  1055. log_parse_error();
  1056. EMIT_EOF;
  1057. }
  1058. ANYTHING_ELSE
  1059. {
  1060. m_current_token.m_tag.attributes.append(HTMLToken::AttributeBuilder());
  1061. m_current_token.m_tag.attributes.last().name_start_position = m_source_positions.last();
  1062. RECONSUME_IN(AttributeName);
  1063. }
  1064. }
  1065. END_STATE
  1066. BEGIN_STATE(BeforeAttributeValue)
  1067. {
  1068. m_current_token.m_tag.attributes.last().value_start_position = nth_last_position(1);
  1069. ON_WHITESPACE
  1070. {
  1071. continue;
  1072. }
  1073. ON('"')
  1074. {
  1075. SWITCH_TO(AttributeValueDoubleQuoted);
  1076. }
  1077. ON('\'')
  1078. {
  1079. SWITCH_TO(AttributeValueSingleQuoted);
  1080. }
  1081. ON('>')
  1082. {
  1083. log_parse_error();
  1084. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1085. }
  1086. ANYTHING_ELSE
  1087. {
  1088. RECONSUME_IN(AttributeValueUnquoted);
  1089. }
  1090. }
  1091. END_STATE
  1092. BEGIN_STATE(AttributeValueDoubleQuoted)
  1093. {
  1094. ON('"')
  1095. {
  1096. SWITCH_TO(AfterAttributeValueQuoted);
  1097. }
  1098. ON('&')
  1099. {
  1100. m_return_state = State::AttributeValueDoubleQuoted;
  1101. SWITCH_TO(CharacterReference);
  1102. }
  1103. ON(0)
  1104. {
  1105. log_parse_error();
  1106. m_current_token.m_tag.attributes.last().value_builder.append_code_point(0xFFFD);
  1107. continue;
  1108. }
  1109. ON_EOF
  1110. {
  1111. log_parse_error();
  1112. EMIT_EOF;
  1113. }
  1114. ANYTHING_ELSE
  1115. {
  1116. m_current_token.m_tag.attributes.last().value_builder.append_code_point(current_input_character.value());
  1117. continue;
  1118. }
  1119. }
  1120. END_STATE
  1121. BEGIN_STATE(AttributeValueSingleQuoted)
  1122. {
  1123. ON('\'')
  1124. {
  1125. SWITCH_TO(AfterAttributeValueQuoted);
  1126. }
  1127. ON('&')
  1128. {
  1129. m_return_state = State::AttributeValueSingleQuoted;
  1130. SWITCH_TO(CharacterReference);
  1131. }
  1132. ON(0)
  1133. {
  1134. log_parse_error();
  1135. m_current_token.m_tag.attributes.last().value_builder.append_code_point(0xFFFD);
  1136. continue;
  1137. }
  1138. ON_EOF
  1139. {
  1140. log_parse_error();
  1141. EMIT_EOF;
  1142. }
  1143. ANYTHING_ELSE
  1144. {
  1145. m_current_token.m_tag.attributes.last().value_builder.append_code_point(current_input_character.value());
  1146. continue;
  1147. }
  1148. }
  1149. END_STATE
  1150. BEGIN_STATE(AttributeValueUnquoted)
  1151. {
  1152. ON_WHITESPACE
  1153. {
  1154. m_current_token.m_tag.attributes.last().value_end_position = nth_last_position(2);
  1155. SWITCH_TO(BeforeAttributeName);
  1156. }
  1157. ON('&')
  1158. {
  1159. m_return_state = State::AttributeValueUnquoted;
  1160. SWITCH_TO(CharacterReference);
  1161. }
  1162. ON('>')
  1163. {
  1164. m_current_token.m_tag.attributes.last().value_end_position = nth_last_position(2);
  1165. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1166. }
  1167. ON(0)
  1168. {
  1169. log_parse_error();
  1170. m_current_token.m_tag.attributes.last().value_builder.append_code_point(0xFFFD);
  1171. continue;
  1172. }
  1173. ON('"')
  1174. {
  1175. log_parse_error();
  1176. goto AnythingElseAttributeValueUnquoted;
  1177. }
  1178. ON('\'')
  1179. {
  1180. log_parse_error();
  1181. goto AnythingElseAttributeValueUnquoted;
  1182. }
  1183. ON('<')
  1184. {
  1185. log_parse_error();
  1186. goto AnythingElseAttributeValueUnquoted;
  1187. }
  1188. ON('=')
  1189. {
  1190. log_parse_error();
  1191. goto AnythingElseAttributeValueUnquoted;
  1192. }
  1193. ON('`')
  1194. {
  1195. log_parse_error();
  1196. goto AnythingElseAttributeValueUnquoted;
  1197. }
  1198. ON_EOF
  1199. {
  1200. log_parse_error();
  1201. EMIT_EOF;
  1202. }
  1203. ANYTHING_ELSE
  1204. {
  1205. AnythingElseAttributeValueUnquoted:
  1206. m_current_token.m_tag.attributes.last().value_builder.append_code_point(current_input_character.value());
  1207. continue;
  1208. }
  1209. }
  1210. END_STATE
  1211. BEGIN_STATE(AfterAttributeValueQuoted)
  1212. {
  1213. m_current_token.m_tag.attributes.last().value_end_position = nth_last_position(2);
  1214. ON_WHITESPACE
  1215. {
  1216. SWITCH_TO(BeforeAttributeName);
  1217. }
  1218. ON('/')
  1219. {
  1220. SWITCH_TO(SelfClosingStartTag);
  1221. }
  1222. ON('>')
  1223. {
  1224. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1225. }
  1226. ON_EOF
  1227. {
  1228. log_parse_error();
  1229. EMIT_EOF;
  1230. }
  1231. ANYTHING_ELSE
  1232. {
  1233. log_parse_error();
  1234. RECONSUME_IN(BeforeAttributeName);
  1235. }
  1236. }
  1237. END_STATE
  1238. BEGIN_STATE(CommentStart)
  1239. {
  1240. ON('-')
  1241. {
  1242. SWITCH_TO(CommentStartDash);
  1243. }
  1244. ON('>')
  1245. {
  1246. log_parse_error();
  1247. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1248. }
  1249. ANYTHING_ELSE
  1250. {
  1251. RECONSUME_IN(Comment);
  1252. }
  1253. }
  1254. END_STATE
  1255. BEGIN_STATE(CommentStartDash)
  1256. {
  1257. ON('-')
  1258. {
  1259. SWITCH_TO(CommentEnd);
  1260. }
  1261. ON('>')
  1262. {
  1263. log_parse_error();
  1264. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1265. }
  1266. ON_EOF
  1267. {
  1268. log_parse_error();
  1269. m_queued_tokens.enqueue(m_current_token);
  1270. EMIT_EOF;
  1271. }
  1272. ANYTHING_ELSE
  1273. {
  1274. m_current_token.m_comment_or_character.data.append('-');
  1275. RECONSUME_IN(Comment);
  1276. }
  1277. }
  1278. END_STATE
  1279. BEGIN_STATE(Comment)
  1280. {
  1281. ON('<')
  1282. {
  1283. m_current_token.m_comment_or_character.data.append_code_point(current_input_character.value());
  1284. SWITCH_TO(CommentLessThanSign);
  1285. }
  1286. ON('-')
  1287. {
  1288. SWITCH_TO(CommentEndDash);
  1289. }
  1290. ON(0)
  1291. {
  1292. log_parse_error();
  1293. m_current_token.m_comment_or_character.data.append_code_point(0xFFFD);
  1294. continue;
  1295. }
  1296. ON_EOF
  1297. {
  1298. log_parse_error();
  1299. m_queued_tokens.enqueue(m_current_token);
  1300. EMIT_EOF;
  1301. }
  1302. ANYTHING_ELSE
  1303. {
  1304. m_current_token.m_comment_or_character.data.append_code_point(current_input_character.value());
  1305. continue;
  1306. }
  1307. }
  1308. END_STATE
  1309. BEGIN_STATE(CommentEnd)
  1310. {
  1311. ON('>')
  1312. {
  1313. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1314. }
  1315. ON('!')
  1316. {
  1317. SWITCH_TO(CommentEndBang);
  1318. }
  1319. ON('-')
  1320. {
  1321. m_current_token.m_comment_or_character.data.append('-');
  1322. continue;
  1323. }
  1324. ON_EOF
  1325. {
  1326. log_parse_error();
  1327. m_queued_tokens.enqueue(m_current_token);
  1328. EMIT_EOF;
  1329. }
  1330. ANYTHING_ELSE
  1331. {
  1332. m_current_token.m_comment_or_character.data.append('-');
  1333. RECONSUME_IN(Comment);
  1334. }
  1335. }
  1336. END_STATE
  1337. BEGIN_STATE(CommentEndBang)
  1338. {
  1339. ON('-')
  1340. {
  1341. m_current_token.m_comment_or_character.data.append("--!");
  1342. SWITCH_TO(CommentEndDash);
  1343. }
  1344. ON('>')
  1345. {
  1346. log_parse_error();
  1347. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1348. }
  1349. ON_EOF
  1350. {
  1351. log_parse_error();
  1352. m_queued_tokens.enqueue(m_current_token);
  1353. EMIT_EOF;
  1354. }
  1355. ANYTHING_ELSE
  1356. {
  1357. m_current_token.m_comment_or_character.data.append("--!");
  1358. RECONSUME_IN(Comment);
  1359. }
  1360. }
  1361. END_STATE
  1362. BEGIN_STATE(CommentEndDash)
  1363. {
  1364. ON('-')
  1365. {
  1366. SWITCH_TO(CommentEnd);
  1367. }
  1368. ON_EOF
  1369. {
  1370. log_parse_error();
  1371. m_queued_tokens.enqueue(m_current_token);
  1372. EMIT_EOF;
  1373. }
  1374. ANYTHING_ELSE
  1375. {
  1376. m_current_token.m_comment_or_character.data.append('-');
  1377. RECONSUME_IN(Comment);
  1378. }
  1379. }
  1380. END_STATE
  1381. BEGIN_STATE(CommentLessThanSign)
  1382. {
  1383. ON('!')
  1384. {
  1385. m_current_token.m_comment_or_character.data.append_code_point(current_input_character.value());
  1386. SWITCH_TO(CommentLessThanSignBang);
  1387. }
  1388. ON('<')
  1389. {
  1390. m_current_token.m_comment_or_character.data.append_code_point(current_input_character.value());
  1391. continue;
  1392. }
  1393. ANYTHING_ELSE
  1394. {
  1395. RECONSUME_IN(Comment);
  1396. }
  1397. }
  1398. END_STATE
  1399. BEGIN_STATE(CommentLessThanSignBang)
  1400. {
  1401. ON('-')
  1402. {
  1403. SWITCH_TO(CommentLessThanSignBangDash);
  1404. }
  1405. ANYTHING_ELSE
  1406. {
  1407. RECONSUME_IN(Comment);
  1408. }
  1409. }
  1410. END_STATE
  1411. BEGIN_STATE(CommentLessThanSignBangDash)
  1412. {
  1413. ON('-')
  1414. {
  1415. SWITCH_TO(CommentLessThanSignBangDashDash);
  1416. }
  1417. ANYTHING_ELSE
  1418. {
  1419. RECONSUME_IN(CommentEndDash);
  1420. }
  1421. }
  1422. END_STATE
  1423. BEGIN_STATE(CommentLessThanSignBangDashDash)
  1424. {
  1425. ON('>')
  1426. {
  1427. RECONSUME_IN(CommentEnd);
  1428. }
  1429. ON_EOF
  1430. {
  1431. RECONSUME_IN(CommentEnd);
  1432. }
  1433. ANYTHING_ELSE
  1434. {
  1435. log_parse_error();
  1436. RECONSUME_IN(CommentEnd);
  1437. }
  1438. }
  1439. END_STATE
  1440. BEGIN_STATE(CharacterReference)
  1441. {
  1442. m_temporary_buffer.clear();
  1443. m_temporary_buffer.append('&');
  1444. ON_ASCII_ALPHANUMERIC
  1445. {
  1446. RECONSUME_IN(NamedCharacterReference);
  1447. }
  1448. ON('#')
  1449. {
  1450. m_temporary_buffer.append(current_input_character.value());
  1451. SWITCH_TO(NumericCharacterReference);
  1452. }
  1453. ANYTHING_ELSE
  1454. {
  1455. FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE;
  1456. RECONSUME_IN_RETURN_STATE;
  1457. }
  1458. }
  1459. END_STATE
  1460. BEGIN_STATE(NamedCharacterReference)
  1461. {
  1462. size_t byte_offset = m_utf8_view.byte_offset_of(m_prev_utf8_iterator);
  1463. auto match = HTML::code_points_from_entity(m_decoded_input.substring_view(byte_offset, m_decoded_input.length() - byte_offset - 1));
  1464. if (match.has_value()) {
  1465. skip(match->entity.length() - 1);
  1466. for (auto ch : match.value().entity)
  1467. m_temporary_buffer.append(ch);
  1468. if (consumed_as_part_of_an_attribute() && !match.value().entity.ends_with(';')) {
  1469. auto next_code_point = peek_code_point(0);
  1470. if (next_code_point.has_value() && (next_code_point.value() == '=' || isalnum(next_code_point.value()))) {
  1471. FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE;
  1472. SWITCH_TO_RETURN_STATE;
  1473. }
  1474. }
  1475. if (!match.value().entity.ends_with(';')) {
  1476. log_parse_error();
  1477. }
  1478. m_temporary_buffer.clear();
  1479. m_temporary_buffer.append(match.value().code_points);
  1480. FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE;
  1481. SWITCH_TO_RETURN_STATE;
  1482. } else {
  1483. FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE;
  1484. // FIXME: This should be SWITCH_TO, but we always lose the first character on this path, so just reconsume it.
  1485. // I can't wrap my head around how to do it as the spec says.
  1486. RECONSUME_IN(AmbiguousAmpersand);
  1487. }
  1488. }
  1489. END_STATE
  1490. BEGIN_STATE(AmbiguousAmpersand)
  1491. {
  1492. ON_ASCII_ALPHANUMERIC
  1493. {
  1494. if (consumed_as_part_of_an_attribute()) {
  1495. m_current_token.m_tag.attributes.last().value_builder.append_code_point(current_input_character.value());
  1496. continue;
  1497. } else {
  1498. EMIT_CURRENT_CHARACTER;
  1499. }
  1500. }
  1501. ON(';')
  1502. {
  1503. log_parse_error();
  1504. RECONSUME_IN_RETURN_STATE;
  1505. }
  1506. ANYTHING_ELSE
  1507. {
  1508. RECONSUME_IN_RETURN_STATE;
  1509. }
  1510. }
  1511. END_STATE
  1512. BEGIN_STATE(NumericCharacterReference)
  1513. {
  1514. m_character_reference_code = 0;
  1515. ON('X')
  1516. {
  1517. m_temporary_buffer.append(current_input_character.value());
  1518. SWITCH_TO(HexadecimalCharacterReferenceStart);
  1519. }
  1520. ON('x')
  1521. {
  1522. m_temporary_buffer.append(current_input_character.value());
  1523. SWITCH_TO(HexadecimalCharacterReferenceStart);
  1524. }
  1525. ANYTHING_ELSE
  1526. {
  1527. RECONSUME_IN(DecimalCharacterReferenceStart);
  1528. }
  1529. }
  1530. END_STATE
  1531. BEGIN_STATE(HexadecimalCharacterReferenceStart)
  1532. {
  1533. ON_ASCII_HEX_DIGIT
  1534. {
  1535. RECONSUME_IN(HexadecimalCharacterReference);
  1536. }
  1537. ANYTHING_ELSE
  1538. {
  1539. log_parse_error();
  1540. FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE;
  1541. RECONSUME_IN_RETURN_STATE;
  1542. }
  1543. }
  1544. END_STATE
  1545. BEGIN_STATE(DecimalCharacterReferenceStart)
  1546. {
  1547. ON_ASCII_DIGIT
  1548. {
  1549. RECONSUME_IN(DecimalCharacterReference);
  1550. }
  1551. ANYTHING_ELSE
  1552. {
  1553. log_parse_error();
  1554. FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE;
  1555. RECONSUME_IN_RETURN_STATE;
  1556. }
  1557. }
  1558. END_STATE
  1559. BEGIN_STATE(HexadecimalCharacterReference)
  1560. {
  1561. ON_ASCII_DIGIT
  1562. {
  1563. m_character_reference_code *= 16;
  1564. m_character_reference_code += current_input_character.value() - 0x30;
  1565. continue;
  1566. }
  1567. ON_ASCII_UPPER_ALPHA
  1568. {
  1569. m_character_reference_code *= 16;
  1570. m_character_reference_code += current_input_character.value() - 0x37;
  1571. continue;
  1572. }
  1573. ON_ASCII_LOWER_ALPHA
  1574. {
  1575. m_character_reference_code *= 16;
  1576. m_character_reference_code += current_input_character.value() - 0x57;
  1577. continue;
  1578. }
  1579. ON(';')
  1580. {
  1581. SWITCH_TO(NumericCharacterReferenceEnd);
  1582. }
  1583. ANYTHING_ELSE
  1584. {
  1585. log_parse_error();
  1586. RECONSUME_IN(NumericCharacterReferenceEnd);
  1587. }
  1588. }
  1589. END_STATE
  1590. BEGIN_STATE(DecimalCharacterReference)
  1591. {
  1592. ON_ASCII_DIGIT
  1593. {
  1594. m_character_reference_code *= 10;
  1595. m_character_reference_code += current_input_character.value() - 0x30;
  1596. continue;
  1597. }
  1598. ON(';')
  1599. {
  1600. SWITCH_TO(NumericCharacterReferenceEnd);
  1601. }
  1602. ANYTHING_ELSE
  1603. {
  1604. log_parse_error();
  1605. RECONSUME_IN(NumericCharacterReferenceEnd);
  1606. }
  1607. }
  1608. END_STATE
  1609. BEGIN_STATE(NumericCharacterReferenceEnd)
  1610. {
  1611. DONT_CONSUME_NEXT_INPUT_CHARACTER;
  1612. if (m_character_reference_code == 0) {
  1613. log_parse_error();
  1614. m_character_reference_code = 0xFFFD;
  1615. }
  1616. if (m_character_reference_code > 0x10ffff) {
  1617. log_parse_error();
  1618. m_character_reference_code = 0xFFFD;
  1619. }
  1620. if (is_surrogate(m_character_reference_code)) {
  1621. log_parse_error();
  1622. m_character_reference_code = 0xFFFD;
  1623. }
  1624. if (is_noncharacter(m_character_reference_code)) {
  1625. log_parse_error();
  1626. }
  1627. if (m_character_reference_code == 0xd || (is_control(m_character_reference_code) && !isspace(m_character_reference_code))) {
  1628. log_parse_error();
  1629. constexpr struct {
  1630. u32 number;
  1631. u32 code_point;
  1632. } conversion_table[] = {
  1633. { 0x80, 0x20AC },
  1634. { 0x82, 0x201A },
  1635. { 0x83, 0x0192 },
  1636. { 0x84, 0x201E },
  1637. { 0x85, 0x2026 },
  1638. { 0x86, 0x2020 },
  1639. { 0x87, 0x2021 },
  1640. { 0x88, 0x02C6 },
  1641. { 0x89, 0x2030 },
  1642. { 0x8A, 0x0160 },
  1643. { 0x8B, 0x2039 },
  1644. { 0x8C, 0x0152 },
  1645. { 0x8E, 0x017D },
  1646. { 0x91, 0x2018 },
  1647. { 0x92, 0x2019 },
  1648. { 0x93, 0x201C },
  1649. { 0x94, 0x201D },
  1650. { 0x95, 0x2022 },
  1651. { 0x96, 0x2013 },
  1652. { 0x97, 0x2014 },
  1653. { 0x98, 0x02DC },
  1654. { 0x99, 0x2122 },
  1655. { 0x9A, 0x0161 },
  1656. { 0x9B, 0x203A },
  1657. { 0x9C, 0x0153 },
  1658. { 0x9E, 0x017E },
  1659. { 0x9F, 0x0178 },
  1660. };
  1661. for (auto& entry : conversion_table) {
  1662. if (m_character_reference_code == entry.number) {
  1663. m_character_reference_code = entry.code_point;
  1664. break;
  1665. }
  1666. }
  1667. }
  1668. m_temporary_buffer.clear();
  1669. m_temporary_buffer.append(m_character_reference_code);
  1670. FLUSH_CODEPOINTS_CONSUMED_AS_A_CHARACTER_REFERENCE;
  1671. SWITCH_TO_RETURN_STATE;
  1672. }
  1673. END_STATE
  1674. BEGIN_STATE(RCDATA)
  1675. {
  1676. ON('&')
  1677. {
  1678. m_return_state = State::RCDATA;
  1679. SWITCH_TO(CharacterReference);
  1680. }
  1681. ON('<')
  1682. {
  1683. SWITCH_TO(RCDATALessThanSign);
  1684. }
  1685. ON(0)
  1686. {
  1687. log_parse_error();
  1688. EMIT_CHARACTER(0xFFFD);
  1689. }
  1690. ON_EOF
  1691. {
  1692. EMIT_EOF;
  1693. }
  1694. ANYTHING_ELSE
  1695. {
  1696. EMIT_CURRENT_CHARACTER;
  1697. }
  1698. }
  1699. END_STATE
  1700. BEGIN_STATE(RCDATALessThanSign)
  1701. {
  1702. ON('/')
  1703. {
  1704. m_temporary_buffer.clear();
  1705. SWITCH_TO(RCDATAEndTagOpen);
  1706. }
  1707. ANYTHING_ELSE
  1708. {
  1709. EMIT_CHARACTER_AND_RECONSUME_IN('<', RCDATA);
  1710. }
  1711. }
  1712. END_STATE
  1713. BEGIN_STATE(RCDATAEndTagOpen)
  1714. {
  1715. ON_ASCII_ALPHA
  1716. {
  1717. create_new_token(HTMLToken::Type::EndTag);
  1718. RECONSUME_IN(RCDATAEndTagName);
  1719. }
  1720. ANYTHING_ELSE
  1721. {
  1722. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1723. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1724. RECONSUME_IN(RCDATA);
  1725. }
  1726. }
  1727. END_STATE
  1728. BEGIN_STATE(RCDATAEndTagName)
  1729. {
  1730. ON_WHITESPACE
  1731. {
  1732. if (!current_end_tag_token_is_appropriate()) {
  1733. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1734. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1735. for (auto code_point : m_temporary_buffer)
  1736. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1737. RECONSUME_IN(RCDATA);
  1738. }
  1739. SWITCH_TO(BeforeAttributeName);
  1740. }
  1741. ON('/')
  1742. {
  1743. if (!current_end_tag_token_is_appropriate()) {
  1744. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1745. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1746. for (auto code_point : m_temporary_buffer)
  1747. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1748. RECONSUME_IN(RCDATA);
  1749. }
  1750. SWITCH_TO(SelfClosingStartTag);
  1751. }
  1752. ON('>')
  1753. {
  1754. if (!current_end_tag_token_is_appropriate()) {
  1755. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1756. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1757. for (auto code_point : m_temporary_buffer)
  1758. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1759. RECONSUME_IN(RCDATA);
  1760. }
  1761. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1762. }
  1763. ON_ASCII_UPPER_ALPHA
  1764. {
  1765. m_current_token.m_tag.tag_name.append(tolower(current_input_character.value()));
  1766. m_temporary_buffer.append(current_input_character.value());
  1767. continue;
  1768. }
  1769. ON_ASCII_LOWER_ALPHA
  1770. {
  1771. m_current_token.m_tag.tag_name.append_code_point(current_input_character.value());
  1772. m_temporary_buffer.append(current_input_character.value());
  1773. continue;
  1774. }
  1775. ANYTHING_ELSE
  1776. {
  1777. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1778. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1779. for (auto code_point : m_temporary_buffer)
  1780. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1781. RECONSUME_IN(RCDATA);
  1782. }
  1783. }
  1784. END_STATE
  1785. BEGIN_STATE(RAWTEXT)
  1786. {
  1787. ON('<')
  1788. {
  1789. SWITCH_TO(RAWTEXTLessThanSign);
  1790. }
  1791. ON(0)
  1792. {
  1793. log_parse_error();
  1794. EMIT_CHARACTER(0xFFFD);
  1795. }
  1796. ON_EOF
  1797. {
  1798. EMIT_EOF;
  1799. }
  1800. ANYTHING_ELSE
  1801. {
  1802. EMIT_CURRENT_CHARACTER;
  1803. }
  1804. }
  1805. END_STATE
  1806. BEGIN_STATE(RAWTEXTLessThanSign)
  1807. {
  1808. ON('/')
  1809. {
  1810. m_temporary_buffer.clear();
  1811. SWITCH_TO(RAWTEXTEndTagOpen);
  1812. }
  1813. ANYTHING_ELSE
  1814. {
  1815. EMIT_CHARACTER_AND_RECONSUME_IN('<', RAWTEXT);
  1816. }
  1817. }
  1818. END_STATE
  1819. BEGIN_STATE(RAWTEXTEndTagOpen)
  1820. {
  1821. ON_ASCII_ALPHA
  1822. {
  1823. create_new_token(HTMLToken::Type::EndTag);
  1824. RECONSUME_IN(RAWTEXTEndTagName);
  1825. }
  1826. ANYTHING_ELSE
  1827. {
  1828. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1829. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1830. RECONSUME_IN(RAWTEXT);
  1831. }
  1832. }
  1833. END_STATE
  1834. BEGIN_STATE(RAWTEXTEndTagName)
  1835. {
  1836. ON_WHITESPACE
  1837. {
  1838. if (!current_end_tag_token_is_appropriate()) {
  1839. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1840. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1841. for (auto code_point : m_temporary_buffer)
  1842. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1843. RECONSUME_IN(RAWTEXT);
  1844. }
  1845. SWITCH_TO(BeforeAttributeName);
  1846. }
  1847. ON('/')
  1848. {
  1849. if (!current_end_tag_token_is_appropriate()) {
  1850. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1851. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1852. for (auto code_point : m_temporary_buffer)
  1853. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1854. RECONSUME_IN(RAWTEXT);
  1855. }
  1856. SWITCH_TO(SelfClosingStartTag);
  1857. }
  1858. ON('>')
  1859. {
  1860. if (!current_end_tag_token_is_appropriate()) {
  1861. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1862. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1863. for (auto code_point : m_temporary_buffer)
  1864. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1865. RECONSUME_IN(RAWTEXT);
  1866. }
  1867. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  1868. }
  1869. ON_ASCII_UPPER_ALPHA
  1870. {
  1871. m_current_token.m_tag.tag_name.append(tolower(current_input_character.value()));
  1872. m_temporary_buffer.append(current_input_character.value());
  1873. continue;
  1874. }
  1875. ON_ASCII_LOWER_ALPHA
  1876. {
  1877. m_current_token.m_tag.tag_name.append(current_input_character.value());
  1878. m_temporary_buffer.append(current_input_character.value());
  1879. continue;
  1880. }
  1881. ANYTHING_ELSE
  1882. {
  1883. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1884. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  1885. for (auto code_point : m_temporary_buffer)
  1886. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  1887. RECONSUME_IN(RAWTEXT);
  1888. }
  1889. }
  1890. END_STATE
  1891. BEGIN_STATE(ScriptData)
  1892. {
  1893. ON('<')
  1894. {
  1895. SWITCH_TO(ScriptDataLessThanSign);
  1896. }
  1897. ON(0)
  1898. {
  1899. log_parse_error();
  1900. EMIT_CHARACTER(0xFFFD);
  1901. }
  1902. ON_EOF
  1903. {
  1904. EMIT_EOF;
  1905. }
  1906. ANYTHING_ELSE
  1907. {
  1908. EMIT_CURRENT_CHARACTER;
  1909. }
  1910. }
  1911. END_STATE
  1912. BEGIN_STATE(PLAINTEXT)
  1913. {
  1914. ON(0)
  1915. {
  1916. log_parse_error();
  1917. EMIT_CHARACTER(0xFFFD);
  1918. }
  1919. ON_EOF
  1920. {
  1921. EMIT_EOF;
  1922. }
  1923. ANYTHING_ELSE
  1924. {
  1925. EMIT_CURRENT_CHARACTER;
  1926. }
  1927. }
  1928. END_STATE
  1929. BEGIN_STATE(ScriptDataLessThanSign)
  1930. {
  1931. ON('/')
  1932. {
  1933. m_temporary_buffer.clear();
  1934. SWITCH_TO(ScriptDataEndTagOpen);
  1935. }
  1936. ON('!')
  1937. {
  1938. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  1939. m_queued_tokens.enqueue(HTMLToken::make_character('!'));
  1940. SWITCH_TO(ScriptDataEscapeStart);
  1941. }
  1942. ANYTHING_ELSE
  1943. {
  1944. EMIT_CHARACTER_AND_RECONSUME_IN('<', ScriptData);
  1945. }
  1946. }
  1947. END_STATE
  1948. BEGIN_STATE(ScriptDataEscapeStart)
  1949. {
  1950. ON('-')
  1951. {
  1952. SWITCH_TO_AND_EMIT_CHARACTER('-', ScriptDataEscapeStartDash);
  1953. }
  1954. ANYTHING_ELSE
  1955. {
  1956. RECONSUME_IN(ScriptData);
  1957. }
  1958. }
  1959. END_STATE
  1960. BEGIN_STATE(ScriptDataEscapeStartDash)
  1961. {
  1962. ON('-')
  1963. {
  1964. SWITCH_TO_AND_EMIT_CHARACTER('-', ScriptDataEscapedDashDash);
  1965. }
  1966. ANYTHING_ELSE
  1967. {
  1968. RECONSUME_IN(ScriptData);
  1969. }
  1970. }
  1971. END_STATE
  1972. BEGIN_STATE(ScriptDataEscapedDashDash)
  1973. {
  1974. ON('-')
  1975. {
  1976. EMIT_CHARACTER('-');
  1977. }
  1978. ON('<')
  1979. {
  1980. SWITCH_TO(ScriptDataEscapedLessThanSign);
  1981. }
  1982. ON('>')
  1983. {
  1984. SWITCH_TO_AND_EMIT_CHARACTER('>', ScriptData);
  1985. }
  1986. ON(0)
  1987. {
  1988. log_parse_error();
  1989. SWITCH_TO_AND_EMIT_CHARACTER(0xFFFD, ScriptDataEscaped);
  1990. }
  1991. ON_EOF
  1992. {
  1993. log_parse_error();
  1994. EMIT_EOF;
  1995. }
  1996. ANYTHING_ELSE
  1997. {
  1998. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  1999. }
  2000. }
  2001. END_STATE
  2002. BEGIN_STATE(ScriptDataEscapedLessThanSign)
  2003. {
  2004. ON('/')
  2005. {
  2006. m_temporary_buffer.clear();
  2007. SWITCH_TO(ScriptDataEscapedEndTagOpen);
  2008. }
  2009. ON_ASCII_ALPHA
  2010. {
  2011. m_temporary_buffer.clear();
  2012. EMIT_CHARACTER_AND_RECONSUME_IN('<', ScriptDataDoubleEscapeStart);
  2013. }
  2014. ANYTHING_ELSE
  2015. {
  2016. EMIT_CHARACTER_AND_RECONSUME_IN('<', ScriptDataEscaped);
  2017. }
  2018. }
  2019. END_STATE
  2020. BEGIN_STATE(ScriptDataEscapedEndTagOpen)
  2021. {
  2022. ON_ASCII_ALPHA
  2023. {
  2024. create_new_token(HTMLToken::Type::EndTag);
  2025. RECONSUME_IN(ScriptDataEscapedEndTagName);
  2026. }
  2027. ANYTHING_ELSE
  2028. {
  2029. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2030. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2031. RECONSUME_IN(ScriptDataEscaped);
  2032. }
  2033. }
  2034. END_STATE
  2035. BEGIN_STATE(ScriptDataEscapedEndTagName)
  2036. {
  2037. ON_WHITESPACE
  2038. {
  2039. if (current_end_tag_token_is_appropriate())
  2040. SWITCH_TO(BeforeAttributeName);
  2041. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2042. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2043. for (auto code_point : m_temporary_buffer) {
  2044. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2045. }
  2046. RECONSUME_IN(ScriptDataEscaped);
  2047. }
  2048. ON('/')
  2049. {
  2050. if (current_end_tag_token_is_appropriate())
  2051. SWITCH_TO(SelfClosingStartTag);
  2052. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2053. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2054. for (auto code_point : m_temporary_buffer) {
  2055. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2056. }
  2057. RECONSUME_IN(ScriptDataEscaped);
  2058. }
  2059. ON('>')
  2060. {
  2061. if (current_end_tag_token_is_appropriate())
  2062. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  2063. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2064. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2065. for (auto code_point : m_temporary_buffer) {
  2066. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2067. }
  2068. RECONSUME_IN(ScriptDataEscaped);
  2069. }
  2070. ON_ASCII_UPPER_ALPHA
  2071. {
  2072. m_current_token.m_tag.tag_name.append(tolower(current_input_character.value()));
  2073. m_temporary_buffer.append(current_input_character.value());
  2074. continue;
  2075. }
  2076. ON_ASCII_LOWER_ALPHA
  2077. {
  2078. m_current_token.m_tag.tag_name.append(current_input_character.value());
  2079. m_temporary_buffer.append(current_input_character.value());
  2080. continue;
  2081. }
  2082. ANYTHING_ELSE
  2083. {
  2084. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2085. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2086. for (auto code_point : m_temporary_buffer) {
  2087. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2088. }
  2089. RECONSUME_IN(ScriptDataEscaped);
  2090. }
  2091. }
  2092. END_STATE
  2093. BEGIN_STATE(ScriptDataDoubleEscapeStart)
  2094. {
  2095. auto temporary_buffer_equal_to_script = [this]() -> bool {
  2096. if (m_temporary_buffer.size() != 6)
  2097. return false;
  2098. // FIXME: Is there a better way of doing this?
  2099. return m_temporary_buffer[0] == 's' && m_temporary_buffer[1] == 'c' && m_temporary_buffer[2] == 'r' && m_temporary_buffer[3] == 'i' && m_temporary_buffer[4] == 'p' && m_temporary_buffer[5] == 't';
  2100. };
  2101. ON_WHITESPACE
  2102. {
  2103. if (temporary_buffer_equal_to_script())
  2104. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2105. else
  2106. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  2107. }
  2108. ON('/')
  2109. {
  2110. if (temporary_buffer_equal_to_script())
  2111. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2112. else
  2113. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  2114. }
  2115. ON('>')
  2116. {
  2117. if (temporary_buffer_equal_to_script())
  2118. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2119. else
  2120. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  2121. }
  2122. ON_ASCII_UPPER_ALPHA
  2123. {
  2124. m_temporary_buffer.append(tolower(current_input_character.value()));
  2125. EMIT_CURRENT_CHARACTER;
  2126. }
  2127. ON_ASCII_LOWER_ALPHA
  2128. {
  2129. m_temporary_buffer.append(current_input_character.value());
  2130. EMIT_CURRENT_CHARACTER;
  2131. }
  2132. ANYTHING_ELSE
  2133. {
  2134. RECONSUME_IN(ScriptDataEscaped);
  2135. }
  2136. }
  2137. END_STATE
  2138. BEGIN_STATE(ScriptDataDoubleEscaped)
  2139. {
  2140. ON('-')
  2141. {
  2142. SWITCH_TO_AND_EMIT_CHARACTER('-', ScriptDataDoubleEscapedDash);
  2143. }
  2144. ON('<')
  2145. {
  2146. SWITCH_TO_AND_EMIT_CHARACTER('<', ScriptDataDoubleEscapedLessThanSign);
  2147. }
  2148. ON(0)
  2149. {
  2150. log_parse_error();
  2151. EMIT_CHARACTER(0xFFFD);
  2152. }
  2153. ON_EOF
  2154. {
  2155. log_parse_error();
  2156. EMIT_EOF;
  2157. }
  2158. ANYTHING_ELSE
  2159. {
  2160. EMIT_CURRENT_CHARACTER;
  2161. }
  2162. }
  2163. END_STATE
  2164. BEGIN_STATE(ScriptDataDoubleEscapedDash)
  2165. {
  2166. ON('-')
  2167. {
  2168. SWITCH_TO_AND_EMIT_CHARACTER('-', ScriptDataDoubleEscapedDashDash);
  2169. }
  2170. ON('<')
  2171. {
  2172. SWITCH_TO_AND_EMIT_CHARACTER('<', ScriptDataDoubleEscapedLessThanSign);
  2173. }
  2174. ON(0)
  2175. {
  2176. log_parse_error();
  2177. SWITCH_TO_AND_EMIT_CHARACTER(0xFFFD, ScriptDataDoubleEscaped);
  2178. }
  2179. ON_EOF
  2180. {
  2181. log_parse_error();
  2182. EMIT_EOF;
  2183. }
  2184. ANYTHING_ELSE
  2185. {
  2186. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2187. }
  2188. }
  2189. END_STATE
  2190. BEGIN_STATE(ScriptDataDoubleEscapedDashDash)
  2191. {
  2192. ON('-')
  2193. {
  2194. EMIT_CHARACTER('-');
  2195. }
  2196. ON('<')
  2197. {
  2198. SWITCH_TO_AND_EMIT_CHARACTER('<', ScriptDataDoubleEscapedLessThanSign);
  2199. }
  2200. ON('>')
  2201. {
  2202. SWITCH_TO_AND_EMIT_CHARACTER('>', ScriptData);
  2203. }
  2204. ON(0)
  2205. {
  2206. log_parse_error();
  2207. SWITCH_TO_AND_EMIT_CHARACTER(0xFFFD, ScriptDataDoubleEscaped);
  2208. }
  2209. ON_EOF
  2210. {
  2211. log_parse_error();
  2212. EMIT_EOF;
  2213. }
  2214. ANYTHING_ELSE
  2215. {
  2216. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2217. }
  2218. }
  2219. END_STATE
  2220. BEGIN_STATE(ScriptDataDoubleEscapedLessThanSign)
  2221. {
  2222. ON('/')
  2223. {
  2224. m_temporary_buffer.clear();
  2225. SWITCH_TO_AND_EMIT_CHARACTER('/', ScriptDataDoubleEscapeEnd);
  2226. }
  2227. ANYTHING_ELSE
  2228. {
  2229. RECONSUME_IN(ScriptDataDoubleEscaped);
  2230. }
  2231. }
  2232. END_STATE
  2233. BEGIN_STATE(ScriptDataDoubleEscapeEnd)
  2234. {
  2235. auto temporary_buffer_equal_to_script = [this]() -> bool {
  2236. if (m_temporary_buffer.size() != 6)
  2237. return false;
  2238. // FIXME: Is there a better way of doing this?
  2239. return m_temporary_buffer[0] == 's' && m_temporary_buffer[1] == 'c' && m_temporary_buffer[2] == 'r' && m_temporary_buffer[3] == 'i' && m_temporary_buffer[4] == 'p' && m_temporary_buffer[5] == 't';
  2240. };
  2241. ON_WHITESPACE
  2242. {
  2243. if (temporary_buffer_equal_to_script())
  2244. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  2245. else
  2246. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2247. }
  2248. ON('/')
  2249. {
  2250. if (temporary_buffer_equal_to_script())
  2251. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  2252. else
  2253. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2254. }
  2255. ON('>')
  2256. {
  2257. if (temporary_buffer_equal_to_script())
  2258. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  2259. else
  2260. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataDoubleEscaped);
  2261. }
  2262. ON_ASCII_UPPER_ALPHA
  2263. {
  2264. m_temporary_buffer.append(tolower(current_input_character.value()));
  2265. EMIT_CURRENT_CHARACTER;
  2266. }
  2267. ON_ASCII_LOWER_ALPHA
  2268. {
  2269. m_temporary_buffer.append(current_input_character.value());
  2270. EMIT_CURRENT_CHARACTER;
  2271. }
  2272. ANYTHING_ELSE
  2273. {
  2274. RECONSUME_IN(ScriptDataDoubleEscaped);
  2275. }
  2276. }
  2277. END_STATE
  2278. BEGIN_STATE(ScriptDataEscapedDash)
  2279. {
  2280. ON('-')
  2281. {
  2282. SWITCH_TO_AND_EMIT_CHARACTER('-', ScriptDataEscapedDashDash);
  2283. }
  2284. ON('<')
  2285. {
  2286. SWITCH_TO(ScriptDataEscapedLessThanSign);
  2287. }
  2288. ON(0)
  2289. {
  2290. log_parse_error();
  2291. SWITCH_TO_AND_EMIT_CHARACTER(0xFFFD, ScriptDataEscaped);
  2292. }
  2293. ON_EOF
  2294. {
  2295. log_parse_error();
  2296. EMIT_EOF;
  2297. }
  2298. ANYTHING_ELSE
  2299. {
  2300. SWITCH_TO_AND_EMIT_CURRENT_CHARACTER(ScriptDataEscaped);
  2301. }
  2302. }
  2303. END_STATE
  2304. BEGIN_STATE(ScriptDataEscaped)
  2305. {
  2306. ON('-')
  2307. {
  2308. SWITCH_TO_AND_EMIT_CHARACTER('-', ScriptDataEscapedDash);
  2309. }
  2310. ON('<')
  2311. {
  2312. SWITCH_TO(ScriptDataEscapedLessThanSign);
  2313. }
  2314. ON(0)
  2315. {
  2316. log_parse_error();
  2317. EMIT_CHARACTER(0xFFFD);
  2318. }
  2319. ON_EOF
  2320. {
  2321. log_parse_error();
  2322. EMIT_EOF;
  2323. }
  2324. ANYTHING_ELSE
  2325. {
  2326. EMIT_CURRENT_CHARACTER;
  2327. }
  2328. }
  2329. END_STATE
  2330. BEGIN_STATE(ScriptDataEndTagOpen)
  2331. {
  2332. ON_ASCII_ALPHA
  2333. {
  2334. create_new_token(HTMLToken::Type::EndTag);
  2335. RECONSUME_IN(ScriptDataEndTagName);
  2336. }
  2337. ANYTHING_ELSE
  2338. {
  2339. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2340. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2341. RECONSUME_IN(ScriptData);
  2342. }
  2343. }
  2344. END_STATE
  2345. BEGIN_STATE(ScriptDataEndTagName)
  2346. {
  2347. ON_WHITESPACE
  2348. {
  2349. if (current_end_tag_token_is_appropriate())
  2350. SWITCH_TO(BeforeAttributeName);
  2351. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2352. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2353. for (auto code_point : m_temporary_buffer)
  2354. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2355. RECONSUME_IN(ScriptData);
  2356. }
  2357. ON('/')
  2358. {
  2359. if (current_end_tag_token_is_appropriate())
  2360. SWITCH_TO(SelfClosingStartTag);
  2361. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2362. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2363. for (auto code_point : m_temporary_buffer)
  2364. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2365. RECONSUME_IN(ScriptData);
  2366. }
  2367. ON('>')
  2368. {
  2369. if (current_end_tag_token_is_appropriate())
  2370. SWITCH_TO_AND_EMIT_CURRENT_TOKEN(Data);
  2371. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2372. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2373. for (auto code_point : m_temporary_buffer)
  2374. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2375. RECONSUME_IN(ScriptData);
  2376. }
  2377. ON_ASCII_UPPER_ALPHA
  2378. {
  2379. m_current_token.m_tag.tag_name.append(tolower(current_input_character.value()));
  2380. m_temporary_buffer.append(current_input_character.value());
  2381. continue;
  2382. }
  2383. ON_ASCII_LOWER_ALPHA
  2384. {
  2385. m_current_token.m_tag.tag_name.append(current_input_character.value());
  2386. m_temporary_buffer.append(current_input_character.value());
  2387. continue;
  2388. }
  2389. ANYTHING_ELSE
  2390. {
  2391. m_queued_tokens.enqueue(HTMLToken::make_character('<'));
  2392. m_queued_tokens.enqueue(HTMLToken::make_character('/'));
  2393. for (auto code_point : m_temporary_buffer)
  2394. m_queued_tokens.enqueue(HTMLToken::make_character(code_point));
  2395. RECONSUME_IN(ScriptData);
  2396. }
  2397. }
  2398. END_STATE
  2399. BEGIN_STATE(CDATASection)
  2400. {
  2401. ON(']')
  2402. {
  2403. SWITCH_TO(CDATASectionBracket);
  2404. }
  2405. ON_EOF
  2406. {
  2407. log_parse_error();
  2408. EMIT_EOF;
  2409. }
  2410. ANYTHING_ELSE
  2411. {
  2412. EMIT_CURRENT_CHARACTER;
  2413. }
  2414. }
  2415. END_STATE
  2416. BEGIN_STATE(CDATASectionBracket)
  2417. {
  2418. ON(']')
  2419. {
  2420. SWITCH_TO(CDATASectionEnd);
  2421. }
  2422. ANYTHING_ELSE
  2423. {
  2424. EMIT_CHARACTER_AND_RECONSUME_IN(']', CDATASection);
  2425. }
  2426. }
  2427. END_STATE
  2428. BEGIN_STATE(CDATASectionEnd)
  2429. {
  2430. ON(']')
  2431. {
  2432. EMIT_CHARACTER(']');
  2433. }
  2434. ON('>')
  2435. {
  2436. SWITCH_TO(Data);
  2437. }
  2438. ANYTHING_ELSE
  2439. {
  2440. m_queued_tokens.enqueue(HTMLToken::make_character(']'));
  2441. m_queued_tokens.enqueue(HTMLToken::make_character(']'));
  2442. RECONSUME_IN(CDATASection);
  2443. }
  2444. }
  2445. END_STATE
  2446. default:
  2447. TODO();
  2448. }
  2449. }
  2450. }
  2451. bool HTMLTokenizer::consume_next_if_match(const StringView& string, CaseSensitivity case_sensitivity)
  2452. {
  2453. for (size_t i = 0; i < string.length(); ++i) {
  2454. auto code_point = peek_code_point(i);
  2455. if (!code_point.has_value())
  2456. return false;
  2457. // FIXME: This should be more Unicode-aware.
  2458. if (case_sensitivity == CaseSensitivity::CaseInsensitive) {
  2459. if (code_point.value() < 0x80) {
  2460. if (tolower(code_point.value()) != tolower(string[i]))
  2461. return false;
  2462. continue;
  2463. }
  2464. }
  2465. if (code_point.value() != (u32)string[i])
  2466. return false;
  2467. }
  2468. skip(string.length());
  2469. return true;
  2470. }
  2471. void HTMLTokenizer::create_new_token(HTMLToken::Type type)
  2472. {
  2473. m_current_token = {};
  2474. m_current_token.m_type = type;
  2475. size_t offset = 0;
  2476. switch (type) {
  2477. case HTMLToken::Type::StartTag:
  2478. offset = 1;
  2479. break;
  2480. case HTMLToken::Type::EndTag:
  2481. offset = 2;
  2482. break;
  2483. default:
  2484. break;
  2485. }
  2486. m_current_token.m_start_position = nth_last_position(offset);
  2487. }
  2488. HTMLTokenizer::HTMLTokenizer(const StringView& input, const String& encoding)
  2489. {
  2490. auto* decoder = TextCodec::decoder_for(encoding);
  2491. VERIFY(decoder);
  2492. m_decoded_input = decoder->to_utf8(input);
  2493. m_utf8_view = Utf8View(m_decoded_input);
  2494. m_utf8_iterator = m_utf8_view.begin();
  2495. m_source_positions.empend(0u, 0u);
  2496. }
  2497. void HTMLTokenizer::will_switch_to([[maybe_unused]] State new_state)
  2498. {
  2499. dbgln_if(TOKENIZER_TRACE_DEBUG, "[{}] Switch to {}", state_name(m_state), state_name(new_state));
  2500. }
  2501. void HTMLTokenizer::will_reconsume_in([[maybe_unused]] State new_state)
  2502. {
  2503. dbgln_if(TOKENIZER_TRACE_DEBUG, "[{}] Reconsume in {}", state_name(m_state), state_name(new_state));
  2504. }
  2505. void HTMLTokenizer::switch_to(Badge<HTMLDocumentParser>, State new_state)
  2506. {
  2507. dbgln_if(TOKENIZER_TRACE_DEBUG, "[{}] Parser switches tokenizer state to {}", state_name(m_state), state_name(new_state));
  2508. m_state = new_state;
  2509. }
  2510. void HTMLTokenizer::will_emit(HTMLToken& token)
  2511. {
  2512. if (token.is_start_tag())
  2513. m_last_emitted_start_tag = token;
  2514. token.m_end_position = m_source_positions.last();
  2515. }
  2516. bool HTMLTokenizer::current_end_tag_token_is_appropriate() const
  2517. {
  2518. VERIFY(m_current_token.is_end_tag());
  2519. if (!m_last_emitted_start_tag.is_start_tag())
  2520. return false;
  2521. return m_current_token.tag_name() == m_last_emitted_start_tag.tag_name();
  2522. }
  2523. bool HTMLTokenizer::consumed_as_part_of_an_attribute() const
  2524. {
  2525. return m_return_state == State::AttributeValueUnquoted || m_return_state == State::AttributeValueSingleQuoted || m_return_state == State::AttributeValueDoubleQuoted;
  2526. }
  2527. void HTMLTokenizer::restore_to(const Utf8CodepointIterator& new_iterator)
  2528. {
  2529. if (new_iterator != m_prev_utf8_iterator) {
  2530. auto diff = m_prev_utf8_iterator - new_iterator;
  2531. if (diff > 0) {
  2532. for (ssize_t i = 0; i < diff; ++i)
  2533. m_source_positions.take_last();
  2534. } else {
  2535. // Going forwards...?
  2536. TODO();
  2537. }
  2538. }
  2539. }
  2540. }