single-file-core.js 88 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree, sessionId = 0;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. if (options.sessionId === undefined) {
  34. options.sessionId = sessionId;
  35. sessionId++;
  36. }
  37. }
  38. async run() {
  39. if (this.options.userScriptEnabled) {
  40. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  41. }
  42. this.runner = new Runner(this.options, true);
  43. await this.runner.loadPage();
  44. await this.runner.initialize();
  45. if (this.options.userScriptEnabled) {
  46. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  47. }
  48. await this.runner.run();
  49. }
  50. cancel() {
  51. this.cancelled = true;
  52. if (this.runner) {
  53. this.runner.cancel();
  54. }
  55. }
  56. getPageData() {
  57. return this.runner.getPageData();
  58. }
  59. }
  60. // -------------
  61. // ProgressEvent
  62. // -------------
  63. const PAGE_LOADING = "page-loading";
  64. const PAGE_LOADED = "page-loaded";
  65. const RESOURCES_INITIALIZING = "resource-initializing";
  66. const RESOURCES_INITIALIZED = "resources-initialized";
  67. const RESOURCE_LOADED = "resource-loaded";
  68. const PAGE_ENDED = "page-ended";
  69. const STAGE_STARTED = "stage-started";
  70. const STAGE_ENDED = "stage-ended";
  71. const STAGE_TASK_STARTED = "stage-task-started";
  72. const STAGE_TASK_ENDED = "stage-task-ended";
  73. class ProgressEvent {
  74. constructor(type, detail) {
  75. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  76. }
  77. }
  78. // ------
  79. // Runner
  80. // ------
  81. const RESOLVE_URLS_STAGE = 0;
  82. const REPLACE_DATA_STAGE = 1;
  83. const REPLACE_DOCS_STAGE = 2;
  84. const POST_PROCESS_STAGE = 3;
  85. const STAGES = [{
  86. sequential: [
  87. { action: "preProcessPage" },
  88. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  89. { action: "replaceStyleContents" },
  90. { action: "resetCharsetMeta" },
  91. { option: "saveFavicon", action: "saveFavicon" },
  92. { action: "replaceCanvasElements" },
  93. { action: "insertFonts" },
  94. { action: "insertShadowRootContents" },
  95. { action: "setInputValues" },
  96. { option: "removeScripts", action: "removeScripts" },
  97. { option: "selected", action: "removeUnselectedElements" },
  98. { option: "removeVideoSrc", action: "insertVideoPosters" },
  99. { option: "removeFrames", action: "removeFrames" },
  100. { option: "removeVideoSrc", action: "removeVideoSources" },
  101. { option: "removeAudioSrc", action: "removeAudioSources" },
  102. { action: "removeDiscardedResources" },
  103. { option: "removeHiddenElements", action: "removeHiddenElements" },
  104. { action: "resolveHrefs" },
  105. { action: "resolveStyleAttributeURLs" }
  106. ],
  107. parallel: [
  108. { action: "resolveStylesheetURLs" },
  109. { option: "!removeFrames", action: "resolveFrameURLs" },
  110. { action: "resolveHtmlImportURLs" }
  111. ]
  112. }, {
  113. sequential: [
  114. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  115. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  116. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  117. ],
  118. parallel: [
  119. { action: "processStylesheets" },
  120. { action: "processStyleAttributes" },
  121. { action: "processPageResources" },
  122. { option: "!removeScripts", action: "processScripts" }
  123. ]
  124. }, {
  125. sequential: [
  126. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  127. ],
  128. parallel: [
  129. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  130. { option: "!removeFrames", action: "processFrames" },
  131. { option: "!removeImports", action: "processHtmlImports" },
  132. ]
  133. }, {
  134. sequential: [
  135. { action: "replaceStylesheets" },
  136. { action: "replaceStyleAttributes" },
  137. { action: "insertVariables" },
  138. { option: "compressHTML", action: "compressHTML" },
  139. { action: "cleanupPage" }
  140. ],
  141. parallel: [
  142. { option: "enableMaff", action: "insertMAFFMetaData" },
  143. { action: "setDocInfo" }
  144. ]
  145. }];
  146. class Runner {
  147. constructor(options, root) {
  148. const rootDocDefined = root && options.doc;
  149. this.root = root;
  150. this.options = options;
  151. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  152. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  153. this.options.rootDocument = root;
  154. this.options.updatedResources = this.options.updatedResources || {};
  155. this.batchRequest = new BatchRequest();
  156. this.processor = new Processor(options, this.batchRequest);
  157. if (rootDocDefined) {
  158. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  159. this.options.canvases = docData.canvases;
  160. this.options.fonts = docData.fonts;
  161. this.options.stylesheets = docData.stylesheets;
  162. this.options.images = docData.images;
  163. this.options.posters = docData.posters;
  164. this.options.usedFonts = docData.usedFonts;
  165. this.options.shadowRoots = docData.shadowRoots;
  166. this.options.imports = docData.imports;
  167. this.options.referrer = docData.referrer;
  168. this.markedElements = docData.markedElements;
  169. }
  170. if (this.options.saveRawPage) {
  171. this.options.removeFrames = true;
  172. }
  173. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  174. this.onprogress = options.onprogress || (() => { });
  175. }
  176. async loadPage() {
  177. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  178. await this.processor.loadPage(this.options.content);
  179. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  180. }
  181. async initialize() {
  182. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  183. await this.executeStage(RESOLVE_URLS_STAGE);
  184. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  185. if (this.root && this.options.doc) {
  186. util.postProcessDoc(this.options.doc, this.markedElements);
  187. }
  188. }
  189. cancel() {
  190. this.cancelled = true;
  191. this.batchRequest.cancel();
  192. if (this.root) {
  193. if (this.options.frames) {
  194. this.options.frames.forEach(cancelRunner);
  195. }
  196. if (this.options.imports) {
  197. this.options.imports.forEach(cancelRunner);
  198. }
  199. }
  200. function cancelRunner(resourceData) {
  201. if (resourceData.runner) {
  202. resourceData.runner.cancel();
  203. }
  204. }
  205. }
  206. async run() {
  207. if (this.root) {
  208. this.processor.initialize(this.batchRequest);
  209. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  210. }
  211. await this.batchRequest.run(detail => {
  212. detail.pageURL = this.options.url;
  213. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  214. }, this.options);
  215. await this.pendingPromises;
  216. this.options.doc = null;
  217. this.options.win = null;
  218. await this.executeStage(REPLACE_DOCS_STAGE);
  219. await this.executeStage(POST_PROCESS_STAGE);
  220. this.processor.finalize();
  221. }
  222. getDocument() {
  223. return this.processor.doc;
  224. }
  225. getStyleSheets() {
  226. return this.processor.stylesheets;
  227. }
  228. getPageData() {
  229. if (this.root) {
  230. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  231. }
  232. return this.processor.getPageData();
  233. }
  234. async executeStage(step) {
  235. if (DEBUG) {
  236. log("**** STARTED STAGE", step, "****");
  237. }
  238. const frame = !this.root;
  239. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  240. STAGES[step].sequential.forEach(task => {
  241. let startTime;
  242. if (DEBUG) {
  243. startTime = Date.now();
  244. log(" -- STARTED task =", task.action);
  245. }
  246. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  247. if (!this.cancelled) {
  248. this.executeTask(task);
  249. }
  250. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  251. if (DEBUG) {
  252. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  253. }
  254. });
  255. let parallelTasksPromise;
  256. if (STAGES[step].parallel) {
  257. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  258. let startTime;
  259. if (DEBUG) {
  260. startTime = Date.now();
  261. log(" // STARTED task =", task.action);
  262. }
  263. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  264. if (!this.cancelled) {
  265. await this.executeTask(task);
  266. }
  267. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  268. if (DEBUG) {
  269. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  270. }
  271. }));
  272. } else {
  273. parallelTasksPromise = Promise.resolve();
  274. }
  275. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  276. if (DEBUG) {
  277. log("**** ENDED STAGE", step, "****");
  278. }
  279. return parallelTasksPromise;
  280. }
  281. executeTask(task) {
  282. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  283. return this.processor[task.action]();
  284. }
  285. }
  286. }
  287. // ------------
  288. // BatchRequest
  289. // ------------
  290. class BatchRequest {
  291. constructor() {
  292. this.requests = new Map();
  293. this.duplicates = new Map();
  294. }
  295. addURL(resourceURL, asBinary, groupDuplicates) {
  296. return new Promise((resolve, reject) => {
  297. const requestKey = JSON.stringify([resourceURL, asBinary]);
  298. let resourceRequests = this.requests.get(requestKey);
  299. if (!resourceRequests) {
  300. resourceRequests = [];
  301. this.requests.set(requestKey, resourceRequests);
  302. }
  303. const callbacks = { resolve, reject };
  304. resourceRequests.push(callbacks);
  305. if (groupDuplicates) {
  306. let duplicateRequests = this.duplicates.get(requestKey);
  307. if (!duplicateRequests) {
  308. duplicateRequests = [];
  309. this.duplicates.set(requestKey, duplicateRequests);
  310. }
  311. duplicateRequests.push(callbacks);
  312. }
  313. });
  314. }
  315. getMaxResources() {
  316. return this.requests.size;
  317. }
  318. run(onloadListener, options) {
  319. const resourceURLs = [...this.requests.keys()];
  320. let indexResource = 0;
  321. return Promise.all(resourceURLs.map(async requestKey => {
  322. const [resourceURL, asBinary] = JSON.parse(requestKey);
  323. const resourceRequests = this.requests.get(requestKey);
  324. try {
  325. const currentIndexResource = indexResource;
  326. indexResource = indexResource + 1;
  327. const content = await util.getContent(resourceURL, {
  328. asBinary,
  329. maxResourceSize: options.maxResourceSize,
  330. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  331. frameId: options.windowId
  332. });
  333. onloadListener({ url: resourceURL });
  334. if (!this.cancelled) {
  335. resourceRequests.forEach(callbacks => {
  336. const duplicateCallbacks = this.duplicates.get(requestKey);
  337. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  338. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  339. });
  340. }
  341. } catch (error) {
  342. indexResource = indexResource + 1;
  343. onloadListener({ url: resourceURL });
  344. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  345. }
  346. this.requests.delete(requestKey);
  347. }));
  348. }
  349. cancel() {
  350. this.cancelled = true;
  351. const resourceURLs = [...this.requests.keys()];
  352. resourceURLs.forEach(requestKey => {
  353. const resourceRequests = this.requests.get(requestKey);
  354. resourceRequests.forEach(callbacks => callbacks.reject());
  355. this.requests.delete(requestKey);
  356. });
  357. }
  358. }
  359. // ---------
  360. // Processor
  361. // ---------
  362. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  363. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  364. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  365. const SCRIPT_TAG_FOUND = /<script/gi;
  366. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  367. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  368. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  369. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  370. const UTF8_CHARSET = "utf-8";
  371. class Processor {
  372. constructor(options, batchRequest) {
  373. this.options = options;
  374. this.stats = new Stats(options);
  375. this.baseURI = normalizeURL(options.baseURI || options.url);
  376. this.batchRequest = batchRequest;
  377. this.stylesheets = new Map();
  378. this.styles = new Map();
  379. this.cssVariables = new Map();
  380. }
  381. initialize() {
  382. this.options.saveDate = new Date();
  383. this.options.saveUrl = this.options.url;
  384. if (this.options.enableMaff) {
  385. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  386. }
  387. this.maxResources = this.batchRequest.getMaxResources();
  388. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  389. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  390. }
  391. if (!this.options.removeImports && this.options.imports) {
  392. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  393. }
  394. this.stats.set("processed", "resources", this.maxResources);
  395. }
  396. async loadPage(pageContent, charset) {
  397. let content;
  398. if (!pageContent || this.options.saveRawPage) {
  399. content = await util.getContent(this.baseURI, {
  400. maxResourceSize: this.options.maxResourceSize,
  401. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  402. charset,
  403. frameId: this.options.windowId
  404. });
  405. pageContent = content.data;
  406. }
  407. this.doc = util.parseDocContent(pageContent, this.baseURI);
  408. if (this.options.saveRawPage) {
  409. let charset;
  410. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  411. const charsetDeclaration = element.content.split(";")[1];
  412. if (charsetDeclaration && !charset) {
  413. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  414. }
  415. });
  416. if (charset && charset != content.charset) {
  417. return this.loadPage(pageContent, charset);
  418. }
  419. }
  420. this.workStyleElement = this.doc.createElement("style");
  421. this.doc.body.appendChild(this.workStyleElement);
  422. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  423. }
  424. finalize() {
  425. if (this.workStyleElement.parentNode) {
  426. this.workStyleElement.remove();
  427. }
  428. }
  429. async getPageData() {
  430. util.postProcessDoc(this.doc);
  431. const url = util.parseURL(this.baseURI);
  432. if (this.options.insertSingleFileComment) {
  433. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  434. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  435. " \n url: " + this.options.saveUrl +
  436. " \n saved date: " + this.options.saveDate +
  437. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  438. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  439. }
  440. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  441. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  442. if (!canonicalLink) {
  443. canonicalLink = this.doc.createElement("link");
  444. canonicalLink.setAttribute("rel", "canonical");
  445. this.doc.head.appendChild(canonicalLink);
  446. }
  447. if (canonicalLink && !canonicalLink.href) {
  448. canonicalLink.href = this.options.saveUrl;
  449. }
  450. }
  451. let size;
  452. if (this.options.displayStats) {
  453. size = util.getContentSize(this.doc.documentElement.outerHTML);
  454. }
  455. const content = util.serialize(this.doc, this.options.compressHTML);
  456. if (this.options.displayStats) {
  457. const contentSize = util.getContentSize(content);
  458. this.stats.set("processed", "HTML bytes", contentSize);
  459. this.stats.add("discarded", "HTML bytes", size - contentSize);
  460. }
  461. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  462. const replacementCharacter = this.options.filenameReplacementCharacter;
  463. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  464. if (!this.options.backgroundSave) {
  465. filename = filename.replace(/\//g, replacementCharacter);
  466. }
  467. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  468. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  469. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  470. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  471. filename = filename + "…" + extension;
  472. }
  473. if (!filename) {
  474. filename = "Unnamed page";
  475. }
  476. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  477. const pageData = {
  478. stats: this.stats.data,
  479. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  480. filename,
  481. content
  482. };
  483. if (this.options.addProof) {
  484. pageData.hash = await util.digest("SHA-256", content);
  485. }
  486. if (this.options.retrieveLinks) {
  487. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  488. }
  489. return pageData;
  490. }
  491. preProcessPage() {
  492. if (this.options.win) {
  493. this.doc.body.querySelectorAll(":not(svg) title, meta, link[href][rel*=\"icon\"]").forEach(element => element instanceof this.options.win.HTMLElement && this.doc.head.appendChild(element));
  494. }
  495. if (this.options.images && !this.options.saveRawPage) {
  496. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  497. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  498. if (attributeValue) {
  499. const imageData = this.options.images[Number(attributeValue)];
  500. if (imageData) {
  501. if (this.options.removeHiddenElements && (
  502. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  503. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  504. )) {
  505. imgElement.setAttribute("src", EMPTY_IMAGE);
  506. } else {
  507. if (imageData.currentSrc) {
  508. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  509. imgElement.setAttribute("src", imageData.currentSrc);
  510. }
  511. if (this.options.loadDeferredImages) {
  512. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  513. imageData.src = imgElement.dataset.src;
  514. imgElement.setAttribute("src", imgElement.dataset.src);
  515. imgElement.removeAttribute("data-src");
  516. }
  517. }
  518. }
  519. }
  520. }
  521. });
  522. if (this.options.loadDeferredImages) {
  523. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  524. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  525. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  526. imgElement.removeAttribute("data-srcset");
  527. }
  528. });
  529. }
  530. }
  531. }
  532. replaceStyleContents() {
  533. if (this.options.stylesheets) {
  534. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  535. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  536. if (attributeValue) {
  537. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  538. if (stylesheetContent) {
  539. styleElement.textContent = stylesheetContent;
  540. }
  541. }
  542. });
  543. }
  544. }
  545. removeUnselectedElements() {
  546. removeUnmarkedElements(this.doc.body);
  547. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  548. function removeUnmarkedElements(element) {
  549. let selectedElementFound = false;
  550. Array.from(element.childNodes).forEach(node => {
  551. if (node.nodeType == 1) {
  552. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  553. selectedElementFound = selectedElementFound || isSelectedElement;
  554. if (isSelectedElement) {
  555. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  556. removeUnmarkedElements(node);
  557. } else if (selectedElementFound) {
  558. removeNode(node);
  559. } else {
  560. hideNode(node);
  561. }
  562. }
  563. });
  564. }
  565. function removeNode(node) {
  566. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  567. node.remove();
  568. } else {
  569. hideNode(node);
  570. }
  571. }
  572. function hideNode(node) {
  573. if (canHideNode(node)) {
  574. node.style.setProperty("display", "none", "important");
  575. Array.from(node.childNodes).forEach(removeNode);
  576. }
  577. }
  578. function canHideNode(node) {
  579. if (node.nodeType == 1) {
  580. const tagName = node.tagName && node.tagName.toLowerCase();
  581. return (tagName != "svg" && tagName != "style" && tagName != "link");
  582. }
  583. }
  584. }
  585. insertVideoPosters() {
  586. if (this.options.posters) {
  587. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  588. let videoElement;
  589. if (element.tagName == "VIDEO") {
  590. videoElement = element;
  591. } else {
  592. videoElement = element.parentElement;
  593. }
  594. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  595. if (attributeValue) {
  596. const posterURL = this.options.posters[Number(attributeValue)];
  597. if (!videoElement.poster && posterURL) {
  598. videoElement.setAttribute("poster", posterURL);
  599. }
  600. }
  601. });
  602. }
  603. }
  604. removeFrames() {
  605. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  606. this.stats.set("discarded", "frames", frameElements.length);
  607. this.stats.set("processed", "frames", frameElements.length);
  608. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  609. }
  610. removeScripts() {
  611. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  612. this.doc.querySelectorAll("[href]").forEach(element => {
  613. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  614. element.setAttribute("href", "");
  615. }
  616. });
  617. this.doc.querySelectorAll("[src]").forEach(element => {
  618. if (element.src && element.src.match(/^\s*javascript:/)) {
  619. element.removeAttribute("src");
  620. }
  621. });
  622. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  623. this.stats.set("discarded", "scripts", scriptElements.length);
  624. this.stats.set("processed", "scripts", scriptElements.length);
  625. scriptElements.forEach(element => element.remove());
  626. }
  627. removeVideoSources() {
  628. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  629. this.stats.set("discarded", "video sources", videoSourceElements.length);
  630. this.stats.set("processed", "video sources", videoSourceElements.length);
  631. videoSourceElements.forEach(element => {
  632. if (element.tagName == "SOURCE") {
  633. element.remove();
  634. } else {
  635. videoSourceElements.forEach(element => element.removeAttribute("src"));
  636. }
  637. });
  638. }
  639. removeAudioSources() {
  640. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  641. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  642. this.stats.set("processed", "audio sources", audioSourceElements.length);
  643. audioSourceElements.forEach(element => {
  644. if (element.tagName == "SOURCE") {
  645. element.remove();
  646. } else {
  647. audioSourceElements.forEach(element => element.removeAttribute("src"));
  648. }
  649. });
  650. }
  651. removeDiscardedResources() {
  652. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  653. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  654. const objectElements = this.doc.querySelectorAll("applet, object[data]:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed[src]:not([src*=\".svg\"]):not([src*=\".pdf\"])");
  655. this.stats.set("discarded", "objects", objectElements.length);
  656. this.stats.set("processed", "objects", objectElements.length);
  657. objectElements.forEach(element => element.remove());
  658. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  659. replacedAttributeValue.forEach(element => {
  660. let regExp;
  661. if (this.options.removeScripts) {
  662. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  663. } else {
  664. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  665. }
  666. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  667. if (relValue.length) {
  668. element.setAttribute("rel", relValue);
  669. } else {
  670. element.remove();
  671. }
  672. });
  673. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  674. if (this.options.compressHTML) {
  675. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  676. }
  677. if (!this.options.saveFavicon) {
  678. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  679. }
  680. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  681. }
  682. resetCharsetMeta() {
  683. let charset;
  684. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  685. const charsetDeclaration = element.content.split(";")[1];
  686. if (charsetDeclaration && !charset) {
  687. charset = charsetDeclaration.split("=")[1];
  688. if (charset) {
  689. this.charset = charset.trim().toLowerCase();
  690. }
  691. }
  692. element.remove();
  693. });
  694. const metaElement = this.doc.createElement("meta");
  695. metaElement.setAttribute("charset", UTF8_CHARSET);
  696. if (this.doc.head.firstChild) {
  697. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  698. } else {
  699. this.doc.head.appendChild(metaElement);
  700. }
  701. }
  702. setInputValues() {
  703. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  704. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  705. input.setAttribute("value", value || "");
  706. });
  707. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  708. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  709. if (value == "true") {
  710. input.setAttribute("checked", "");
  711. }
  712. });
  713. this.doc.querySelectorAll("textarea").forEach(textarea => {
  714. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  715. textarea.textContent = value || "";
  716. });
  717. this.doc.querySelectorAll("select").forEach(select => {
  718. select.querySelectorAll("option").forEach(option => {
  719. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  720. if (selected) {
  721. option.setAttribute("selected", "");
  722. }
  723. });
  724. });
  725. }
  726. saveFavicon() {
  727. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  728. if (!faviconElement) {
  729. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  730. }
  731. if (!faviconElement) {
  732. faviconElement = this.doc.createElement("link");
  733. faviconElement.setAttribute("type", "image/x-icon");
  734. faviconElement.setAttribute("rel", "shortcut icon");
  735. faviconElement.setAttribute("href", "/favicon.ico");
  736. }
  737. this.doc.head.appendChild(faviconElement);
  738. }
  739. replaceCanvasElements() {
  740. if (this.options.canvases) {
  741. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  742. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  743. if (attributeValue) {
  744. const canvasData = this.options.canvases[Number(attributeValue)];
  745. if (canvasData) {
  746. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  747. this.stats.add("processed", "canvas", 1);
  748. }
  749. }
  750. });
  751. }
  752. }
  753. insertFonts() {
  754. if (this.options.fonts && this.options.fonts.length) {
  755. let stylesheetContent = "";
  756. this.options.fonts.forEach(fontData => {
  757. if (fontData["font-family"] && fontData.src) {
  758. stylesheetContent += "@font-face{";
  759. let stylesContent = "";
  760. Object.keys(fontData).forEach(fontStyle => {
  761. if (stylesContent) {
  762. stylesContent += ";";
  763. }
  764. stylesContent += fontStyle + ":" + fontData[fontStyle];
  765. });
  766. stylesheetContent += stylesContent + "}";
  767. }
  768. });
  769. if (stylesheetContent) {
  770. const styleElement = this.doc.createElement("style");
  771. styleElement.textContent = stylesheetContent;
  772. const existingStyleElement = this.doc.querySelector("style");
  773. if (existingStyleElement) {
  774. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  775. } else {
  776. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  777. }
  778. }
  779. }
  780. }
  781. removeHiddenElements() {
  782. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  783. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  784. this.stats.set("discarded", "hidden elements", removedElements.length);
  785. this.stats.set("processed", "hidden elements", removedElements.length);
  786. if (hiddenElements.length) {
  787. const styleElement = this.doc.createElement("style");
  788. styleElement.textContent = ".sf-hidden{display:none!important;}";
  789. this.doc.head.appendChild(styleElement);
  790. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  791. }
  792. removedElements.forEach(element => element.remove());
  793. }
  794. resolveHrefs() {
  795. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  796. const href = element.getAttribute("href").trim();
  797. if (!testIgnoredPath(href)) {
  798. let resolvedURL;
  799. try {
  800. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  801. } catch (error) {
  802. // ignored
  803. }
  804. if (resolvedURL) {
  805. const url = normalizeURL(this.options.url);
  806. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  807. resolvedURL = resolvedURL.substring(url.length);
  808. }
  809. try {
  810. element.setAttribute("href", resolvedURL);
  811. } catch (error) {
  812. // ignored
  813. }
  814. }
  815. }
  816. });
  817. }
  818. resolveStyleAttributeURLs() {
  819. this.doc.querySelectorAll("[style]").forEach(element => {
  820. let styleContent = element.getAttribute("style");
  821. if (this.options.compressCSS) {
  822. styleContent = util.compressCSS(styleContent);
  823. }
  824. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  825. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  826. this.styles.set(element, declarationList);
  827. });
  828. }
  829. async resolveStylesheetURLs() {
  830. const options = {
  831. maxResourceSize: this.options.maxResourceSize,
  832. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  833. url: this.options.url,
  834. charset: this.charset,
  835. compressCSS: this.options.compressCSS,
  836. updatedResources: this.options.updatedResources,
  837. rootDocument: this.options.rootDocument,
  838. frameId: this.options.windowId
  839. };
  840. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  841. .map(async element => {
  842. let mediaText;
  843. if (element.media) {
  844. mediaText = element.media.toLowerCase();
  845. }
  846. const stylesheetInfo = { mediaText };
  847. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  848. stylesheetInfo.scoped = true;
  849. }
  850. if (element.tagName.toLowerCase() == "link") {
  851. if (element.charset) {
  852. options.charset = element.charset;
  853. }
  854. }
  855. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  856. }));
  857. if (options.rootDocument) {
  858. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  859. await Promise.all(newResources.map(async resource => {
  860. resource.retrieved = true;
  861. const stylesheetInfo = {};
  862. const element = this.doc.createElement("style");
  863. this.doc.body.appendChild(element);
  864. element.textContent = resource.content;
  865. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  866. }));
  867. }
  868. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  869. stylesheets.set(element, stylesheetInfo);
  870. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  871. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  872. if (match && match[1] && match[1] != options.charset) {
  873. options.charset = match[1];
  874. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  875. }
  876. let stylesheet;
  877. try {
  878. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  879. } catch (error) {
  880. // ignored
  881. }
  882. if (stylesheet && stylesheet.children) {
  883. if (options.compressCSS) {
  884. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  885. }
  886. stylesheetInfo.stylesheet = stylesheet;
  887. } else {
  888. stylesheets.delete(element);
  889. }
  890. }
  891. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  892. let content;
  893. if (element.tagName.toLowerCase() == "link") {
  894. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  895. } else {
  896. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  897. }
  898. return content || "";
  899. }
  900. }
  901. async resolveFrameURLs() {
  902. if (!this.options.saveRawPage) {
  903. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  904. await Promise.all(frameElements.map(async frameElement => {
  905. if (frameElement.tagName == "OBJECT") {
  906. frameElement.setAttribute("data", "data:text/html,");
  907. } else {
  908. frameElement.removeAttribute("src");
  909. frameElement.removeAttribute("srcdoc");
  910. }
  911. Array.from(frameElement.childNodes).forEach(node => node.remove());
  912. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  913. if (this.options.frames && frameWindowId) {
  914. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  915. if (frameData) {
  916. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  917. }
  918. }
  919. }));
  920. }
  921. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  922. options.insertSingleFileComment = false;
  923. options.insertCanonicalLink = false;
  924. options.saveFavicon = false;
  925. options.url = frameData.baseURI;
  926. options.windowId = frameWindowId;
  927. if (frameData.content) {
  928. options.content = frameData.content;
  929. options.canvases = frameData.canvases;
  930. options.fonts = frameData.fonts;
  931. options.stylesheets = frameData.stylesheets;
  932. options.images = frameData.images;
  933. options.posters = frameData.posters;
  934. options.usedFonts = frameData.usedFonts;
  935. options.shadowRoots = frameData.shadowRoots;
  936. options.imports = frameData.imports;
  937. frameData.runner = new Runner(options);
  938. frameData.frameElement = frameElement;
  939. await frameData.runner.loadPage();
  940. await frameData.runner.initialize();
  941. frameData.maxResources = batchRequest.getMaxResources();
  942. }
  943. }
  944. }
  945. insertShadowRootContents() {
  946. const doc = this.doc;
  947. const options = this.options;
  948. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  949. processElement(this.doc);
  950. if (this.options.removeScripts) {
  951. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  952. }
  953. const scriptElement = doc.createElement("script");
  954. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  955. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{if (!element.parentElement.shadowRoot) {const shadowRoot=element.parentElement.attachShadow({mode:element.getAttribute("${SHADOW_MODE_ATTRIBUTE_NAME}"),delegatesFocus:Boolean(element.getAttribute("${SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME}"))});shadowRoot.innerHTML=element.innerHTML;element.remove();processNode(shadowRoot)}})}})()`;
  956. doc.body.appendChild(scriptElement);
  957. }
  958. function processElement(element) {
  959. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  960. shadowRootElements.forEach(element => {
  961. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  962. if (attributeValue) {
  963. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  964. if (shadowRootData) {
  965. const templateElement = doc.createElement("template");
  966. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  967. if (shadowRootData.delegatesFocus) {
  968. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  969. }
  970. if (shadowRootData.adoptedStyleSheets) {
  971. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  972. const styleElement = doc.createElement("style");
  973. styleElement.textContent = stylesheetContent;
  974. templateElement.appendChild(styleElement);
  975. });
  976. }
  977. const shadowDoc = util.parseDocContent(shadowRootData.content);
  978. if (shadowDoc.head) {
  979. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  980. if (metaCharset) {
  981. metaCharset.remove();
  982. }
  983. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  984. }
  985. if (shadowDoc.body) {
  986. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  987. }
  988. processElement(templateElement);
  989. if (element.firstChild) {
  990. element.insertBefore(templateElement, element.firstChild);
  991. } else {
  992. element.appendChild(templateElement);
  993. }
  994. }
  995. }
  996. });
  997. }
  998. }
  999. async resolveHtmlImportURLs() {
  1000. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1001. await Promise.all(linkElements.map(async linkElement => {
  1002. const resourceURL = linkElement.href;
  1003. linkElement.removeAttribute("href");
  1004. const options = Object.create(this.options);
  1005. options.insertSingleFileComment = false;
  1006. options.insertCanonicalLink = false;
  1007. options.saveFavicon = false;
  1008. options.removeUnusedStyles = false;
  1009. options.removeAlternativeMedias = false;
  1010. options.removeUnusedFonts = false;
  1011. options.removeHiddenElements = false;
  1012. options.url = resourceURL;
  1013. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1014. if (attributeValue) {
  1015. const importData = options.imports[Number(attributeValue)];
  1016. if (importData) {
  1017. options.content = importData.content;
  1018. importData.runner = new Runner(options);
  1019. await importData.runner.loadPage();
  1020. await importData.runner.initialize();
  1021. if (!options.removeImports) {
  1022. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1023. }
  1024. importData.runner.getStyleSheets().forEach(stylesheet => {
  1025. const importedStyleElement = this.doc.createElement("style");
  1026. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1027. this.stylesheets.set(importedStyleElement, stylesheet);
  1028. });
  1029. }
  1030. }
  1031. if (options.removeImports) {
  1032. linkElement.remove();
  1033. this.stats.add("discarded", "HTML imports", 1);
  1034. }
  1035. }));
  1036. }
  1037. removeUnusedStyles() {
  1038. if (!this.mediaAllInfo) {
  1039. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1040. }
  1041. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1042. this.stats.set("processed", "CSS rules", stats.processed);
  1043. this.stats.set("discarded", "CSS rules", stats.discarded);
  1044. }
  1045. removeUnusedFonts() {
  1046. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1047. }
  1048. removeAlternativeMedias() {
  1049. const stats = util.minifyMedias(this.stylesheets);
  1050. this.stats.set("processed", "medias", stats.processed);
  1051. this.stats.set("discarded", "medias", stats.discarded);
  1052. }
  1053. async processStylesheets() {
  1054. this.options.fontURLs = new Map();
  1055. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1056. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1057. ));
  1058. }
  1059. async processStyleAttributes() {
  1060. return Promise.all([...this.styles].map(([, declarationList]) =>
  1061. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1062. ));
  1063. }
  1064. async processPageResources() {
  1065. const processAttributeArgs = [
  1066. ["link[href][rel*=\"icon\"]", "href", false, true],
  1067. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1068. ["img[src], input[src][type=image]", "src", true],
  1069. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1070. ["video[poster]", "poster"],
  1071. ["*[background]", "background"],
  1072. ["image", "xlink:href"],
  1073. ["image", "href"]
  1074. ];
  1075. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1076. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1077. );
  1078. resourcePromises = resourcePromises.concat([
  1079. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1080. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1081. ]);
  1082. if (!this.options.removeAudioSrc) {
  1083. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1084. }
  1085. if (!this.options.removeVideoSrc) {
  1086. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1087. }
  1088. await Promise.all(resourcePromises);
  1089. if (this.options.saveFavicon) {
  1090. ProcessorHelper.processShortcutIcons(this.doc);
  1091. }
  1092. }
  1093. async processScripts() {
  1094. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1095. let resourceURL;
  1096. let scriptSrc;
  1097. if (element.tagName == "SCRIPT") {
  1098. scriptSrc = element.getAttribute("src");
  1099. element.removeAttribute("src");
  1100. } else {
  1101. scriptSrc = element.getAttribute("href");
  1102. element.removeAttribute("href");
  1103. }
  1104. element.removeAttribute("integrity");
  1105. element.textContent = "";
  1106. try {
  1107. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1108. } catch (error) {
  1109. // ignored
  1110. }
  1111. if (testValidURL(resourceURL)) {
  1112. this.stats.add("processed", "scripts", 1);
  1113. const content = await util.getContent(resourceURL, {
  1114. asBinary: true,
  1115. charset: this.charset != UTF8_CHARSET && this.charset,
  1116. maxResourceSize: this.options.maxResourceSize,
  1117. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1118. frameId: this.options.windowId
  1119. });
  1120. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1121. if (element.tagName == "SCRIPT") {
  1122. element.setAttribute("src", content.data);
  1123. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1124. element.setAttribute("async", "");
  1125. }
  1126. } else {
  1127. const scriptElement = this.doc.createElement("script");
  1128. scriptElement.setAttribute("src", content.data);
  1129. scriptElement.setAttribute("async", "");
  1130. element.parentElement.replaceChild(scriptElement, element);
  1131. }
  1132. }
  1133. }));
  1134. }
  1135. removeAlternativeImages() {
  1136. util.removeAlternativeImages(this.doc);
  1137. }
  1138. async removeAlternativeFonts() {
  1139. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs);
  1140. }
  1141. async processFrames() {
  1142. if (this.options.frames) {
  1143. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1144. await Promise.all(frameElements.map(async frameElement => {
  1145. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1146. if (frameWindowId) {
  1147. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1148. if (frameData) {
  1149. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1150. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1151. this.stats.add("processed", "frames", 1);
  1152. await frameData.runner.run();
  1153. const pageData = await frameData.runner.getPageData();
  1154. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1155. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1156. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1157. sandbox += " allow-scripts allow-same-origin";
  1158. }
  1159. frameElement.setAttribute("sandbox", sandbox);
  1160. if (frameElement.tagName == "OBJECT") {
  1161. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1162. } else {
  1163. if (frameElement.tagName == "FRAME") {
  1164. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1165. } else {
  1166. frameElement.setAttribute("srcdoc", pageData.content);
  1167. frameElement.removeAttribute("src");
  1168. }
  1169. }
  1170. this.stats.addAll(pageData);
  1171. } else {
  1172. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1173. this.stats.add("discarded", "frames", 1);
  1174. }
  1175. }
  1176. }
  1177. }));
  1178. }
  1179. }
  1180. async processHtmlImports() {
  1181. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1182. await Promise.all(linkElements.map(async linkElement => {
  1183. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1184. if (attributeValue) {
  1185. const importData = this.options.imports[Number(attributeValue)];
  1186. if (importData.runner) {
  1187. this.stats.add("processed", "HTML imports", 1);
  1188. await importData.runner.run();
  1189. const pageData = await importData.runner.getPageData();
  1190. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1191. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1192. this.stats.addAll(pageData);
  1193. } else {
  1194. this.stats.add("discarded", "HTML imports", 1);
  1195. }
  1196. }
  1197. }));
  1198. }
  1199. replaceStylesheets() {
  1200. this.doc.querySelectorAll("style").forEach(styleElement => {
  1201. const stylesheetInfo = this.stylesheets.get(styleElement);
  1202. if (stylesheetInfo) {
  1203. this.stylesheets.delete(styleElement);
  1204. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1205. styleElement.textContent = stylesheetContent;
  1206. if (stylesheetInfo.mediaText) {
  1207. styleElement.media = stylesheetInfo.mediaText;
  1208. }
  1209. } else {
  1210. styleElement.remove();
  1211. }
  1212. });
  1213. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1214. const stylesheetInfo = this.stylesheets.get(linkElement);
  1215. if (stylesheetInfo) {
  1216. this.stylesheets.delete(linkElement);
  1217. const styleElement = this.doc.createElement("style");
  1218. if (stylesheetInfo.mediaText) {
  1219. styleElement.media = stylesheetInfo.mediaText;
  1220. }
  1221. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1222. styleElement.textContent = stylesheetContent;
  1223. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1224. } else {
  1225. linkElement.remove();
  1226. }
  1227. });
  1228. }
  1229. replaceStyleAttributes() {
  1230. this.doc.querySelectorAll("[style]").forEach(element => {
  1231. const declarations = this.styles.get(element);
  1232. if (declarations) {
  1233. this.styles.delete(element);
  1234. let styleContent = cssTree.generate(declarations);
  1235. element.setAttribute("style", styleContent);
  1236. } else {
  1237. element.setAttribute("style", "");
  1238. }
  1239. });
  1240. }
  1241. insertVariables() {
  1242. if (this.cssVariables.size) {
  1243. const styleElement = this.doc.createElement("style");
  1244. const firstStyleElement = this.doc.head.querySelector("style");
  1245. if (firstStyleElement) {
  1246. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1247. } else {
  1248. this.doc.head.appendChild(styleElement);
  1249. }
  1250. let stylesheetContent = "";
  1251. this.cssVariables.forEach((content, indexResource) => {
  1252. this.cssVariables.delete(indexResource);
  1253. if (stylesheetContent) {
  1254. stylesheetContent += ";";
  1255. }
  1256. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1257. });
  1258. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1259. }
  1260. }
  1261. compressHTML() {
  1262. let size;
  1263. if (this.options.displayStats) {
  1264. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1265. }
  1266. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1267. if (this.options.displayStats) {
  1268. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1269. }
  1270. }
  1271. cleanupPage() {
  1272. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1273. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1274. if (metaCharset) {
  1275. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1276. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1277. this.doc.head.querySelector("meta[charset]").remove();
  1278. }
  1279. }
  1280. }
  1281. resetZoomLevel() {
  1282. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1283. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1284. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1285. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1286. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1287. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1288. this.doc.documentElement.style.removeProperty("-sf-transform");
  1289. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1290. }
  1291. async insertMAFFMetaData() {
  1292. const maffMetaData = await this.maffMetaDataPromise;
  1293. if (maffMetaData && maffMetaData.content) {
  1294. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1295. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1296. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1297. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1298. if (originalURLElement) {
  1299. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1300. }
  1301. if (archiveTimeElement) {
  1302. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1303. if (value) {
  1304. const date = new Date(value);
  1305. if (!isNaN(date.getTime())) {
  1306. this.options.saveDate = new Date(value);
  1307. }
  1308. }
  1309. }
  1310. }
  1311. }
  1312. async setDocInfo() {
  1313. const titleElement = this.doc.querySelector("title");
  1314. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1315. const authorElement = this.doc.querySelector("meta[name=author]");
  1316. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1317. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1318. const headingElement = this.doc.querySelector("h1");
  1319. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1320. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1321. this.options.info = {
  1322. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1323. lang: this.doc.documentElement.lang,
  1324. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1325. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1326. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1327. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1328. };
  1329. }
  1330. }
  1331. // ---------------
  1332. // ProcessorHelper
  1333. // ---------------
  1334. const DATA_URI_PREFIX = "data:";
  1335. const ABOUT_BLANK_URI = "about:blank";
  1336. const EMPTY_DATA_URI = "data:null;base64,";
  1337. const REGEXP_URL_HASH = /(#.+?)$/;
  1338. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1339. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1340. class ProcessorHelper {
  1341. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1342. const date = options.saveDate;
  1343. const url = util.parseURL(options.saveUrl);
  1344. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1345. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1346. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1347. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1348. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1349. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1350. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1351. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1352. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1353. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1354. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1357. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1358. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1359. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1360. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1361. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1362. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1363. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1364. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1365. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1366. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1367. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1368. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1370. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1371. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1376. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1377. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1383. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1384. for (const [name, value] of params) {
  1385. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1386. }
  1387. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1388. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1392. if (content) {
  1393. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1396. }
  1397. return template.trim();
  1398. function decode(value) {
  1399. try {
  1400. return decodeURI(value);
  1401. } catch (error) {
  1402. return value;
  1403. }
  1404. }
  1405. }
  1406. static setBackgroundImage(element, url, style) {
  1407. element.style.setProperty("background-blend-mode", "normal", "important");
  1408. element.style.setProperty("background-clip", "content-box", "important");
  1409. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1410. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1411. element.style.setProperty("background-image", url, "important");
  1412. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1413. element.style.setProperty("background-origin", "content-box", "important");
  1414. element.style.setProperty("background-repeat", "no-repeat", "important");
  1415. }
  1416. static processShortcutIcons(doc) {
  1417. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1418. if (!shortcutIcon) {
  1419. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1420. if (shortcutIcon) {
  1421. shortcutIcon.rel = "icon";
  1422. }
  1423. }
  1424. if (shortcutIcon) {
  1425. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1426. if (linkElement != shortcutIcon) {
  1427. linkElement.remove();
  1428. }
  1429. });
  1430. }
  1431. }
  1432. static removeSingleLineCssComments(stylesheet) {
  1433. const removedRules = [];
  1434. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1435. const ruleData = cssRule.data;
  1436. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1437. removedRules.push(cssRule);
  1438. }
  1439. }
  1440. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1441. }
  1442. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1443. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1444. const imports = getImportFunctions(stylesheetContent);
  1445. await Promise.all(imports.map(async cssImport => {
  1446. const match = matchImport(cssImport);
  1447. if (match) {
  1448. const regExpCssImport = getRegExp(cssImport);
  1449. let resourceURL = normalizeURL(match.resourceURL);
  1450. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1451. try {
  1452. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1453. } catch (error) {
  1454. // ignored
  1455. }
  1456. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1457. const content = await util.getContent(resourceURL, {
  1458. maxResourceSize: options.maxResourceSize,
  1459. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1460. validateTextContentType: true,
  1461. frameId: options.frameId
  1462. });
  1463. resourceURL = content.resourceURL;
  1464. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1465. let importedStylesheetContent = removeCssComments(content.data);
  1466. if (options.compressCSS) {
  1467. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1468. }
  1469. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1470. if (stylesheetContent.includes(cssImport)) {
  1471. const ancestorStyleSheets = new Set(importedStyleSheets);
  1472. ancestorStyleSheets.add(resourceURL);
  1473. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1474. workStylesheet.textContent = importedStylesheetContent;
  1475. if (workStylesheet.sheet.cssRules.length) {
  1476. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1477. } else {
  1478. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1479. }
  1480. }
  1481. } else {
  1482. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1483. }
  1484. } else {
  1485. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1486. }
  1487. }
  1488. }));
  1489. return stylesheetContent;
  1490. }
  1491. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1492. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1493. urlFunctions.map(urlFunction => {
  1494. const originalResourceURL = matchURL(urlFunction);
  1495. const resourceURL = normalizeURL(originalResourceURL);
  1496. if (!testIgnoredPath(resourceURL)) {
  1497. if (!resourceURL || testValidPath(resourceURL)) {
  1498. let resolvedURL;
  1499. if (!originalResourceURL.startsWith("#")) {
  1500. try {
  1501. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1502. } catch (error) {
  1503. // ignored
  1504. }
  1505. }
  1506. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1507. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1508. }
  1509. } else {
  1510. let newUrlFunction;
  1511. if (originalResourceURL) {
  1512. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1513. } else {
  1514. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1515. }
  1516. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1517. }
  1518. }
  1519. });
  1520. return stylesheetContent;
  1521. }
  1522. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1523. resourceURL = normalizeURL(resourceURL);
  1524. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1525. const content = await util.getContent(resourceURL, {
  1526. maxResourceSize: options.maxResourceSize,
  1527. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1528. charset: options.charset,
  1529. frameId: options.frameId,
  1530. validateTextContentType: true
  1531. });
  1532. resourceURL = content.resourceURL;
  1533. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1534. let stylesheetContent = removeCssComments(content.data);
  1535. if (options.compressCSS) {
  1536. stylesheetContent = util.compressCSS(stylesheetContent);
  1537. }
  1538. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1539. return stylesheetContent;
  1540. }
  1541. }
  1542. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1543. const promises = [];
  1544. const removedRules = [];
  1545. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1546. const ruleData = cssRule.data;
  1547. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1548. removedRules.push(cssRule);
  1549. } else if (ruleData.block && ruleData.block.children) {
  1550. if (ruleData.type == "Rule") {
  1551. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1552. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1553. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1554. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1555. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1556. }
  1557. }
  1558. }
  1559. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1560. await Promise.all(promises);
  1561. async function processFontFaceRule(ruleData, fontURLs) {
  1562. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1563. if (declaration.type == "Declaration" && declaration.value.children) {
  1564. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1565. await Promise.all(urlFunctions.map(async urlFunction => {
  1566. const originalResourceURL = matchURL(urlFunction);
  1567. const resourceURL = normalizeURL(originalResourceURL);
  1568. if (!testIgnoredPath(resourceURL)) {
  1569. if (testValidURL(resourceURL)) {
  1570. let { content } = await batchRequest.addURL(resourceURL, true);
  1571. let resourceURLs = fontURLs.get(declaration);
  1572. if (!resourceURLs) {
  1573. resourceURLs = [];
  1574. fontURLs.set(declaration, resourceURLs);
  1575. }
  1576. resourceURLs.push(resourceURL);
  1577. replaceURLs(declaration, originalResourceURL, content);
  1578. }
  1579. }
  1580. }));
  1581. }
  1582. }));
  1583. function replaceURLs(declaration, oldURL, newURL) {
  1584. declaration.value.children.forEach(token => {
  1585. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1586. token.value.value = newURL;
  1587. }
  1588. });
  1589. }
  1590. }
  1591. }
  1592. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1593. await Promise.all(declarations.map(async declaration => {
  1594. if (declaration.type == "Declaration" && declaration.value.children) {
  1595. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1596. await Promise.all(urlFunctions.map(async urlFunction => {
  1597. const originalResourceURL = matchURL(urlFunction);
  1598. const resourceURL = normalizeURL(originalResourceURL);
  1599. if (!testIgnoredPath(resourceURL)) {
  1600. if (testValidURL(resourceURL)) {
  1601. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1602. let variableDefined;
  1603. const tokens = [];
  1604. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1605. if (!originalResourceURL.startsWith("#")) {
  1606. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1607. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1608. tokens.push({ parent, token, value });
  1609. variableDefined = true;
  1610. } else {
  1611. token.data.value.value = content;
  1612. }
  1613. }
  1614. });
  1615. if (variableDefined) {
  1616. cssVariables.set(indexResource, content);
  1617. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1618. }
  1619. }
  1620. }
  1621. }));
  1622. }
  1623. }));
  1624. function findURLToken(url, children, callback, depth = 0) {
  1625. for (let token = children.head; token; token = token.next) {
  1626. if (token.data.children) {
  1627. findURLToken(url, token.data.children, callback, depth + 1);
  1628. }
  1629. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1630. callback(token, children, depth == 0);
  1631. }
  1632. }
  1633. }
  1634. }
  1635. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1636. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1637. let resourceURL = resourceElement.getAttribute(attributeName);
  1638. if (resourceURL != null) {
  1639. resourceURL = normalizeURL(resourceURL);
  1640. if (!testIgnoredPath(resourceURL)) {
  1641. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1642. if (testValidPath(resourceURL)) {
  1643. try {
  1644. resourceURL = util.resolveURL(resourceURL, baseURI);
  1645. } catch (error) {
  1646. // ignored
  1647. }
  1648. if (testValidURL(resourceURL)) {
  1649. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1650. if (resourceElement.dataset.singleFileOriginURL) {
  1651. let originURL = resourceElement.dataset.singleFileOriginURL;
  1652. delete resourceElement.dataset.singleFileOriginURL;
  1653. if (content == EMPTY_DATA_URI) {
  1654. try {
  1655. originURL = util.resolveURL(originURL, baseURI);
  1656. } catch (error) {
  1657. // ignored
  1658. }
  1659. try {
  1660. resourceURL = originURL;
  1661. content = (await util.getContent(resourceURL, {
  1662. asBinary: true,
  1663. maxResourceSize: options.maxResourceSize,
  1664. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1665. frameId: options.windowId
  1666. })).data;
  1667. } catch (error) {
  1668. // ignored
  1669. }
  1670. }
  1671. }
  1672. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1673. resourceElement.remove();
  1674. } else {
  1675. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1676. if (!forbiddenPrefixFound) {
  1677. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1678. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1679. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1680. cssVariables.set(indexResource, content);
  1681. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1682. styles.set(resourceElement, declarationList);
  1683. } else {
  1684. resourceElement.setAttribute(attributeName, content);
  1685. }
  1686. } else {
  1687. resourceElement.setAttribute(attributeName, content);
  1688. }
  1689. }
  1690. }
  1691. }
  1692. }
  1693. }
  1694. }
  1695. }));
  1696. }
  1697. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1698. let attributeName = "xlink:href";
  1699. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1700. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1701. if (originalResourceURL == null) {
  1702. attributeName = "href";
  1703. originalResourceURL = resourceElement.getAttribute(attributeName);
  1704. }
  1705. let resourceURL = normalizeURL(originalResourceURL);
  1706. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1707. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1708. try {
  1709. resourceURL = util.resolveURL(resourceURL, baseURI);
  1710. } catch (error) {
  1711. // ignored
  1712. }
  1713. if (testValidURL(resourceURL)) {
  1714. const { content } = await batchRequest.addURL(resourceURL);
  1715. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1716. if (hashMatch && hashMatch[0]) {
  1717. let symbolElement;
  1718. try {
  1719. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1720. } catch (error) {
  1721. // ignored
  1722. }
  1723. if (symbolElement) {
  1724. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1725. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1726. }
  1727. } else {
  1728. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1729. }
  1730. }
  1731. } else if (resourceURL == options.url) {
  1732. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1733. }
  1734. }));
  1735. }
  1736. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1737. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1738. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1739. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1740. let resourceURL = normalizeURL(srcsetValue.url);
  1741. if (!testIgnoredPath(resourceURL)) {
  1742. if (testValidPath(resourceURL)) {
  1743. try {
  1744. resourceURL = util.resolveURL(resourceURL, baseURI);
  1745. } catch (error) {
  1746. // ignored
  1747. }
  1748. if (testValidURL(resourceURL)) {
  1749. const { content } = await batchRequest.addURL(resourceURL, true);
  1750. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1751. if (forbiddenPrefixFound) {
  1752. return "";
  1753. }
  1754. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1755. } else {
  1756. return "";
  1757. }
  1758. } else {
  1759. return "";
  1760. }
  1761. } else {
  1762. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1763. }
  1764. }));
  1765. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1766. }));
  1767. }
  1768. static replaceImageSource(imgElement, variableName, options) {
  1769. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1770. if (attributeValue) {
  1771. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1772. if (imageData && imageData.replaceable) {
  1773. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1774. const backgroundStyle = {};
  1775. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1776. if (backgroundSize) {
  1777. backgroundStyle["background-size"] = imageData.objectFit;
  1778. }
  1779. if (imageData.objectPosition) {
  1780. backgroundStyle["background-position"] = imageData.objectPosition;
  1781. }
  1782. if (imageData.backgroundColor) {
  1783. backgroundStyle["background-color"] = imageData.backgroundColor;
  1784. }
  1785. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1786. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1787. return true;
  1788. }
  1789. }
  1790. }
  1791. }
  1792. // ----
  1793. // Util
  1794. // ----
  1795. const BLOB_URI_PREFIX = "blob:";
  1796. const HTTP_URI_PREFIX = /^https?:\/\//;
  1797. const FILE_URI_PREFIX = /^file:\/\//;
  1798. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1799. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1800. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1801. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1802. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1803. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1804. const REGEXP_IMPORT_FN = /(@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|}))|(@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|}))|(@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|}))|(@import\s*'(.*?)'\s*(.*?)(;|$|}))|(@import\s*"(.*?)"\s*(.*?)(;|$|}))|(@import\s*(.*?)\s*(.*?)(;|$|}))/gi;
  1805. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1806. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1807. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1808. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1809. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1810. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1811. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1812. function getUpdatedResourceContent(resourceURL, content, options) {
  1813. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1814. options.updatedResources[resourceURL].retrieved = true;
  1815. return options.updatedResources[resourceURL].content;
  1816. } else {
  1817. return content.data || "";
  1818. }
  1819. }
  1820. function normalizeURL(url) {
  1821. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1822. return url;
  1823. } else {
  1824. return url.split("#")[0];
  1825. }
  1826. }
  1827. function getCSSValue(value) {
  1828. let result = "";
  1829. try {
  1830. result = cssTree.generate(value);
  1831. } catch (error) {
  1832. // ignored
  1833. }
  1834. return result;
  1835. }
  1836. function getOnEventAttributeNames(doc) {
  1837. const element = doc.createElement("div");
  1838. const attributeNames = [];
  1839. for (const propertyName in element) {
  1840. if (propertyName.startsWith("on")) {
  1841. attributeNames.push(propertyName);
  1842. }
  1843. }
  1844. attributeNames.push("onunload");
  1845. return attributeNames;
  1846. }
  1847. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1848. let maxLength;
  1849. if (template) {
  1850. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1851. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1852. if (template.match(replaceRegExp)) {
  1853. const matchedLength = template.match(replaceRegExp)[0];
  1854. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1855. if (isNaN(maxLength) || maxLength <= 0) {
  1856. maxLength = null;
  1857. }
  1858. } else {
  1859. replaceRegExp = new RegExp(regExpVariable, "g");
  1860. }
  1861. if (template.match(replaceRegExp)) {
  1862. let value = await valueGetter();
  1863. if (!dontReplaceSlash) {
  1864. value = value.replace(/\/+/g, replacementCharacter);
  1865. }
  1866. if (maxLength) {
  1867. value = await util.truncateText(value, maxLength);
  1868. }
  1869. return template.replace(replaceRegExp, value);
  1870. }
  1871. }
  1872. return template;
  1873. }
  1874. function getLastSegment(url, replacementCharacter) {
  1875. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1876. if (!lastSegment) {
  1877. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1878. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1879. }
  1880. if (!lastSegment) {
  1881. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1882. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1883. }
  1884. if (!lastSegment) {
  1885. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1886. }
  1887. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1888. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1889. lastSegment = lastSegmentMatch[1];
  1890. }
  1891. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1892. return lastSegment;
  1893. }
  1894. function getRegExp(string) {
  1895. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1896. }
  1897. function getUrlFunctions(stylesheetContent, unique) {
  1898. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1899. if (unique) {
  1900. return [...new Set(result)];
  1901. } else {
  1902. return result;
  1903. }
  1904. }
  1905. function getImportFunctions(stylesheetContent) {
  1906. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1907. }
  1908. function findShortcutIcon(shortcutIcons) {
  1909. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1910. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1911. return shortcutIcons[0];
  1912. }
  1913. function matchURL(stylesheetContent) {
  1914. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1915. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1916. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1917. return match && match[1];
  1918. }
  1919. function testIgnoredPath(resourceURL) {
  1920. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1921. }
  1922. function testValidPath(resourceURL) {
  1923. return resourceURL && !resourceURL.match(EMPTY_URL);
  1924. }
  1925. function testValidURL(resourceURL) {
  1926. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1927. }
  1928. function matchImport(stylesheetContent) {
  1929. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1930. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1931. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1932. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1933. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1934. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1935. if (match) {
  1936. const [, resourceURL, media] = match;
  1937. return { resourceURL, media };
  1938. }
  1939. }
  1940. function removeCssComments(stylesheetContent) {
  1941. try {
  1942. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  1943. } catch (error) {
  1944. let start, end;
  1945. do {
  1946. start = stylesheetContent.indexOf("/*");
  1947. end = stylesheetContent.indexOf("*/", start + 2);
  1948. if (start != -1 && end != -1) {
  1949. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1950. }
  1951. } while (start != -1 && end != -1);
  1952. return stylesheetContent;
  1953. }
  1954. }
  1955. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  1956. if (mediaQuery) {
  1957. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1958. } else {
  1959. return stylesheetContent;
  1960. }
  1961. }
  1962. function log(...args) {
  1963. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1964. }
  1965. // -----
  1966. // Stats
  1967. // -----
  1968. const STATS_DEFAULT_VALUES = {
  1969. discarded: {
  1970. "HTML bytes": 0,
  1971. "hidden elements": 0,
  1972. "HTML imports": 0,
  1973. scripts: 0,
  1974. objects: 0,
  1975. "audio sources": 0,
  1976. "video sources": 0,
  1977. frames: 0,
  1978. "CSS rules": 0,
  1979. canvas: 0,
  1980. stylesheets: 0,
  1981. resources: 0,
  1982. medias: 0
  1983. },
  1984. processed: {
  1985. "HTML bytes": 0,
  1986. "hidden elements": 0,
  1987. "HTML imports": 0,
  1988. scripts: 0,
  1989. objects: 0,
  1990. "audio sources": 0,
  1991. "video sources": 0,
  1992. frames: 0,
  1993. "CSS rules": 0,
  1994. canvas: 0,
  1995. stylesheets: 0,
  1996. resources: 0,
  1997. medias: 0
  1998. }
  1999. };
  2000. class Stats {
  2001. constructor(options) {
  2002. this.options = options;
  2003. if (options.displayStats) {
  2004. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2005. }
  2006. }
  2007. set(type, subType, value) {
  2008. if (this.options.displayStats) {
  2009. this.data[type][subType] = value;
  2010. }
  2011. }
  2012. add(type, subType, value) {
  2013. if (this.options.displayStats) {
  2014. this.data[type][subType] += value;
  2015. }
  2016. }
  2017. addAll(pageData) {
  2018. if (this.options.displayStats) {
  2019. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2020. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2021. }
  2022. }
  2023. }
  2024. return { getClass };
  2025. })();