single-file-core.js 89 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165
  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;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. }
  34. async run() {
  35. if (this.options.userScriptEnabled) {
  36. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  37. }
  38. this.runner = new Runner(this.options, true);
  39. await this.runner.loadPage();
  40. await this.runner.initialize();
  41. if (this.options.userScriptEnabled) {
  42. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  43. }
  44. await this.runner.run();
  45. }
  46. cancel() {
  47. this.cancelled = true;
  48. if (this.runner) {
  49. this.runner.cancel();
  50. }
  51. }
  52. getPageData() {
  53. return this.runner.getPageData();
  54. }
  55. }
  56. // -------------
  57. // ProgressEvent
  58. // -------------
  59. const PAGE_LOADING = "page-loading";
  60. const PAGE_LOADED = "page-loaded";
  61. const RESOURCES_INITIALIZING = "resource-initializing";
  62. const RESOURCES_INITIALIZED = "resources-initialized";
  63. const RESOURCE_LOADED = "resource-loaded";
  64. const PAGE_ENDED = "page-ended";
  65. const STAGE_STARTED = "stage-started";
  66. const STAGE_ENDED = "stage-ended";
  67. const STAGE_TASK_STARTED = "stage-task-started";
  68. const STAGE_TASK_ENDED = "stage-task-ended";
  69. class ProgressEvent {
  70. constructor(type, detail) {
  71. 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 };
  72. }
  73. }
  74. // ------
  75. // Runner
  76. // ------
  77. const RESOLVE_URLS_STAGE = 0;
  78. const REPLACE_DATA_STAGE = 1;
  79. const REPLACE_DOCS_STAGE = 2;
  80. const POST_PROCESS_STAGE = 3;
  81. const STAGES = [{
  82. sequential: [
  83. { action: "preProcessPage" },
  84. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  85. { action: "replaceStyleContents" },
  86. { action: "resetCharsetMeta" },
  87. { option: "saveFavicon", action: "saveFavicon" },
  88. { action: "replaceCanvasElements" },
  89. { action: "insertFonts" },
  90. { action: "insertShadowRootContents" },
  91. { action: "setInputValues" },
  92. { option: "removeScripts", action: "removeScripts" },
  93. { option: "selected", action: "removeUnselectedElements" },
  94. { option: "removeVideoSrc", action: "insertVideoPosters" },
  95. { option: "removeFrames", action: "removeFrames" },
  96. { option: "removeVideoSrc", action: "removeVideoSources" },
  97. { option: "removeAudioSrc", action: "removeAudioSources" },
  98. { action: "removeDiscardedResources" },
  99. { option: "removeHiddenElements", action: "removeHiddenElements" },
  100. { action: "resolveHrefs" },
  101. { action: "resolveStyleAttributeURLs" }
  102. ],
  103. parallel: [
  104. { action: "resolveStylesheetURLs" },
  105. { option: "!removeFrames", action: "resolveFrameURLs" },
  106. { action: "resolveHtmlImportURLs" }
  107. ]
  108. }, {
  109. sequential: [
  110. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  111. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  112. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  113. ],
  114. parallel: [
  115. { action: "processStylesheets" },
  116. { action: "processStyleAttributes" },
  117. { action: "processPageResources" },
  118. { option: "!removeScripts", action: "processScripts" }
  119. ]
  120. }, {
  121. sequential: [
  122. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  123. ],
  124. parallel: [
  125. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  126. { option: "!removeFrames", action: "processFrames" },
  127. { option: "!removeImports", action: "processHtmlImports" },
  128. ]
  129. }, {
  130. sequential: [
  131. { action: "replaceStylesheets" },
  132. { action: "replaceStyleAttributes" },
  133. { action: "insertVariables" },
  134. { option: "compressHTML", action: "compressHTML" },
  135. { action: "cleanupPage" }
  136. ],
  137. parallel: [
  138. { option: "enableMaff", action: "insertMAFFMetaData" },
  139. { action: "setDocInfo" }
  140. ]
  141. }];
  142. class Runner {
  143. constructor(options, root) {
  144. const rootDocDefined = root && options.doc;
  145. this.root = root;
  146. this.options = options;
  147. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  148. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  149. this.options.rootDocument = root;
  150. this.options.updatedResources = this.options.updatedResources || {};
  151. this.options.fontTests = new Map();
  152. this.batchRequest = new BatchRequest();
  153. this.processor = new Processor(options, this.batchRequest);
  154. if (rootDocDefined) {
  155. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  156. this.options.canvases = docData.canvases;
  157. this.options.fonts = docData.fonts;
  158. this.options.stylesheets = docData.stylesheets;
  159. this.options.images = docData.images;
  160. this.options.posters = docData.posters;
  161. this.options.usedFonts = docData.usedFonts;
  162. this.options.shadowRoots = docData.shadowRoots;
  163. this.options.imports = docData.imports;
  164. this.options.referrer = docData.referrer;
  165. this.markedElements = docData.markedElements;
  166. }
  167. if (this.options.saveRawPage) {
  168. this.options.removeFrames = true;
  169. }
  170. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  171. this.onprogress = options.onprogress || (() => { });
  172. }
  173. async loadPage() {
  174. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  175. await this.processor.loadPage(this.options.content);
  176. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  177. }
  178. async initialize() {
  179. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  180. await this.executeStage(RESOLVE_URLS_STAGE);
  181. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  182. if (this.root && this.options.doc) {
  183. util.postProcessDoc(this.options.doc, this.markedElements);
  184. }
  185. }
  186. cancel() {
  187. this.cancelled = true;
  188. this.batchRequest.cancel();
  189. if (this.root) {
  190. if (this.options.frames) {
  191. this.options.frames.forEach(cancelRunner);
  192. }
  193. if (this.options.imports) {
  194. this.options.imports.forEach(cancelRunner);
  195. }
  196. }
  197. function cancelRunner(resourceData) {
  198. if (resourceData.runner) {
  199. resourceData.runner.cancel();
  200. }
  201. }
  202. }
  203. async run() {
  204. if (this.root) {
  205. this.processor.initialize(this.batchRequest);
  206. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  207. }
  208. await this.batchRequest.run(detail => {
  209. detail.pageURL = this.options.url;
  210. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  211. }, this.options);
  212. await this.pendingPromises;
  213. this.options.doc = null;
  214. this.options.win = null;
  215. await this.executeStage(REPLACE_DOCS_STAGE);
  216. await this.executeStage(POST_PROCESS_STAGE);
  217. this.processor.finalize();
  218. }
  219. getDocument() {
  220. return this.processor.doc;
  221. }
  222. getStyleSheets() {
  223. return this.processor.stylesheets;
  224. }
  225. getPageData() {
  226. if (this.root) {
  227. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  228. }
  229. return this.processor.getPageData();
  230. }
  231. async executeStage(step) {
  232. if (DEBUG) {
  233. log("**** STARTED STAGE", step, "****");
  234. }
  235. const frame = !this.root;
  236. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  237. STAGES[step].sequential.forEach(task => {
  238. let startTime;
  239. if (DEBUG) {
  240. startTime = Date.now();
  241. log(" -- STARTED task =", task.action);
  242. }
  243. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  244. if (!this.cancelled) {
  245. this.executeTask(task);
  246. }
  247. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  248. if (DEBUG) {
  249. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  250. }
  251. });
  252. let parallelTasksPromise;
  253. if (STAGES[step].parallel) {
  254. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  255. let startTime;
  256. if (DEBUG) {
  257. startTime = Date.now();
  258. log(" // STARTED task =", task.action);
  259. }
  260. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  261. if (!this.cancelled) {
  262. await this.executeTask(task);
  263. }
  264. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  265. if (DEBUG) {
  266. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  267. }
  268. }));
  269. } else {
  270. parallelTasksPromise = Promise.resolve();
  271. }
  272. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  273. if (DEBUG) {
  274. log("**** ENDED STAGE", step, "****");
  275. }
  276. return parallelTasksPromise;
  277. }
  278. executeTask(task) {
  279. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  280. return this.processor[task.action]();
  281. }
  282. }
  283. }
  284. // ------------
  285. // BatchRequest
  286. // ------------
  287. class BatchRequest {
  288. constructor() {
  289. this.requests = new Map();
  290. this.duplicates = new Map();
  291. }
  292. addURL(resourceURL, asBinary, groupDuplicates) {
  293. return new Promise((resolve, reject) => {
  294. const requestKey = JSON.stringify([resourceURL, asBinary]);
  295. let resourceRequests = this.requests.get(requestKey);
  296. if (!resourceRequests) {
  297. resourceRequests = [];
  298. this.requests.set(requestKey, resourceRequests);
  299. }
  300. const callbacks = { resolve, reject };
  301. resourceRequests.push(callbacks);
  302. if (groupDuplicates) {
  303. let duplicateRequests = this.duplicates.get(requestKey);
  304. if (!duplicateRequests) {
  305. duplicateRequests = [];
  306. this.duplicates.set(requestKey, duplicateRequests);
  307. }
  308. duplicateRequests.push(callbacks);
  309. }
  310. });
  311. }
  312. getMaxResources() {
  313. return this.requests.size;
  314. }
  315. run(onloadListener, options) {
  316. const resourceURLs = [...this.requests.keys()];
  317. let indexResource = 0;
  318. return Promise.all(resourceURLs.map(async requestKey => {
  319. const [resourceURL, asBinary] = JSON.parse(requestKey);
  320. const resourceRequests = this.requests.get(requestKey);
  321. try {
  322. const currentIndexResource = indexResource;
  323. indexResource = indexResource + 1;
  324. const content = await util.getContent(resourceURL, {
  325. asBinary,
  326. maxResourceSize: options.maxResourceSize,
  327. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  328. frameId: options.windowId
  329. });
  330. onloadListener({ url: resourceURL });
  331. if (!this.cancelled) {
  332. resourceRequests.forEach(callbacks => {
  333. const duplicateCallbacks = this.duplicates.get(requestKey);
  334. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  335. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  336. });
  337. }
  338. } catch (error) {
  339. indexResource = indexResource + 1;
  340. onloadListener({ url: resourceURL });
  341. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  342. }
  343. this.requests.delete(requestKey);
  344. }));
  345. }
  346. cancel() {
  347. this.cancelled = true;
  348. const resourceURLs = [...this.requests.keys()];
  349. resourceURLs.forEach(requestKey => {
  350. const resourceRequests = this.requests.get(requestKey);
  351. resourceRequests.forEach(callbacks => callbacks.reject());
  352. this.requests.delete(requestKey);
  353. });
  354. }
  355. }
  356. // ---------
  357. // Processor
  358. // ---------
  359. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  360. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  361. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  362. const SCRIPT_TAG_FOUND = /<script/gi;
  363. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  364. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  365. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  366. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  367. const UTF8_CHARSET = "utf-8";
  368. class Processor {
  369. constructor(options, batchRequest) {
  370. this.options = options;
  371. this.stats = new Stats(options);
  372. this.baseURI = normalizeURL(options.baseURI || options.url);
  373. this.batchRequest = batchRequest;
  374. this.stylesheets = new Map();
  375. this.styles = new Map();
  376. this.cssVariables = new Map();
  377. this.fontTests = options.fontTests;
  378. }
  379. initialize() {
  380. this.options.saveDate = new Date();
  381. this.options.saveUrl = this.options.url;
  382. if (this.options.enableMaff) {
  383. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  384. }
  385. this.maxResources = this.batchRequest.getMaxResources();
  386. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  387. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  388. }
  389. if (!this.options.removeImports && this.options.imports) {
  390. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  391. }
  392. this.stats.set("processed", "resources", this.maxResources);
  393. }
  394. async loadPage(pageContent, charset) {
  395. let content;
  396. if (!pageContent || this.options.saveRawPage) {
  397. content = await util.getContent(this.baseURI, {
  398. maxResourceSize: this.options.maxResourceSize,
  399. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  400. charset,
  401. frameId: this.options.windowId
  402. });
  403. pageContent = content.data;
  404. }
  405. this.doc = util.parseDocContent(pageContent, this.baseURI);
  406. if (this.options.saveRawPage) {
  407. let charset;
  408. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  409. const charsetDeclaration = element.content.split(";")[1];
  410. if (charsetDeclaration && !charset) {
  411. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  412. }
  413. });
  414. if (charset && charset != content.charset) {
  415. return this.loadPage(pageContent, charset);
  416. }
  417. }
  418. this.workStyleElement = this.doc.createElement("style");
  419. this.doc.body.appendChild(this.workStyleElement);
  420. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  421. }
  422. finalize() {
  423. if (this.workStyleElement.parentNode) {
  424. this.workStyleElement.remove();
  425. }
  426. }
  427. async getPageData() {
  428. util.postProcessDoc(this.doc);
  429. const url = util.parseURL(this.baseURI);
  430. if (this.options.insertSingleFileComment) {
  431. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  432. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  433. " \n url: " + this.options.saveUrl +
  434. " \n saved date: " + this.options.saveDate +
  435. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  436. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  437. }
  438. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  439. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  440. if (!canonicalLink) {
  441. canonicalLink = this.doc.createElement("link");
  442. canonicalLink.setAttribute("rel", "canonical");
  443. this.doc.head.appendChild(canonicalLink);
  444. }
  445. if (canonicalLink && !canonicalLink.href) {
  446. canonicalLink.href = this.options.saveUrl;
  447. }
  448. }
  449. if (this.options.insertMetaNoIndex) {
  450. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  451. if (!metaElement) {
  452. metaElement = this.doc.createElement("meta");
  453. metaElement.setAttribute("name", "robots");
  454. metaElement.setAttribute("content", "noindex");
  455. this.doc.head.appendChild(metaElement);
  456. }
  457. }
  458. let size;
  459. if (this.options.displayStats) {
  460. size = util.getContentSize(this.doc.documentElement.outerHTML);
  461. }
  462. const content = util.serialize(this.doc, this.options.compressHTML);
  463. if (this.options.displayStats) {
  464. const contentSize = util.getContentSize(content);
  465. this.stats.set("processed", "HTML bytes", contentSize);
  466. this.stats.add("discarded", "HTML bytes", size - contentSize);
  467. }
  468. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  469. const replacementCharacter = this.options.filenameReplacementCharacter;
  470. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  471. if (!this.options.backgroundSave) {
  472. filename = filename.replace(/\//g, replacementCharacter);
  473. }
  474. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  475. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  476. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  477. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  478. filename = filename + "…" + extension;
  479. }
  480. if (!filename) {
  481. filename = "Unnamed page";
  482. }
  483. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  484. const pageData = {
  485. stats: this.stats.data,
  486. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  487. filename,
  488. content
  489. };
  490. if (this.options.addProof) {
  491. pageData.hash = await util.digest("SHA-256", content);
  492. }
  493. if (this.options.retrieveLinks) {
  494. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  495. }
  496. return pageData;
  497. }
  498. preProcessPage() {
  499. if (this.options.win) {
  500. 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));
  501. }
  502. if (this.options.images && !this.options.saveRawPage) {
  503. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  504. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  505. if (attributeValue) {
  506. const imageData = this.options.images[Number(attributeValue)];
  507. if (imageData) {
  508. if (this.options.removeHiddenElements && (
  509. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  510. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  511. )) {
  512. imgElement.setAttribute("src", EMPTY_IMAGE);
  513. } else {
  514. if (imageData.currentSrc) {
  515. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  516. imgElement.setAttribute("src", imageData.currentSrc);
  517. }
  518. if (this.options.loadDeferredImages) {
  519. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  520. imageData.src = imgElement.dataset.src;
  521. imgElement.setAttribute("src", imgElement.dataset.src);
  522. imgElement.removeAttribute("data-src");
  523. }
  524. }
  525. }
  526. }
  527. }
  528. });
  529. if (this.options.loadDeferredImages) {
  530. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  531. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  532. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  533. imgElement.removeAttribute("data-srcset");
  534. }
  535. });
  536. }
  537. }
  538. }
  539. replaceStyleContents() {
  540. if (this.options.stylesheets) {
  541. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  542. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  543. if (attributeValue) {
  544. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  545. if (stylesheetContent) {
  546. styleElement.textContent = stylesheetContent;
  547. }
  548. }
  549. });
  550. }
  551. }
  552. removeUnselectedElements() {
  553. removeUnmarkedElements(this.doc.body);
  554. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  555. function removeUnmarkedElements(element) {
  556. let selectedElementFound = false;
  557. Array.from(element.childNodes).forEach(node => {
  558. if (node.nodeType == 1) {
  559. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  560. selectedElementFound = selectedElementFound || isSelectedElement;
  561. if (isSelectedElement) {
  562. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  563. removeUnmarkedElements(node);
  564. } else if (selectedElementFound) {
  565. removeNode(node);
  566. } else {
  567. hideNode(node);
  568. }
  569. }
  570. });
  571. }
  572. function removeNode(node) {
  573. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  574. node.remove();
  575. } else {
  576. hideNode(node);
  577. }
  578. }
  579. function hideNode(node) {
  580. if (canHideNode(node)) {
  581. node.style.setProperty("display", "none", "important");
  582. Array.from(node.childNodes).forEach(removeNode);
  583. }
  584. }
  585. function canHideNode(node) {
  586. if (node.nodeType == 1) {
  587. const tagName = node.tagName && node.tagName.toLowerCase();
  588. return (tagName != "svg" && tagName != "style" && tagName != "link");
  589. }
  590. }
  591. }
  592. insertVideoPosters() {
  593. if (this.options.posters) {
  594. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  595. let videoElement;
  596. if (element.tagName == "VIDEO") {
  597. videoElement = element;
  598. } else {
  599. videoElement = element.parentElement;
  600. }
  601. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  602. if (attributeValue) {
  603. const posterURL = this.options.posters[Number(attributeValue)];
  604. if (!videoElement.poster && posterURL) {
  605. videoElement.setAttribute("poster", posterURL);
  606. }
  607. }
  608. });
  609. }
  610. }
  611. removeFrames() {
  612. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  613. this.stats.set("discarded", "frames", frameElements.length);
  614. this.stats.set("processed", "frames", frameElements.length);
  615. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  616. }
  617. removeScripts() {
  618. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  619. this.doc.querySelectorAll("[href]").forEach(element => {
  620. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  621. element.setAttribute("href", "");
  622. }
  623. });
  624. this.doc.querySelectorAll("[src]").forEach(element => {
  625. if (element.src && element.src.match(/^\s*javascript:/)) {
  626. element.removeAttribute("src");
  627. }
  628. });
  629. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  630. this.stats.set("discarded", "scripts", scriptElements.length);
  631. this.stats.set("processed", "scripts", scriptElements.length);
  632. scriptElements.forEach(element => element.remove());
  633. }
  634. removeVideoSources() {
  635. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  636. this.stats.set("discarded", "video sources", videoSourceElements.length);
  637. this.stats.set("processed", "video sources", videoSourceElements.length);
  638. videoSourceElements.forEach(element => {
  639. if (element.tagName == "SOURCE") {
  640. element.remove();
  641. } else {
  642. videoSourceElements.forEach(element => element.removeAttribute("src"));
  643. }
  644. });
  645. }
  646. removeAudioSources() {
  647. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  648. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  649. this.stats.set("processed", "audio sources", audioSourceElements.length);
  650. audioSourceElements.forEach(element => {
  651. if (element.tagName == "SOURCE") {
  652. element.remove();
  653. } else {
  654. audioSourceElements.forEach(element => element.removeAttribute("src"));
  655. }
  656. });
  657. }
  658. removeDiscardedResources() {
  659. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  660. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  661. 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\"])");
  662. this.stats.set("discarded", "objects", objectElements.length);
  663. this.stats.set("processed", "objects", objectElements.length);
  664. objectElements.forEach(element => element.remove());
  665. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  666. replacedAttributeValue.forEach(element => {
  667. let regExp;
  668. if (this.options.removeScripts) {
  669. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  670. } else {
  671. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  672. }
  673. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  674. if (relValue.length) {
  675. element.setAttribute("rel", relValue);
  676. } else {
  677. element.remove();
  678. }
  679. });
  680. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  681. if (this.options.removeHiddenElements) {
  682. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  683. }
  684. if (!this.options.saveFavicon) {
  685. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  686. }
  687. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  688. }
  689. resetCharsetMeta() {
  690. let charset;
  691. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  692. const charsetDeclaration = element.content.split(";")[1];
  693. if (charsetDeclaration && !charset) {
  694. charset = charsetDeclaration.split("=")[1];
  695. if (charset) {
  696. this.charset = charset.trim().toLowerCase();
  697. }
  698. }
  699. element.remove();
  700. });
  701. const metaElement = this.doc.createElement("meta");
  702. metaElement.setAttribute("charset", UTF8_CHARSET);
  703. if (this.doc.head.firstChild) {
  704. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  705. } else {
  706. this.doc.head.appendChild(metaElement);
  707. }
  708. }
  709. setInputValues() {
  710. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  711. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  712. input.setAttribute("value", value || "");
  713. });
  714. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  715. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  716. if (value == "true") {
  717. input.setAttribute("checked", "");
  718. }
  719. });
  720. this.doc.querySelectorAll("textarea").forEach(textarea => {
  721. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  722. textarea.textContent = value || "";
  723. });
  724. this.doc.querySelectorAll("select").forEach(select => {
  725. select.querySelectorAll("option").forEach(option => {
  726. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  727. if (selected) {
  728. option.setAttribute("selected", "");
  729. }
  730. });
  731. });
  732. }
  733. saveFavicon() {
  734. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  735. if (!faviconElement) {
  736. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  737. }
  738. if (!faviconElement) {
  739. faviconElement = this.doc.createElement("link");
  740. faviconElement.setAttribute("type", "image/x-icon");
  741. faviconElement.setAttribute("rel", "shortcut icon");
  742. faviconElement.setAttribute("href", "/favicon.ico");
  743. }
  744. this.doc.head.appendChild(faviconElement);
  745. }
  746. replaceCanvasElements() {
  747. if (this.options.canvases) {
  748. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  749. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  750. if (attributeValue) {
  751. const canvasData = this.options.canvases[Number(attributeValue)];
  752. if (canvasData) {
  753. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  754. this.stats.add("processed", "canvas", 1);
  755. }
  756. }
  757. });
  758. }
  759. }
  760. insertFonts() {
  761. if (this.options.fonts && this.options.fonts.length) {
  762. let stylesheetContent = "";
  763. this.options.fonts.forEach(fontData => {
  764. if (fontData["font-family"] && fontData.src) {
  765. stylesheetContent += "@font-face{";
  766. let stylesContent = "";
  767. Object.keys(fontData).forEach(fontStyle => {
  768. if (stylesContent) {
  769. stylesContent += ";";
  770. }
  771. stylesContent += fontStyle + ":" + fontData[fontStyle];
  772. });
  773. stylesheetContent += stylesContent + "}";
  774. }
  775. });
  776. if (stylesheetContent) {
  777. const styleElement = this.doc.createElement("style");
  778. styleElement.textContent = stylesheetContent;
  779. const existingStyleElement = this.doc.querySelector("style");
  780. if (existingStyleElement) {
  781. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  782. } else {
  783. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  784. }
  785. }
  786. }
  787. }
  788. removeHiddenElements() {
  789. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  790. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  791. this.stats.set("discarded", "hidden elements", removedElements.length);
  792. this.stats.set("processed", "hidden elements", removedElements.length);
  793. if (hiddenElements.length) {
  794. const styleElement = this.doc.createElement("style");
  795. styleElement.textContent = ".sf-hidden{display:none!important;}";
  796. this.doc.head.appendChild(styleElement);
  797. hiddenElements.forEach(element => {
  798. if (element.style.getPropertyValue("display") != "none") {
  799. if (element.style.getPropertyPriority("display") == "important") {
  800. element.style.setProperty("display", "none", "important");
  801. } else {
  802. element.classList.add("sf-hidden");
  803. }
  804. }
  805. });
  806. }
  807. removedElements.forEach(element => element.remove());
  808. }
  809. resolveHrefs() {
  810. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  811. const href = element.getAttribute("href").trim();
  812. if (!testIgnoredPath(href)) {
  813. let resolvedURL;
  814. try {
  815. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  816. } catch (error) {
  817. // ignored
  818. }
  819. if (resolvedURL) {
  820. const url = normalizeURL(this.options.url);
  821. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  822. resolvedURL = resolvedURL.substring(url.length);
  823. }
  824. try {
  825. element.setAttribute("href", resolvedURL);
  826. } catch (error) {
  827. // ignored
  828. }
  829. }
  830. }
  831. });
  832. }
  833. resolveStyleAttributeURLs() {
  834. this.doc.querySelectorAll("[style]").forEach(element => {
  835. let styleContent = element.getAttribute("style");
  836. if (this.options.compressCSS) {
  837. styleContent = util.compressCSS(styleContent);
  838. }
  839. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  840. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  841. this.styles.set(element, declarationList);
  842. });
  843. }
  844. async resolveStylesheetURLs() {
  845. const options = {
  846. maxResourceSize: this.options.maxResourceSize,
  847. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  848. url: this.options.url,
  849. charset: this.charset,
  850. compressCSS: this.options.compressCSS,
  851. updatedResources: this.options.updatedResources,
  852. rootDocument: this.options.rootDocument,
  853. frameId: this.options.windowId
  854. };
  855. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  856. .map(async element => {
  857. let mediaText;
  858. if (element.media) {
  859. mediaText = element.media.toLowerCase();
  860. }
  861. const stylesheetInfo = { mediaText };
  862. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  863. stylesheetInfo.scoped = true;
  864. }
  865. if (element.tagName.toLowerCase() == "link") {
  866. if (element.charset) {
  867. options.charset = element.charset;
  868. }
  869. }
  870. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  871. }));
  872. if (options.rootDocument) {
  873. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  874. await Promise.all(newResources.map(async resource => {
  875. resource.retrieved = true;
  876. const stylesheetInfo = {};
  877. const element = this.doc.createElement("style");
  878. this.doc.body.appendChild(element);
  879. element.textContent = resource.content;
  880. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  881. }));
  882. }
  883. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  884. stylesheets.set(element, stylesheetInfo);
  885. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  886. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  887. if (match && match[1] && match[1] != options.charset) {
  888. options.charset = match[1];
  889. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  890. }
  891. let stylesheet;
  892. try {
  893. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  894. } catch (error) {
  895. // ignored
  896. }
  897. if (stylesheet && stylesheet.children) {
  898. if (options.compressCSS) {
  899. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  900. }
  901. stylesheetInfo.stylesheet = stylesheet;
  902. } else {
  903. stylesheets.delete(element);
  904. }
  905. }
  906. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  907. let content;
  908. if (element.tagName.toLowerCase() == "link") {
  909. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  910. } else {
  911. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  912. }
  913. return content || "";
  914. }
  915. }
  916. async resolveFrameURLs() {
  917. if (!this.options.saveRawPage) {
  918. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  919. await Promise.all(frameElements.map(async frameElement => {
  920. if (frameElement.tagName == "OBJECT") {
  921. frameElement.setAttribute("data", "data:text/html,");
  922. } else {
  923. frameElement.removeAttribute("src");
  924. frameElement.removeAttribute("srcdoc");
  925. }
  926. Array.from(frameElement.childNodes).forEach(node => node.remove());
  927. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  928. if (this.options.frames && frameWindowId) {
  929. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  930. if (frameData) {
  931. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  932. }
  933. }
  934. }));
  935. }
  936. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  937. options.insertSingleFileComment = false;
  938. options.insertCanonicalLink = false;
  939. options.insertMetaNoIndex = false;
  940. options.saveFavicon = false;
  941. options.url = frameData.baseURI;
  942. options.windowId = frameWindowId;
  943. if (frameData.content) {
  944. options.content = frameData.content;
  945. options.canvases = frameData.canvases;
  946. options.fonts = frameData.fonts;
  947. options.stylesheets = frameData.stylesheets;
  948. options.images = frameData.images;
  949. options.posters = frameData.posters;
  950. options.usedFonts = frameData.usedFonts;
  951. options.shadowRoots = frameData.shadowRoots;
  952. options.imports = frameData.imports;
  953. frameData.runner = new Runner(options);
  954. frameData.frameElement = frameElement;
  955. await frameData.runner.loadPage();
  956. await frameData.runner.initialize();
  957. frameData.maxResources = batchRequest.getMaxResources();
  958. }
  959. }
  960. }
  961. insertShadowRootContents() {
  962. const doc = this.doc;
  963. const options = this.options;
  964. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  965. processElement(this.doc);
  966. if (this.options.removeScripts) {
  967. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  968. }
  969. const scriptElement = doc.createElement("script");
  970. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  971. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{let shadowRoot = element.parentElement.shadowRoot;if (!shadowRoot) {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)}})}})()`;
  972. doc.body.appendChild(scriptElement);
  973. }
  974. function processElement(element) {
  975. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  976. shadowRootElements.forEach(element => {
  977. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  978. if (attributeValue) {
  979. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  980. if (shadowRootData) {
  981. const templateElement = doc.createElement("template");
  982. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  983. if (shadowRootData.delegatesFocus) {
  984. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  985. }
  986. if (shadowRootData.adoptedStyleSheets) {
  987. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  988. const styleElement = doc.createElement("style");
  989. styleElement.textContent = stylesheetContent;
  990. templateElement.appendChild(styleElement);
  991. });
  992. }
  993. const shadowDoc = util.parseDocContent(shadowRootData.content);
  994. if (shadowDoc.head) {
  995. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  996. if (metaCharset) {
  997. metaCharset.remove();
  998. }
  999. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1000. }
  1001. if (shadowDoc.body) {
  1002. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1003. }
  1004. processElement(templateElement);
  1005. if (element.firstChild) {
  1006. element.insertBefore(templateElement, element.firstChild);
  1007. } else {
  1008. element.appendChild(templateElement);
  1009. }
  1010. }
  1011. }
  1012. });
  1013. }
  1014. }
  1015. async resolveHtmlImportURLs() {
  1016. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1017. await Promise.all(linkElements.map(async linkElement => {
  1018. const resourceURL = linkElement.href;
  1019. linkElement.removeAttribute("href");
  1020. const options = Object.create(this.options);
  1021. options.insertSingleFileComment = false;
  1022. options.insertCanonicalLink = false;
  1023. options.insertMetaNoIndex = false;
  1024. options.saveFavicon = false;
  1025. options.removeUnusedStyles = false;
  1026. options.removeAlternativeMedias = false;
  1027. options.removeUnusedFonts = false;
  1028. options.removeHiddenElements = false;
  1029. options.url = resourceURL;
  1030. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1031. if (attributeValue) {
  1032. const importData = options.imports[Number(attributeValue)];
  1033. if (importData) {
  1034. options.content = importData.content;
  1035. importData.runner = new Runner(options);
  1036. await importData.runner.loadPage();
  1037. await importData.runner.initialize();
  1038. if (!options.removeImports) {
  1039. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1040. }
  1041. importData.runner.getStyleSheets().forEach(stylesheet => {
  1042. const importedStyleElement = this.doc.createElement("style");
  1043. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1044. this.stylesheets.set(importedStyleElement, stylesheet);
  1045. });
  1046. }
  1047. }
  1048. if (options.removeImports) {
  1049. linkElement.remove();
  1050. this.stats.add("discarded", "HTML imports", 1);
  1051. }
  1052. }));
  1053. }
  1054. removeUnusedStyles() {
  1055. if (!this.mediaAllInfo) {
  1056. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1057. }
  1058. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1059. this.stats.set("processed", "CSS rules", stats.processed);
  1060. this.stats.set("discarded", "CSS rules", stats.discarded);
  1061. }
  1062. removeUnusedFonts() {
  1063. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1064. }
  1065. removeAlternativeMedias() {
  1066. const stats = util.minifyMedias(this.stylesheets);
  1067. this.stats.set("processed", "medias", stats.processed);
  1068. this.stats.set("discarded", "medias", stats.discarded);
  1069. }
  1070. async processStylesheets() {
  1071. this.options.fontURLs = new Map();
  1072. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1073. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1074. ));
  1075. }
  1076. async processStyleAttributes() {
  1077. return Promise.all([...this.styles].map(([, declarationList]) =>
  1078. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1079. ));
  1080. }
  1081. async processPageResources() {
  1082. const processAttributeArgs = [
  1083. ["link[href][rel*=\"icon\"]", "href", false, true],
  1084. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1085. ["img[src], input[src][type=image]", "src", true],
  1086. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1087. ["video[poster]", "poster"],
  1088. ["*[background]", "background"],
  1089. ["image", "xlink:href"],
  1090. ["image", "href"]
  1091. ];
  1092. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1093. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1094. );
  1095. resourcePromises = resourcePromises.concat([
  1096. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1097. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1098. ]);
  1099. if (!this.options.removeAudioSrc) {
  1100. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1101. }
  1102. if (!this.options.removeVideoSrc) {
  1103. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1104. }
  1105. await Promise.all(resourcePromises);
  1106. if (this.options.saveFavicon) {
  1107. ProcessorHelper.processShortcutIcons(this.doc);
  1108. }
  1109. }
  1110. async processScripts() {
  1111. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1112. let resourceURL;
  1113. let scriptSrc;
  1114. if (element.tagName == "SCRIPT") {
  1115. scriptSrc = element.getAttribute("src");
  1116. element.removeAttribute("src");
  1117. } else {
  1118. scriptSrc = element.getAttribute("href");
  1119. element.removeAttribute("href");
  1120. }
  1121. element.removeAttribute("integrity");
  1122. element.textContent = "";
  1123. try {
  1124. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1125. } catch (error) {
  1126. // ignored
  1127. }
  1128. if (testValidURL(resourceURL)) {
  1129. this.stats.add("processed", "scripts", 1);
  1130. const content = await util.getContent(resourceURL, {
  1131. asBinary: true,
  1132. charset: this.charset != UTF8_CHARSET && this.charset,
  1133. maxResourceSize: this.options.maxResourceSize,
  1134. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1135. frameId: this.options.windowId
  1136. });
  1137. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1138. if (element.tagName == "SCRIPT") {
  1139. element.setAttribute("src", content.data);
  1140. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1141. element.setAttribute("async", "");
  1142. }
  1143. } else {
  1144. const scriptElement = this.doc.createElement("script");
  1145. scriptElement.setAttribute("src", content.data);
  1146. scriptElement.setAttribute("async", "");
  1147. element.parentElement.replaceChild(scriptElement, element);
  1148. }
  1149. }
  1150. }));
  1151. }
  1152. removeAlternativeImages() {
  1153. util.removeAlternativeImages(this.doc);
  1154. }
  1155. async removeAlternativeFonts() {
  1156. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1157. }
  1158. async processFrames() {
  1159. if (this.options.frames) {
  1160. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1161. await Promise.all(frameElements.map(async frameElement => {
  1162. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1163. if (frameWindowId) {
  1164. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1165. if (frameData) {
  1166. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1167. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1168. this.stats.add("processed", "frames", 1);
  1169. await frameData.runner.run();
  1170. const pageData = await frameData.runner.getPageData();
  1171. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1172. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1173. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1174. sandbox += " allow-scripts allow-same-origin";
  1175. }
  1176. frameElement.setAttribute("sandbox", sandbox);
  1177. if (frameElement.tagName == "OBJECT") {
  1178. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1179. } else {
  1180. if (frameElement.tagName == "FRAME") {
  1181. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1182. } else {
  1183. frameElement.setAttribute("srcdoc", pageData.content);
  1184. frameElement.removeAttribute("src");
  1185. }
  1186. }
  1187. this.stats.addAll(pageData);
  1188. } else {
  1189. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1190. this.stats.add("discarded", "frames", 1);
  1191. }
  1192. }
  1193. }
  1194. }));
  1195. }
  1196. }
  1197. async processHtmlImports() {
  1198. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1199. await Promise.all(linkElements.map(async linkElement => {
  1200. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1201. if (attributeValue) {
  1202. const importData = this.options.imports[Number(attributeValue)];
  1203. if (importData.runner) {
  1204. this.stats.add("processed", "HTML imports", 1);
  1205. await importData.runner.run();
  1206. const pageData = await importData.runner.getPageData();
  1207. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1208. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1209. this.stats.addAll(pageData);
  1210. } else {
  1211. this.stats.add("discarded", "HTML imports", 1);
  1212. }
  1213. }
  1214. }));
  1215. }
  1216. replaceStylesheets() {
  1217. this.doc.querySelectorAll("style").forEach(styleElement => {
  1218. const stylesheetInfo = this.stylesheets.get(styleElement);
  1219. if (stylesheetInfo) {
  1220. this.stylesheets.delete(styleElement);
  1221. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1222. styleElement.textContent = stylesheetContent;
  1223. if (stylesheetInfo.mediaText) {
  1224. styleElement.media = stylesheetInfo.mediaText;
  1225. }
  1226. } else {
  1227. styleElement.remove();
  1228. }
  1229. });
  1230. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1231. const stylesheetInfo = this.stylesheets.get(linkElement);
  1232. if (stylesheetInfo) {
  1233. this.stylesheets.delete(linkElement);
  1234. const styleElement = this.doc.createElement("style");
  1235. if (stylesheetInfo.mediaText) {
  1236. styleElement.media = stylesheetInfo.mediaText;
  1237. }
  1238. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1239. styleElement.textContent = stylesheetContent;
  1240. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1241. } else {
  1242. linkElement.remove();
  1243. }
  1244. });
  1245. }
  1246. replaceStyleAttributes() {
  1247. this.doc.querySelectorAll("[style]").forEach(element => {
  1248. const declarations = this.styles.get(element);
  1249. if (declarations) {
  1250. this.styles.delete(element);
  1251. let styleContent = cssTree.generate(declarations);
  1252. element.setAttribute("style", styleContent);
  1253. } else {
  1254. element.setAttribute("style", "");
  1255. }
  1256. });
  1257. }
  1258. insertVariables() {
  1259. if (this.cssVariables.size) {
  1260. const styleElement = this.doc.createElement("style");
  1261. const firstStyleElement = this.doc.head.querySelector("style");
  1262. if (firstStyleElement) {
  1263. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1264. } else {
  1265. this.doc.head.appendChild(styleElement);
  1266. }
  1267. let stylesheetContent = "";
  1268. this.cssVariables.forEach((content, indexResource) => {
  1269. this.cssVariables.delete(indexResource);
  1270. if (stylesheetContent) {
  1271. stylesheetContent += ";";
  1272. }
  1273. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1274. });
  1275. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1276. }
  1277. }
  1278. compressHTML() {
  1279. let size;
  1280. if (this.options.displayStats) {
  1281. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1282. }
  1283. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1284. if (this.options.displayStats) {
  1285. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1286. }
  1287. }
  1288. cleanupPage() {
  1289. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1290. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1291. if (metaCharset) {
  1292. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1293. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1294. this.doc.head.querySelector("meta[charset]").remove();
  1295. }
  1296. }
  1297. }
  1298. resetZoomLevel() {
  1299. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1300. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1301. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1302. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1303. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1304. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1305. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1306. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1307. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1308. this.doc.documentElement.style.removeProperty("-sf-transform");
  1309. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1310. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1311. }
  1312. async insertMAFFMetaData() {
  1313. const maffMetaData = await this.maffMetaDataPromise;
  1314. if (maffMetaData && maffMetaData.content) {
  1315. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1316. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1317. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1318. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1319. if (originalURLElement) {
  1320. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1321. }
  1322. if (archiveTimeElement) {
  1323. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1324. if (value) {
  1325. const date = new Date(value);
  1326. if (!isNaN(date.getTime())) {
  1327. this.options.saveDate = new Date(value);
  1328. }
  1329. }
  1330. }
  1331. }
  1332. }
  1333. async setDocInfo() {
  1334. const titleElement = this.doc.querySelector("title");
  1335. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1336. const authorElement = this.doc.querySelector("meta[name=author]");
  1337. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1338. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1339. const headingElement = this.doc.querySelector("h1");
  1340. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1341. this.options.info = {
  1342. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1343. lang: this.doc.documentElement.lang,
  1344. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1345. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1346. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1347. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1348. };
  1349. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1350. }
  1351. }
  1352. // ---------------
  1353. // ProcessorHelper
  1354. // ---------------
  1355. const DATA_URI_PREFIX = "data:";
  1356. const ABOUT_BLANK_URI = "about:blank";
  1357. const EMPTY_DATA_URI = "data:null;base64,";
  1358. const REGEXP_URL_HASH = /(#.+?)$/;
  1359. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1360. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1361. class ProcessorHelper {
  1362. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1363. const date = options.saveDate;
  1364. const url = util.parseURL(options.saveUrl);
  1365. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1366. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1367. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1368. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1370. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1371. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1376. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1377. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1387. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1388. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1397. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1398. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1401. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1402. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1403. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1404. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1405. for (const [name, value] of params) {
  1406. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1407. }
  1408. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1409. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1410. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1411. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1412. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1413. if (content) {
  1414. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1415. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1416. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1417. }
  1418. return template.trim();
  1419. function decode(value) {
  1420. try {
  1421. return decodeURI(value);
  1422. } catch (error) {
  1423. return value;
  1424. }
  1425. }
  1426. }
  1427. static setBackgroundImage(element, url, style) {
  1428. element.style.setProperty("background-blend-mode", "normal", "important");
  1429. element.style.setProperty("background-clip", "content-box", "important");
  1430. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1431. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1432. element.style.setProperty("background-image", url, "important");
  1433. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1434. element.style.setProperty("background-origin", "content-box", "important");
  1435. element.style.setProperty("background-repeat", "no-repeat", "important");
  1436. }
  1437. static processShortcutIcons(doc) {
  1438. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1439. if (!shortcutIcon) {
  1440. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1441. if (shortcutIcon) {
  1442. shortcutIcon.rel = "icon";
  1443. }
  1444. }
  1445. if (shortcutIcon) {
  1446. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1447. if (linkElement != shortcutIcon) {
  1448. linkElement.remove();
  1449. }
  1450. });
  1451. }
  1452. }
  1453. static removeSingleLineCssComments(stylesheet) {
  1454. const removedRules = [];
  1455. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1456. const ruleData = cssRule.data;
  1457. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1458. removedRules.push(cssRule);
  1459. }
  1460. }
  1461. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1462. }
  1463. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1464. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1465. const imports = getImportFunctions(stylesheetContent);
  1466. await Promise.all(imports.map(async cssImport => {
  1467. const match = matchImport(cssImport);
  1468. if (match) {
  1469. const regExpCssImport = getRegExp(cssImport);
  1470. let resourceURL = normalizeURL(match.resourceURL);
  1471. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1472. try {
  1473. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1474. } catch (error) {
  1475. // ignored
  1476. }
  1477. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1478. const content = await util.getContent(resourceURL, {
  1479. maxResourceSize: options.maxResourceSize,
  1480. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1481. validateTextContentType: true,
  1482. frameId: options.frameId
  1483. });
  1484. resourceURL = content.resourceURL;
  1485. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1486. let importedStylesheetContent = removeCssComments(content.data);
  1487. if (options.compressCSS) {
  1488. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1489. }
  1490. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1491. if (stylesheetContent.includes(cssImport)) {
  1492. const ancestorStyleSheets = new Set(importedStyleSheets);
  1493. ancestorStyleSheets.add(resourceURL);
  1494. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1495. workStylesheet.textContent = importedStylesheetContent;
  1496. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1497. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1498. } else {
  1499. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1500. }
  1501. }
  1502. } else {
  1503. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1504. }
  1505. } else {
  1506. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1507. }
  1508. }
  1509. }));
  1510. return stylesheetContent;
  1511. }
  1512. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1513. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1514. urlFunctions.map(urlFunction => {
  1515. const originalResourceURL = matchURL(urlFunction);
  1516. const resourceURL = normalizeURL(originalResourceURL);
  1517. if (!testIgnoredPath(resourceURL)) {
  1518. if (!resourceURL || testValidPath(resourceURL)) {
  1519. let resolvedURL;
  1520. if (!originalResourceURL.startsWith("#")) {
  1521. try {
  1522. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1523. } catch (error) {
  1524. // ignored
  1525. }
  1526. }
  1527. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1528. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1529. }
  1530. } else {
  1531. let newUrlFunction;
  1532. if (originalResourceURL) {
  1533. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1534. } else {
  1535. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1536. }
  1537. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1538. }
  1539. }
  1540. });
  1541. return stylesheetContent;
  1542. }
  1543. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1544. resourceURL = normalizeURL(resourceURL);
  1545. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1546. const content = await util.getContent(resourceURL, {
  1547. maxResourceSize: options.maxResourceSize,
  1548. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1549. charset: options.charset,
  1550. frameId: options.frameId,
  1551. validateTextContentType: true
  1552. });
  1553. resourceURL = content.resourceURL;
  1554. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1555. let stylesheetContent = removeCssComments(content.data);
  1556. if (options.compressCSS) {
  1557. stylesheetContent = util.compressCSS(stylesheetContent);
  1558. }
  1559. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1560. return stylesheetContent;
  1561. }
  1562. }
  1563. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1564. const promises = [];
  1565. const removedRules = [];
  1566. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1567. const ruleData = cssRule.data;
  1568. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1569. removedRules.push(cssRule);
  1570. } else if (ruleData.block && ruleData.block.children) {
  1571. if (ruleData.type == "Rule") {
  1572. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1573. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1574. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1575. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1576. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1577. }
  1578. }
  1579. }
  1580. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1581. await Promise.all(promises);
  1582. async function processFontFaceRule(ruleData, fontURLs) {
  1583. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1584. if (declaration.type == "Declaration" && declaration.value.children) {
  1585. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1586. await Promise.all(urlFunctions.map(async urlFunction => {
  1587. const originalResourceURL = matchURL(urlFunction);
  1588. const resourceURL = normalizeURL(originalResourceURL);
  1589. if (!testIgnoredPath(resourceURL)) {
  1590. if (testValidURL(resourceURL)) {
  1591. let { content } = await batchRequest.addURL(resourceURL, true);
  1592. let resourceURLs = fontURLs.get(declaration);
  1593. if (!resourceURLs) {
  1594. resourceURLs = [];
  1595. fontURLs.set(declaration, resourceURLs);
  1596. }
  1597. resourceURLs.push(resourceURL);
  1598. replaceURLs(declaration, originalResourceURL, content);
  1599. }
  1600. }
  1601. }));
  1602. }
  1603. }));
  1604. function replaceURLs(declaration, oldURL, newURL) {
  1605. declaration.value.children.forEach(token => {
  1606. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1607. token.value.value = newURL;
  1608. }
  1609. });
  1610. }
  1611. }
  1612. }
  1613. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1614. await Promise.all(declarations.map(async declaration => {
  1615. if (declaration.type == "Declaration" && declaration.value.children) {
  1616. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1617. await Promise.all(urlFunctions.map(async urlFunction => {
  1618. const originalResourceURL = matchURL(urlFunction);
  1619. const resourceURL = normalizeURL(originalResourceURL);
  1620. if (!testIgnoredPath(resourceURL)) {
  1621. if (testValidURL(resourceURL)) {
  1622. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1623. let variableDefined;
  1624. const tokens = [];
  1625. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1626. if (!originalResourceURL.startsWith("#")) {
  1627. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1628. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1629. tokens.push({ parent, token, value });
  1630. variableDefined = true;
  1631. } else {
  1632. token.data.value.value = content;
  1633. }
  1634. }
  1635. });
  1636. if (variableDefined) {
  1637. cssVariables.set(indexResource, content);
  1638. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1639. }
  1640. }
  1641. }
  1642. }));
  1643. }
  1644. }));
  1645. function findURLToken(url, children, callback, depth = 0) {
  1646. for (let token = children.head; token; token = token.next) {
  1647. if (token.data.children) {
  1648. findURLToken(url, token.data.children, callback, depth + 1);
  1649. }
  1650. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1651. callback(token, children, depth == 0);
  1652. }
  1653. }
  1654. }
  1655. }
  1656. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1657. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1658. let resourceURL = resourceElement.getAttribute(attributeName);
  1659. if (resourceURL != null) {
  1660. resourceURL = normalizeURL(resourceURL);
  1661. let originURL = resourceElement.dataset.singleFileOriginURL;
  1662. delete resourceElement.dataset.singleFileOriginURL;
  1663. if (!testIgnoredPath(resourceURL)) {
  1664. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1665. if (testValidPath(resourceURL)) {
  1666. try {
  1667. resourceURL = util.resolveURL(resourceURL, baseURI);
  1668. } catch (error) {
  1669. // ignored
  1670. }
  1671. if (testValidURL(resourceURL)) {
  1672. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1673. if (originURL) {
  1674. if (content == EMPTY_DATA_URI) {
  1675. try {
  1676. originURL = util.resolveURL(originURL, baseURI);
  1677. } catch (error) {
  1678. // ignored
  1679. }
  1680. try {
  1681. resourceURL = originURL;
  1682. content = (await util.getContent(resourceURL, {
  1683. asBinary: true,
  1684. maxResourceSize: options.maxResourceSize,
  1685. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1686. frameId: options.windowId
  1687. })).data;
  1688. } catch (error) {
  1689. // ignored
  1690. }
  1691. }
  1692. }
  1693. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1694. resourceElement.remove();
  1695. } else {
  1696. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1697. if (!forbiddenPrefixFound) {
  1698. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1699. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1700. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1701. cssVariables.set(indexResource, content);
  1702. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1703. styles.set(resourceElement, declarationList);
  1704. } else {
  1705. resourceElement.setAttribute(attributeName, content);
  1706. }
  1707. } else {
  1708. resourceElement.setAttribute(attributeName, content);
  1709. }
  1710. }
  1711. }
  1712. }
  1713. }
  1714. }
  1715. }
  1716. }));
  1717. }
  1718. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1719. let attributeName = "xlink:href";
  1720. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1721. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1722. if (originalResourceURL == null) {
  1723. attributeName = "href";
  1724. originalResourceURL = resourceElement.getAttribute(attributeName);
  1725. }
  1726. let resourceURL = normalizeURL(originalResourceURL);
  1727. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1728. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1729. try {
  1730. resourceURL = util.resolveURL(resourceURL, baseURI);
  1731. } catch (error) {
  1732. // ignored
  1733. }
  1734. if (testValidURL(resourceURL)) {
  1735. const { content } = await batchRequest.addURL(resourceURL);
  1736. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1737. if (hashMatch && hashMatch[0]) {
  1738. let symbolElement;
  1739. try {
  1740. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1741. } catch (error) {
  1742. // ignored
  1743. }
  1744. if (symbolElement) {
  1745. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1746. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1747. }
  1748. } else {
  1749. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1750. }
  1751. }
  1752. } else if (resourceURL == options.url) {
  1753. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1754. }
  1755. }));
  1756. }
  1757. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1758. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1759. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1760. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1761. let resourceURL = normalizeURL(srcsetValue.url);
  1762. if (!testIgnoredPath(resourceURL)) {
  1763. if (testValidPath(resourceURL)) {
  1764. try {
  1765. resourceURL = util.resolveURL(resourceURL, baseURI);
  1766. } catch (error) {
  1767. // ignored
  1768. }
  1769. if (testValidURL(resourceURL)) {
  1770. const { content } = await batchRequest.addURL(resourceURL, true);
  1771. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1772. if (forbiddenPrefixFound) {
  1773. return "";
  1774. }
  1775. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1776. } else {
  1777. return "";
  1778. }
  1779. } else {
  1780. return "";
  1781. }
  1782. } else {
  1783. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1784. }
  1785. }));
  1786. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1787. }));
  1788. }
  1789. static replaceImageSource(imgElement, variableName, options) {
  1790. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1791. if (attributeValue) {
  1792. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1793. if (imageData && imageData.replaceable) {
  1794. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1795. const backgroundStyle = {};
  1796. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1797. if (backgroundSize) {
  1798. backgroundStyle["background-size"] = imageData.objectFit;
  1799. }
  1800. if (imageData.objectPosition) {
  1801. backgroundStyle["background-position"] = imageData.objectPosition;
  1802. }
  1803. if (imageData.backgroundColor) {
  1804. backgroundStyle["background-color"] = imageData.backgroundColor;
  1805. }
  1806. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1807. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1808. return true;
  1809. }
  1810. }
  1811. }
  1812. }
  1813. // ----
  1814. // Util
  1815. // ----
  1816. const BLOB_URI_PREFIX = "blob:";
  1817. const HTTP_URI_PREFIX = /^https?:\/\//;
  1818. const FILE_URI_PREFIX = /^file:\/\//;
  1819. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1820. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1821. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1822. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1823. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1824. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1825. 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;
  1826. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1827. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1828. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1829. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1830. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1831. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1832. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1833. function getUpdatedResourceContent(resourceURL, content, options) {
  1834. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1835. options.updatedResources[resourceURL].retrieved = true;
  1836. return options.updatedResources[resourceURL].content;
  1837. } else {
  1838. return content.data || "";
  1839. }
  1840. }
  1841. function normalizeURL(url) {
  1842. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1843. return url;
  1844. } else {
  1845. return url.split("#")[0];
  1846. }
  1847. }
  1848. function getCSSValue(value) {
  1849. let result = "";
  1850. try {
  1851. result = cssTree.generate(value);
  1852. } catch (error) {
  1853. // ignored
  1854. }
  1855. return result;
  1856. }
  1857. function getOnEventAttributeNames(doc) {
  1858. const element = doc.createElement("div");
  1859. const attributeNames = [];
  1860. for (const propertyName in element) {
  1861. if (propertyName.startsWith("on")) {
  1862. attributeNames.push(propertyName);
  1863. }
  1864. }
  1865. attributeNames.push("onunload");
  1866. return attributeNames;
  1867. }
  1868. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1869. let maxLength;
  1870. if (template) {
  1871. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1872. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1873. if (template.match(replaceRegExp)) {
  1874. const matchedLength = template.match(replaceRegExp)[0];
  1875. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1876. if (isNaN(maxLength) || maxLength <= 0) {
  1877. maxLength = null;
  1878. }
  1879. } else {
  1880. replaceRegExp = new RegExp(regExpVariable, "g");
  1881. }
  1882. if (template.match(replaceRegExp)) {
  1883. let value = await valueGetter();
  1884. if (!dontReplaceSlash) {
  1885. value = value.replace(/\/+/g, replacementCharacter);
  1886. }
  1887. if (maxLength) {
  1888. value = await util.truncateText(value, maxLength);
  1889. }
  1890. return template.replace(replaceRegExp, value);
  1891. }
  1892. }
  1893. return template;
  1894. }
  1895. function getLastSegment(url, replacementCharacter) {
  1896. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1897. if (!lastSegment) {
  1898. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1899. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1900. }
  1901. if (!lastSegment) {
  1902. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1903. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1904. }
  1905. if (!lastSegment) {
  1906. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1907. }
  1908. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1909. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1910. lastSegment = lastSegmentMatch[1];
  1911. }
  1912. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1913. return lastSegment;
  1914. }
  1915. function getRegExp(string) {
  1916. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1917. }
  1918. function getUrlFunctions(stylesheetContent, unique) {
  1919. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1920. if (unique) {
  1921. return [...new Set(result)];
  1922. } else {
  1923. return result;
  1924. }
  1925. }
  1926. function getImportFunctions(stylesheetContent) {
  1927. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1928. }
  1929. function findShortcutIcon(shortcutIcons) {
  1930. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1931. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1932. return shortcutIcons[0];
  1933. }
  1934. function matchURL(stylesheetContent) {
  1935. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1936. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1937. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1938. return match && match[1];
  1939. }
  1940. function testIgnoredPath(resourceURL) {
  1941. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1942. }
  1943. function testValidPath(resourceURL) {
  1944. return resourceURL && !resourceURL.match(EMPTY_URL);
  1945. }
  1946. function testValidURL(resourceURL) {
  1947. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1948. }
  1949. function matchImport(stylesheetContent) {
  1950. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1951. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1952. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1953. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1954. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1955. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1956. if (match) {
  1957. const [, resourceURL, media] = match;
  1958. return { resourceURL, media };
  1959. }
  1960. }
  1961. function removeCssComments(stylesheetContent) {
  1962. try {
  1963. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  1964. } catch (error) {
  1965. let start, end;
  1966. do {
  1967. start = stylesheetContent.indexOf("/*");
  1968. end = stylesheetContent.indexOf("*/", start + 2);
  1969. if (start != -1 && end != -1) {
  1970. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1971. }
  1972. } while (start != -1 && end != -1);
  1973. return stylesheetContent;
  1974. }
  1975. }
  1976. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  1977. if (mediaQuery) {
  1978. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1979. } else {
  1980. return stylesheetContent;
  1981. }
  1982. }
  1983. function log(...args) {
  1984. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1985. }
  1986. // -----
  1987. // Stats
  1988. // -----
  1989. const STATS_DEFAULT_VALUES = {
  1990. discarded: {
  1991. "HTML bytes": 0,
  1992. "hidden elements": 0,
  1993. "HTML imports": 0,
  1994. scripts: 0,
  1995. objects: 0,
  1996. "audio sources": 0,
  1997. "video sources": 0,
  1998. frames: 0,
  1999. "CSS rules": 0,
  2000. canvas: 0,
  2001. stylesheets: 0,
  2002. resources: 0,
  2003. medias: 0
  2004. },
  2005. processed: {
  2006. "HTML bytes": 0,
  2007. "hidden elements": 0,
  2008. "HTML imports": 0,
  2009. scripts: 0,
  2010. objects: 0,
  2011. "audio sources": 0,
  2012. "video sources": 0,
  2013. frames: 0,
  2014. "CSS rules": 0,
  2015. canvas: 0,
  2016. stylesheets: 0,
  2017. resources: 0,
  2018. medias: 0
  2019. }
  2020. };
  2021. class Stats {
  2022. constructor(options) {
  2023. this.options = options;
  2024. if (options.displayStats) {
  2025. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2026. }
  2027. }
  2028. set(type, subType, value) {
  2029. if (this.options.displayStats) {
  2030. this.data[type][subType] = value;
  2031. }
  2032. }
  2033. add(type, subType, value) {
  2034. if (this.options.displayStats) {
  2035. this.data[type][subType] += value;
  2036. }
  2037. }
  2038. addAll(pageData) {
  2039. if (this.options.displayStats) {
  2040. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2041. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2042. }
  2043. }
  2044. }
  2045. return { getClass };
  2046. })();