single-file-core.js 91 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237
  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. /* global globalThis */
  24. const DEBUG = false;
  25. const Set = globalThis.Set;
  26. const Map = globalThis.Map;
  27. let util, cssTree;
  28. function getClass(...args) {
  29. [util, cssTree] = args;
  30. return SingleFileClass;
  31. }
  32. class SingleFileClass {
  33. constructor(options) {
  34. this.options = options;
  35. }
  36. async run() {
  37. if (this.options.userScriptEnabled && util.waitForUserScript.callback) {
  38. await util.waitForUserScript.callback(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  39. }
  40. this.runner = new Runner(this.options, true);
  41. await this.runner.loadPage();
  42. await this.runner.initialize();
  43. if (this.options.userScriptEnabled && util.waitForUserScript.callback) {
  44. await util.waitForUserScript.callback(util.ON_AFTER_CAPTURE_EVENT_NAME);
  45. }
  46. await this.runner.run();
  47. }
  48. cancel() {
  49. this.cancelled = true;
  50. if (this.runner) {
  51. this.runner.cancel();
  52. }
  53. }
  54. getPageData() {
  55. return this.runner.getPageData();
  56. }
  57. }
  58. // -------------
  59. // ProgressEvent
  60. // -------------
  61. const PAGE_LOADING = "page-loading";
  62. const PAGE_LOADED = "page-loaded";
  63. const RESOURCES_INITIALIZING = "resource-initializing";
  64. const RESOURCES_INITIALIZED = "resources-initialized";
  65. const RESOURCE_LOADED = "resource-loaded";
  66. const PAGE_ENDED = "page-ended";
  67. const STAGE_STARTED = "stage-started";
  68. const STAGE_ENDED = "stage-ended";
  69. const STAGE_TASK_STARTED = "stage-task-started";
  70. const STAGE_TASK_ENDED = "stage-task-ended";
  71. class ProgressEvent {
  72. constructor(type, detail) {
  73. 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 };
  74. }
  75. }
  76. // ------
  77. // Runner
  78. // ------
  79. const RESOLVE_URLS_STAGE = 0;
  80. const REPLACE_DATA_STAGE = 1;
  81. const REPLACE_DOCS_STAGE = 2;
  82. const POST_PROCESS_STAGE = 3;
  83. const STAGES = [{
  84. sequential: [
  85. { action: "preProcessPage" },
  86. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  87. { action: "replaceStyleContents" },
  88. { action: "resetCharsetMeta" },
  89. { option: "saveFavicon", action: "saveFavicon" },
  90. { action: "replaceCanvasElements" },
  91. { action: "insertFonts" },
  92. { action: "insertShadowRootContents" },
  93. { action: "setInputValues" },
  94. { option: "removeScripts", action: "removeScripts" },
  95. { option: "selected", action: "removeUnselectedElements" },
  96. { option: "removeVideoSrc", action: "insertVideoPosters" },
  97. { option: "removeFrames", action: "removeFrames" },
  98. { option: "removeVideoSrc", action: "removeVideoSources" },
  99. { option: "removeAudioSrc", action: "removeAudioSources" },
  100. { action: "removeDiscardedResources" },
  101. { option: "removeHiddenElements", action: "removeHiddenElements" },
  102. { action: "resolveHrefs" },
  103. { action: "resolveStyleAttributeURLs" }
  104. ],
  105. parallel: [
  106. { action: "resolveStylesheetURLs" },
  107. { option: "!removeFrames", action: "resolveFrameURLs" },
  108. { action: "resolveHtmlImportURLs" }
  109. ]
  110. }, {
  111. sequential: [
  112. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  113. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  114. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  115. ],
  116. parallel: [
  117. { action: "processStylesheets" },
  118. { action: "processStyleAttributes" },
  119. { action: "processPageResources" },
  120. { option: "!removeScripts", action: "processScripts" }
  121. ]
  122. }, {
  123. sequential: [
  124. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  125. ],
  126. parallel: [
  127. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  128. { option: "!removeFrames", action: "processFrames" },
  129. { option: "!removeImports", action: "processHtmlImports" },
  130. ]
  131. }, {
  132. sequential: [
  133. { action: "replaceStylesheets" },
  134. { action: "replaceStyleAttributes" },
  135. { action: "insertVariables" },
  136. { option: "compressHTML", action: "compressHTML" },
  137. { action: "cleanupPage" }
  138. ],
  139. parallel: [
  140. { option: "enableMaff", action: "insertMAFFMetaData" },
  141. { action: "setDocInfo" }
  142. ]
  143. }];
  144. class Runner {
  145. constructor(options, root) {
  146. const rootDocDefined = root && options.doc;
  147. this.root = root;
  148. this.options = options;
  149. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  150. const matchResourceReferrer = this.options.url.match(/^.*\//);
  151. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  152. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  153. this.options.rootDocument = root;
  154. this.options.updatedResources = this.options.updatedResources || {};
  155. this.options.fontTests = new Map();
  156. this.batchRequest = new BatchRequest();
  157. this.processor = new Processor(options, this.batchRequest);
  158. if (rootDocDefined) {
  159. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  160. this.options.canvases = docData.canvases;
  161. this.options.fonts = docData.fonts;
  162. this.options.stylesheets = docData.stylesheets;
  163. this.options.images = docData.images;
  164. this.options.posters = docData.posters;
  165. this.options.usedFonts = docData.usedFonts;
  166. this.options.shadowRoots = docData.shadowRoots;
  167. this.options.imports = docData.imports;
  168. this.options.referrer = docData.referrer;
  169. this.markedElements = docData.markedElements;
  170. }
  171. if (this.options.saveRawPage) {
  172. this.options.removeFrames = true;
  173. }
  174. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  175. this.onprogress = options.onprogress || (() => { });
  176. }
  177. async loadPage() {
  178. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  179. await this.processor.loadPage(this.options.content);
  180. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  181. }
  182. async initialize() {
  183. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  184. await this.executeStage(RESOLVE_URLS_STAGE);
  185. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  186. if (this.root && this.options.doc) {
  187. util.postProcessDoc(this.options.doc, this.markedElements);
  188. }
  189. }
  190. cancel() {
  191. this.cancelled = true;
  192. this.batchRequest.cancel();
  193. if (this.root) {
  194. if (this.options.frames) {
  195. this.options.frames.forEach(cancelRunner);
  196. }
  197. if (this.options.imports) {
  198. this.options.imports.forEach(cancelRunner);
  199. }
  200. }
  201. function cancelRunner(resourceData) {
  202. if (resourceData.runner) {
  203. resourceData.runner.cancel();
  204. }
  205. }
  206. }
  207. async run() {
  208. if (this.root) {
  209. this.processor.initialize(this.batchRequest);
  210. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  211. }
  212. await this.batchRequest.run(detail => {
  213. detail.pageURL = this.options.url;
  214. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  215. }, this.options);
  216. await this.pendingPromises;
  217. this.options.doc = null;
  218. this.options.win = null;
  219. await this.executeStage(REPLACE_DOCS_STAGE);
  220. await this.executeStage(POST_PROCESS_STAGE);
  221. this.processor.finalize();
  222. }
  223. getDocument() {
  224. return this.processor.doc;
  225. }
  226. getStyleSheets() {
  227. return this.processor.stylesheets;
  228. }
  229. getPageData() {
  230. if (this.root) {
  231. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  232. }
  233. return this.processor.getPageData();
  234. }
  235. async executeStage(step) {
  236. if (DEBUG) {
  237. log("**** STARTED STAGE", step, "****");
  238. }
  239. const frame = !this.root;
  240. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  241. STAGES[step].sequential.forEach(task => {
  242. let startTime;
  243. if (DEBUG) {
  244. startTime = Date.now();
  245. log(" -- STARTED task =", task.action);
  246. }
  247. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  248. if (!this.cancelled) {
  249. this.executeTask(task);
  250. }
  251. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  252. if (DEBUG) {
  253. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  254. }
  255. });
  256. let parallelTasksPromise;
  257. if (STAGES[step].parallel) {
  258. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  259. let startTime;
  260. if (DEBUG) {
  261. startTime = Date.now();
  262. log(" // STARTED task =", task.action);
  263. }
  264. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  265. if (!this.cancelled) {
  266. await this.executeTask(task);
  267. }
  268. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  269. if (DEBUG) {
  270. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  271. }
  272. }));
  273. } else {
  274. parallelTasksPromise = Promise.resolve();
  275. }
  276. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  277. if (DEBUG) {
  278. log("**** ENDED STAGE", step, "****");
  279. }
  280. return parallelTasksPromise;
  281. }
  282. executeTask(task) {
  283. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  284. return this.processor[task.action]();
  285. }
  286. }
  287. }
  288. // ------------
  289. // BatchRequest
  290. // ------------
  291. class BatchRequest {
  292. constructor() {
  293. this.requests = new Map();
  294. this.duplicates = new Map();
  295. }
  296. addURL(resourceURL, asBinary, expectedType, groupDuplicates) {
  297. return new Promise((resolve, reject) => {
  298. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType]);
  299. let resourceRequests = this.requests.get(requestKey);
  300. if (!resourceRequests) {
  301. resourceRequests = [];
  302. this.requests.set(requestKey, resourceRequests);
  303. }
  304. const callbacks = { resolve, reject };
  305. resourceRequests.push(callbacks);
  306. if (groupDuplicates) {
  307. let duplicateRequests = this.duplicates.get(requestKey);
  308. if (!duplicateRequests) {
  309. duplicateRequests = [];
  310. this.duplicates.set(requestKey, duplicateRequests);
  311. }
  312. duplicateRequests.push(callbacks);
  313. }
  314. });
  315. }
  316. getMaxResources() {
  317. return this.requests.size;
  318. }
  319. run(onloadListener, options) {
  320. const resourceURLs = [...this.requests.keys()];
  321. let indexResource = 0;
  322. return Promise.all(resourceURLs.map(async requestKey => {
  323. const [resourceURL, asBinary, expectedType] = JSON.parse(requestKey);
  324. const resourceRequests = this.requests.get(requestKey);
  325. try {
  326. const currentIndexResource = indexResource;
  327. indexResource = indexResource + 1;
  328. const content = await util.getContent(resourceURL, {
  329. asBinary,
  330. expectedType,
  331. maxResourceSize: options.maxResourceSize,
  332. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  333. frameId: options.windowId,
  334. resourceReferrer: options.resourceReferrer
  335. });
  336. onloadListener({ url: resourceURL });
  337. if (!this.cancelled) {
  338. resourceRequests.forEach(callbacks => {
  339. const duplicateCallbacks = this.duplicates.get(requestKey);
  340. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  341. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  342. });
  343. }
  344. } catch (error) {
  345. indexResource = indexResource + 1;
  346. onloadListener({ url: resourceURL });
  347. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  348. }
  349. this.requests.delete(requestKey);
  350. }));
  351. }
  352. cancel() {
  353. this.cancelled = true;
  354. const resourceURLs = [...this.requests.keys()];
  355. resourceURLs.forEach(requestKey => {
  356. const resourceRequests = this.requests.get(requestKey);
  357. resourceRequests.forEach(callbacks => callbacks.reject());
  358. this.requests.delete(requestKey);
  359. });
  360. }
  361. }
  362. // ---------
  363. // Processor
  364. // ---------
  365. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  366. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  367. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  368. const SCRIPT_TAG_FOUND = /<script/gi;
  369. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  370. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  371. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  372. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  373. const UTF8_CHARSET = "utf-8";
  374. class Processor {
  375. constructor(options, batchRequest) {
  376. this.options = options;
  377. this.stats = new Stats(options);
  378. this.baseURI = normalizeURL(options.baseURI || options.url);
  379. this.batchRequest = batchRequest;
  380. this.stylesheets = new Map();
  381. this.styles = new Map();
  382. this.cssVariables = new Map();
  383. this.fontTests = options.fontTests;
  384. }
  385. initialize() {
  386. this.options.saveDate = new Date();
  387. this.options.saveUrl = this.options.url;
  388. if (this.options.enableMaff) {
  389. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url));
  390. }
  391. this.maxResources = this.batchRequest.getMaxResources();
  392. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  393. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  394. }
  395. if (!this.options.removeImports && this.options.imports) {
  396. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  397. }
  398. this.stats.set("processed", "resources", this.maxResources);
  399. }
  400. async loadPage(pageContent, charset) {
  401. let content;
  402. if (!pageContent || this.options.saveRawPage) {
  403. content = await util.getContent(this.baseURI, {
  404. maxResourceSize: this.options.maxResourceSize,
  405. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  406. charset,
  407. frameId: this.options.windowId,
  408. resourceReferrer: this.options.resourceReferrer
  409. });
  410. pageContent = content.data;
  411. }
  412. this.doc = util.parseDocContent(pageContent, this.baseURI);
  413. if (this.options.saveRawPage) {
  414. let charset;
  415. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  416. const charsetDeclaration = element.content.split(";")[1];
  417. if (charsetDeclaration && !charset) {
  418. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  419. }
  420. });
  421. if (charset && charset != content.charset) {
  422. return this.loadPage(pageContent, charset);
  423. }
  424. }
  425. this.workStyleElement = this.doc.createElement("style");
  426. this.doc.body.appendChild(this.workStyleElement);
  427. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  428. }
  429. finalize() {
  430. if (this.workStyleElement.parentNode) {
  431. this.workStyleElement.remove();
  432. }
  433. }
  434. async getPageData() {
  435. util.postProcessDoc(this.doc);
  436. const url = util.parseURL(this.baseURI);
  437. if (this.options.insertSingleFileComment) {
  438. const firstComment = this.doc.documentElement.firstChild;
  439. let infobarURL = this.options.saveUrl, infobarSaveDate = this.options.saveDate;
  440. if (firstComment.nodeType == 8 && (firstComment.textContent.includes(util.COMMENT_HEADER_LEGACY) || firstComment.textContent.includes(util.COMMENT_HEADER))) {
  441. const info = this.doc.documentElement.firstChild.textContent.split("\n");
  442. const [, , url, saveDate] = info;
  443. infobarURL = url.split("url: ")[1];
  444. infobarSaveDate = saveDate.split("saved date: ")[1];
  445. firstComment.remove();
  446. }
  447. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  448. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  449. " \n url: " + infobarURL +
  450. " \n saved date: " + infobarSaveDate +
  451. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  452. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  453. }
  454. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  455. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  456. if (!canonicalLink) {
  457. canonicalLink = this.doc.createElement("link");
  458. canonicalLink.setAttribute("rel", "canonical");
  459. this.doc.head.appendChild(canonicalLink);
  460. }
  461. if (canonicalLink && !canonicalLink.href) {
  462. canonicalLink.href = this.options.saveUrl;
  463. }
  464. }
  465. if (this.options.insertMetaNoIndex) {
  466. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  467. if (!metaElement) {
  468. metaElement = this.doc.createElement("meta");
  469. metaElement.setAttribute("name", "robots");
  470. metaElement.setAttribute("content", "noindex");
  471. this.doc.head.appendChild(metaElement);
  472. }
  473. }
  474. let size;
  475. if (this.options.displayStats) {
  476. size = util.getContentSize(this.doc.documentElement.outerHTML);
  477. }
  478. const content = util.serialize(this.doc, this.options.compressHTML);
  479. if (this.options.displayStats) {
  480. const contentSize = util.getContentSize(content);
  481. this.stats.set("processed", "HTML bytes", contentSize);
  482. this.stats.add("discarded", "HTML bytes", size - contentSize);
  483. }
  484. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  485. const replacementCharacter = this.options.filenameReplacementCharacter;
  486. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  487. if (!this.options.backgroundSave) {
  488. filename = filename.replace(/\//g, replacementCharacter);
  489. }
  490. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  491. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  492. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  493. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  494. filename = filename + "…" + extension;
  495. }
  496. if (!filename) {
  497. filename = "Unnamed page";
  498. }
  499. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  500. const pageData = {
  501. stats: this.stats.data,
  502. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  503. filename,
  504. content
  505. };
  506. if (this.options.addProof) {
  507. pageData.hash = await util.digest("SHA-256", content);
  508. }
  509. if (this.options.retrieveLinks) {
  510. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  511. }
  512. return pageData;
  513. }
  514. preProcessPage() {
  515. if (this.options.win) {
  516. 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));
  517. }
  518. if (this.options.images && !this.options.saveRawPage) {
  519. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  520. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  521. if (attributeValue) {
  522. const imageData = this.options.images[Number(attributeValue)];
  523. if (imageData) {
  524. if (this.options.removeHiddenElements && (
  525. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  526. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  527. )) {
  528. imgElement.setAttribute("src", EMPTY_IMAGE);
  529. } else {
  530. if (imageData.currentSrc) {
  531. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  532. imgElement.setAttribute("src", imageData.currentSrc);
  533. }
  534. if (this.options.loadDeferredImages) {
  535. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  536. imageData.src = imgElement.dataset.src;
  537. imgElement.setAttribute("src", imgElement.dataset.src);
  538. imgElement.removeAttribute("data-src");
  539. }
  540. }
  541. }
  542. }
  543. }
  544. });
  545. if (this.options.loadDeferredImages) {
  546. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  547. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  548. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  549. imgElement.removeAttribute("data-srcset");
  550. }
  551. });
  552. }
  553. }
  554. }
  555. replaceStyleContents() {
  556. if (this.options.stylesheets) {
  557. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  558. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  559. if (attributeValue) {
  560. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  561. if (stylesheetContent) {
  562. styleElement.textContent = stylesheetContent;
  563. }
  564. }
  565. });
  566. }
  567. }
  568. removeUnselectedElements() {
  569. removeUnmarkedElements(this.doc.body);
  570. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  571. function removeUnmarkedElements(element) {
  572. let selectedElementFound = false;
  573. Array.from(element.childNodes).forEach(node => {
  574. if (node.nodeType == 1) {
  575. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  576. selectedElementFound = selectedElementFound || isSelectedElement;
  577. if (isSelectedElement) {
  578. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  579. removeUnmarkedElements(node);
  580. } else if (selectedElementFound) {
  581. removeNode(node);
  582. } else {
  583. hideNode(node);
  584. }
  585. }
  586. });
  587. }
  588. function removeNode(node) {
  589. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  590. node.remove();
  591. } else {
  592. hideNode(node);
  593. }
  594. }
  595. function hideNode(node) {
  596. if (canHideNode(node)) {
  597. node.style.setProperty("display", "none", "important");
  598. Array.from(node.childNodes).forEach(removeNode);
  599. }
  600. }
  601. function canHideNode(node) {
  602. if (node.nodeType == 1) {
  603. const tagName = node.tagName && node.tagName.toLowerCase();
  604. return (tagName != "svg" && tagName != "style" && tagName != "link");
  605. }
  606. }
  607. }
  608. insertVideoPosters() {
  609. if (this.options.posters) {
  610. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  611. let videoElement;
  612. if (element.tagName == "VIDEO") {
  613. videoElement = element;
  614. } else {
  615. videoElement = element.parentElement;
  616. }
  617. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  618. if (attributeValue) {
  619. const posterURL = this.options.posters[Number(attributeValue)];
  620. if (!videoElement.poster && posterURL) {
  621. videoElement.setAttribute("poster", posterURL);
  622. }
  623. }
  624. });
  625. }
  626. }
  627. removeFrames() {
  628. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  629. this.stats.set("discarded", "frames", frameElements.length);
  630. this.stats.set("processed", "frames", frameElements.length);
  631. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  632. }
  633. removeScripts() {
  634. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  635. this.doc.querySelectorAll("[href]").forEach(element => {
  636. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  637. element.setAttribute("href", "");
  638. }
  639. });
  640. this.doc.querySelectorAll("[src]").forEach(element => {
  641. if (element.src && element.src.match(/^\s*javascript:/)) {
  642. element.removeAttribute("src");
  643. }
  644. });
  645. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  646. this.stats.set("discarded", "scripts", scriptElements.length);
  647. this.stats.set("processed", "scripts", scriptElements.length);
  648. scriptElements.forEach(element => element.remove());
  649. }
  650. removeVideoSources() {
  651. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  652. this.stats.set("discarded", "video sources", videoSourceElements.length);
  653. this.stats.set("processed", "video sources", videoSourceElements.length);
  654. videoSourceElements.forEach(element => {
  655. if (element.tagName == "SOURCE") {
  656. element.remove();
  657. } else {
  658. videoSourceElements.forEach(element => element.removeAttribute("src"));
  659. }
  660. });
  661. }
  662. removeAudioSources() {
  663. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  664. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  665. this.stats.set("processed", "audio sources", audioSourceElements.length);
  666. audioSourceElements.forEach(element => {
  667. if (element.tagName == "SOURCE") {
  668. element.remove();
  669. } else {
  670. audioSourceElements.forEach(element => element.removeAttribute("src"));
  671. }
  672. });
  673. }
  674. removeDiscardedResources() {
  675. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  676. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  677. 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\"])");
  678. this.stats.set("discarded", "objects", objectElements.length);
  679. this.stats.set("processed", "objects", objectElements.length);
  680. objectElements.forEach(element => element.remove());
  681. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  682. replacedAttributeValue.forEach(element => {
  683. let regExp;
  684. if (this.options.removeScripts) {
  685. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  686. } else {
  687. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  688. }
  689. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  690. if (relValue.length) {
  691. element.setAttribute("rel", relValue);
  692. } else {
  693. element.remove();
  694. }
  695. });
  696. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  697. if (this.options.removeHiddenElements) {
  698. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  699. }
  700. if (!this.options.saveFavicon) {
  701. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  702. }
  703. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  704. }
  705. resetCharsetMeta() {
  706. let charset;
  707. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  708. const charsetDeclaration = element.content.split(";")[1];
  709. if (charsetDeclaration && !charset) {
  710. charset = charsetDeclaration.split("=")[1];
  711. if (charset) {
  712. this.charset = charset.trim().toLowerCase();
  713. }
  714. }
  715. element.remove();
  716. });
  717. const metaElement = this.doc.createElement("meta");
  718. metaElement.setAttribute("charset", UTF8_CHARSET);
  719. if (this.doc.head.firstChild) {
  720. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  721. } else {
  722. this.doc.head.appendChild(metaElement);
  723. }
  724. }
  725. setInputValues() {
  726. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  727. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  728. input.setAttribute("value", value || "");
  729. });
  730. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  731. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  732. if (value == "true") {
  733. input.setAttribute("checked", "");
  734. }
  735. });
  736. this.doc.querySelectorAll("textarea").forEach(textarea => {
  737. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  738. textarea.textContent = value || "";
  739. });
  740. this.doc.querySelectorAll("select").forEach(select => {
  741. select.querySelectorAll("option").forEach(option => {
  742. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  743. if (selected) {
  744. option.setAttribute("selected", "");
  745. }
  746. });
  747. });
  748. }
  749. saveFavicon() {
  750. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  751. if (!faviconElement) {
  752. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  753. }
  754. if (!faviconElement) {
  755. faviconElement = this.doc.createElement("link");
  756. faviconElement.setAttribute("type", "image/x-icon");
  757. faviconElement.setAttribute("rel", "shortcut icon");
  758. faviconElement.setAttribute("href", "/favicon.ico");
  759. }
  760. this.doc.head.appendChild(faviconElement);
  761. }
  762. replaceCanvasElements() {
  763. if (this.options.canvases) {
  764. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  765. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  766. if (attributeValue) {
  767. const canvasData = this.options.canvases[Number(attributeValue)];
  768. if (canvasData) {
  769. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  770. this.stats.add("processed", "canvas", 1);
  771. }
  772. }
  773. });
  774. }
  775. }
  776. insertFonts() {
  777. if (this.options.fonts && this.options.fonts.length) {
  778. let stylesheetContent = "";
  779. this.options.fonts.forEach(fontData => {
  780. if (fontData["font-family"] && fontData.src) {
  781. stylesheetContent += "@font-face{";
  782. let stylesContent = "";
  783. Object.keys(fontData).forEach(fontStyle => {
  784. if (stylesContent) {
  785. stylesContent += ";";
  786. }
  787. stylesContent += fontStyle + ":" + fontData[fontStyle];
  788. });
  789. stylesheetContent += stylesContent + "}";
  790. }
  791. });
  792. if (stylesheetContent) {
  793. const styleElement = this.doc.createElement("style");
  794. styleElement.textContent = stylesheetContent;
  795. const existingStyleElement = this.doc.querySelector("style");
  796. if (existingStyleElement) {
  797. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  798. } else {
  799. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  800. }
  801. }
  802. }
  803. }
  804. removeHiddenElements() {
  805. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  806. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  807. this.stats.set("discarded", "hidden elements", removedElements.length);
  808. this.stats.set("processed", "hidden elements", removedElements.length);
  809. if (hiddenElements.length) {
  810. const styleElement = this.doc.createElement("style");
  811. styleElement.textContent = ".sf-hidden{display:none!important;}";
  812. this.doc.head.appendChild(styleElement);
  813. hiddenElements.forEach(element => {
  814. if (element.style.getPropertyValue("display") != "none") {
  815. if (element.style.getPropertyPriority("display") == "important") {
  816. element.style.setProperty("display", "none", "important");
  817. } else {
  818. element.classList.add("sf-hidden");
  819. }
  820. }
  821. });
  822. }
  823. removedElements.forEach(element => element.remove());
  824. }
  825. resolveHrefs() {
  826. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  827. const href = element.getAttribute("href").trim();
  828. if (!testIgnoredPath(href)) {
  829. let resolvedURL;
  830. try {
  831. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  832. } catch (error) {
  833. // ignored
  834. }
  835. if (resolvedURL) {
  836. const url = normalizeURL(this.options.url);
  837. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  838. resolvedURL = resolvedURL.substring(url.length);
  839. }
  840. try {
  841. element.setAttribute("href", resolvedURL);
  842. } catch (error) {
  843. // ignored
  844. }
  845. }
  846. }
  847. });
  848. }
  849. resolveStyleAttributeURLs() {
  850. this.doc.querySelectorAll("[style]").forEach(element => {
  851. let styleContent = element.getAttribute("style");
  852. if (this.options.compressCSS) {
  853. styleContent = util.compressCSS(styleContent);
  854. }
  855. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement);
  856. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  857. this.styles.set(element, declarationList);
  858. });
  859. }
  860. async resolveStylesheetURLs() {
  861. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]")).map(async element => {
  862. const options = {
  863. maxResourceSize: this.options.maxResourceSize,
  864. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  865. url: this.options.url,
  866. charset: this.charset,
  867. compressCSS: this.options.compressCSS,
  868. updatedResources: this.options.updatedResources,
  869. rootDocument: this.options.rootDocument,
  870. frameId: this.options.windowId,
  871. resourceReferrer: this.options.resourceReferrer
  872. };
  873. let mediaText;
  874. if (element.media) {
  875. mediaText = element.media.toLowerCase();
  876. }
  877. const stylesheetInfo = { mediaText };
  878. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  879. stylesheetInfo.scoped = true;
  880. }
  881. if (element.tagName == "LINK" && element.charset) {
  882. options.charset = element.charset;
  883. }
  884. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  885. }));
  886. if (this.options.rootDocument) {
  887. const newResources = Object.keys(this.options.updatedResources).filter(url => this.options.updatedResources[url].type == "stylesheet" && !this.options.updatedResources[url].retrieved).map(url => this.options.updatedResources[url]);
  888. await Promise.all(newResources.map(async resource => {
  889. resource.retrieved = true;
  890. const stylesheetInfo = {};
  891. const element = this.doc.createElement("style");
  892. this.doc.body.appendChild(element);
  893. element.textContent = resource.content;
  894. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, this.options, this.workStyleElement);
  895. }));
  896. }
  897. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  898. stylesheets.set(element, stylesheetInfo);
  899. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  900. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  901. options = Object.assign({}, options, { charset: getCharset(stylesheetContent) });
  902. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  903. }
  904. let stylesheet;
  905. try {
  906. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  907. } catch (error) {
  908. // ignored
  909. }
  910. if (stylesheet && stylesheet.children) {
  911. if (options.compressCSS) {
  912. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  913. }
  914. stylesheetInfo.stylesheet = stylesheet;
  915. } else {
  916. stylesheets.delete(element);
  917. }
  918. }
  919. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  920. let content;
  921. if (element.tagName == "LINK") {
  922. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  923. } else {
  924. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  925. }
  926. return content || "";
  927. }
  928. }
  929. async resolveFrameURLs() {
  930. if (!this.options.saveRawPage) {
  931. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  932. await Promise.all(frameElements.map(async frameElement => {
  933. if (frameElement.tagName == "OBJECT") {
  934. frameElement.setAttribute("data", "data:text/html,");
  935. } else {
  936. frameElement.removeAttribute("src");
  937. frameElement.removeAttribute("srcdoc");
  938. }
  939. Array.from(frameElement.childNodes).forEach(node => node.remove());
  940. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  941. if (this.options.frames && frameWindowId) {
  942. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  943. if (frameData) {
  944. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  945. }
  946. }
  947. }));
  948. }
  949. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  950. options.insertSingleFileComment = false;
  951. options.insertCanonicalLink = false;
  952. options.insertMetaNoIndex = false;
  953. options.saveFavicon = false;
  954. options.url = frameData.baseURI;
  955. options.windowId = frameWindowId;
  956. if (frameData.content) {
  957. options.content = frameData.content;
  958. options.canvases = frameData.canvases;
  959. options.fonts = frameData.fonts;
  960. options.stylesheets = frameData.stylesheets;
  961. options.images = frameData.images;
  962. options.posters = frameData.posters;
  963. options.usedFonts = frameData.usedFonts;
  964. options.shadowRoots = frameData.shadowRoots;
  965. options.imports = frameData.imports;
  966. frameData.runner = new Runner(options);
  967. frameData.frameElement = frameElement;
  968. await frameData.runner.loadPage();
  969. await frameData.runner.initialize();
  970. frameData.maxResources = batchRequest.getMaxResources();
  971. }
  972. }
  973. }
  974. insertShadowRootContents() {
  975. const doc = this.doc;
  976. const options = this.options;
  977. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  978. processElement(this.doc);
  979. if (this.options.removeScripts) {
  980. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  981. }
  982. const scriptElement = doc.createElement("script");
  983. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  984. 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) {try {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()} catch (error) {} if (shadowRoot) {processNode(shadowRoot)}}})}})()`;
  985. doc.body.appendChild(scriptElement);
  986. }
  987. function processElement(element) {
  988. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  989. shadowRootElements.forEach(element => {
  990. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  991. if (attributeValue) {
  992. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  993. if (shadowRootData) {
  994. const templateElement = doc.createElement("template");
  995. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  996. if (shadowRootData.delegatesFocus) {
  997. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  998. }
  999. if (shadowRootData.adoptedStyleSheets) {
  1000. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  1001. const styleElement = doc.createElement("style");
  1002. styleElement.textContent = stylesheetContent;
  1003. templateElement.appendChild(styleElement);
  1004. });
  1005. }
  1006. const shadowDoc = util.parseDocContent(shadowRootData.content);
  1007. if (shadowDoc.head) {
  1008. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  1009. if (metaCharset) {
  1010. metaCharset.remove();
  1011. }
  1012. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1013. }
  1014. if (shadowDoc.body) {
  1015. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1016. }
  1017. processElement(templateElement);
  1018. if (element.firstChild) {
  1019. element.insertBefore(templateElement, element.firstChild);
  1020. } else {
  1021. element.appendChild(templateElement);
  1022. }
  1023. }
  1024. }
  1025. });
  1026. }
  1027. }
  1028. async resolveHtmlImportURLs() {
  1029. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1030. await Promise.all(linkElements.map(async linkElement => {
  1031. const resourceURL = linkElement.href;
  1032. linkElement.removeAttribute("href");
  1033. const options = Object.create(this.options);
  1034. options.insertSingleFileComment = false;
  1035. options.insertCanonicalLink = false;
  1036. options.insertMetaNoIndex = false;
  1037. options.saveFavicon = false;
  1038. options.removeUnusedStyles = false;
  1039. options.removeAlternativeMedias = false;
  1040. options.removeUnusedFonts = false;
  1041. options.removeHiddenElements = false;
  1042. options.url = resourceURL;
  1043. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1044. if (attributeValue) {
  1045. const importData = options.imports[Number(attributeValue)];
  1046. if (importData) {
  1047. options.content = importData.content;
  1048. importData.runner = new Runner(options);
  1049. await importData.runner.loadPage();
  1050. await importData.runner.initialize();
  1051. if (!options.removeImports) {
  1052. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1053. }
  1054. importData.runner.getStyleSheets().forEach(stylesheet => {
  1055. const importedStyleElement = this.doc.createElement("style");
  1056. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1057. this.stylesheets.set(importedStyleElement, stylesheet);
  1058. });
  1059. }
  1060. }
  1061. if (options.removeImports) {
  1062. linkElement.remove();
  1063. this.stats.add("discarded", "HTML imports", 1);
  1064. }
  1065. }));
  1066. }
  1067. removeUnusedStyles() {
  1068. if (!this.mediaAllInfo) {
  1069. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1070. }
  1071. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1072. this.stats.set("processed", "CSS rules", stats.processed);
  1073. this.stats.set("discarded", "CSS rules", stats.discarded);
  1074. }
  1075. removeUnusedFonts() {
  1076. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1077. }
  1078. removeAlternativeMedias() {
  1079. const stats = util.minifyMedias(this.stylesheets);
  1080. this.stats.set("processed", "medias", stats.processed);
  1081. this.stats.set("discarded", "medias", stats.discarded);
  1082. }
  1083. async processStylesheets() {
  1084. this.options.fontURLs = new Map();
  1085. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1086. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1087. ));
  1088. }
  1089. async processStyleAttributes() {
  1090. return Promise.all([...this.styles].map(([, declarationList]) =>
  1091. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1092. ));
  1093. }
  1094. async processPageResources() {
  1095. const processAttributeArgs = [
  1096. ["link[href][rel*=\"icon\"]", "href", false, true],
  1097. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1098. ["img[src], input[src][type=image]", "src", true],
  1099. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1100. ["video[poster]", "poster"],
  1101. ["*[background]", "background"],
  1102. ["image", "xlink:href"],
  1103. ["image", "href"]
  1104. ];
  1105. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1106. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1107. );
  1108. resourcePromises = resourcePromises.concat([
  1109. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1110. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1111. ]);
  1112. if (!this.options.removeAudioSrc) {
  1113. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1114. }
  1115. if (!this.options.removeVideoSrc) {
  1116. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1117. }
  1118. await Promise.all(resourcePromises);
  1119. if (this.options.saveFavicon) {
  1120. ProcessorHelper.processShortcutIcons(this.doc);
  1121. }
  1122. }
  1123. async processScripts() {
  1124. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1125. let resourceURL;
  1126. let scriptSrc;
  1127. if (element.tagName == "SCRIPT") {
  1128. scriptSrc = element.getAttribute("src");
  1129. element.removeAttribute("src");
  1130. } else {
  1131. scriptSrc = element.getAttribute("href");
  1132. element.removeAttribute("href");
  1133. }
  1134. element.removeAttribute("integrity");
  1135. element.textContent = "";
  1136. try {
  1137. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1138. } catch (error) {
  1139. // ignored
  1140. }
  1141. if (testValidURL(resourceURL)) {
  1142. this.stats.add("processed", "scripts", 1);
  1143. const content = await util.getContent(resourceURL, {
  1144. asBinary: true,
  1145. charset: this.charset != UTF8_CHARSET && this.charset,
  1146. maxResourceSize: this.options.maxResourceSize,
  1147. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1148. frameId: this.options.windowId,
  1149. resourceReferrer: this.options.resourceReferrer
  1150. });
  1151. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1152. if (element.tagName == "SCRIPT") {
  1153. element.setAttribute("src", content.data);
  1154. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1155. element.setAttribute("async", "");
  1156. }
  1157. } else {
  1158. const scriptElement = this.doc.createElement("script");
  1159. scriptElement.setAttribute("src", content.data);
  1160. scriptElement.setAttribute("async", "");
  1161. element.parentElement.replaceChild(scriptElement, element);
  1162. }
  1163. }
  1164. }));
  1165. }
  1166. removeAlternativeImages() {
  1167. util.removeAlternativeImages(this.doc);
  1168. }
  1169. async removeAlternativeFonts() {
  1170. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1171. }
  1172. async processFrames() {
  1173. if (this.options.frames) {
  1174. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1175. await Promise.all(frameElements.map(async frameElement => {
  1176. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1177. if (frameWindowId) {
  1178. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1179. if (frameData) {
  1180. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1181. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1182. this.stats.add("processed", "frames", 1);
  1183. await frameData.runner.run();
  1184. const pageData = await frameData.runner.getPageData();
  1185. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1186. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1187. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1188. sandbox += " allow-scripts allow-same-origin";
  1189. }
  1190. frameElement.setAttribute("sandbox", sandbox);
  1191. if (frameElement.tagName == "OBJECT") {
  1192. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1193. } else {
  1194. if (frameElement.tagName == "FRAME") {
  1195. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1196. } else {
  1197. frameElement.setAttribute("srcdoc", pageData.content);
  1198. frameElement.removeAttribute("src");
  1199. }
  1200. }
  1201. this.stats.addAll(pageData);
  1202. } else {
  1203. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1204. this.stats.add("discarded", "frames", 1);
  1205. }
  1206. }
  1207. }
  1208. }));
  1209. }
  1210. }
  1211. async processHtmlImports() {
  1212. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1213. await Promise.all(linkElements.map(async linkElement => {
  1214. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1215. if (attributeValue) {
  1216. const importData = this.options.imports[Number(attributeValue)];
  1217. if (importData.runner) {
  1218. this.stats.add("processed", "HTML imports", 1);
  1219. await importData.runner.run();
  1220. const pageData = await importData.runner.getPageData();
  1221. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1222. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1223. this.stats.addAll(pageData);
  1224. } else {
  1225. this.stats.add("discarded", "HTML imports", 1);
  1226. }
  1227. }
  1228. }));
  1229. }
  1230. replaceStylesheets() {
  1231. this.doc.querySelectorAll("style").forEach(styleElement => {
  1232. const stylesheetInfo = this.stylesheets.get(styleElement);
  1233. if (stylesheetInfo) {
  1234. this.stylesheets.delete(styleElement);
  1235. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1236. styleElement.textContent = stylesheetContent;
  1237. if (stylesheetInfo.mediaText) {
  1238. styleElement.media = stylesheetInfo.mediaText;
  1239. }
  1240. } else {
  1241. styleElement.remove();
  1242. }
  1243. });
  1244. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1245. const stylesheetInfo = this.stylesheets.get(linkElement);
  1246. if (stylesheetInfo) {
  1247. this.stylesheets.delete(linkElement);
  1248. const styleElement = this.doc.createElement("style");
  1249. if (stylesheetInfo.mediaText) {
  1250. styleElement.media = stylesheetInfo.mediaText;
  1251. }
  1252. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1253. styleElement.textContent = stylesheetContent;
  1254. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1255. } else {
  1256. linkElement.remove();
  1257. }
  1258. });
  1259. }
  1260. replaceStyleAttributes() {
  1261. this.doc.querySelectorAll("[style]").forEach(element => {
  1262. const declarations = this.styles.get(element);
  1263. if (declarations) {
  1264. this.styles.delete(element);
  1265. let styleContent = cssTree.generate(declarations);
  1266. element.setAttribute("style", styleContent);
  1267. } else {
  1268. element.setAttribute("style", "");
  1269. }
  1270. });
  1271. }
  1272. insertVariables() {
  1273. if (this.cssVariables.size) {
  1274. const styleElement = this.doc.createElement("style");
  1275. const firstStyleElement = this.doc.head.querySelector("style");
  1276. if (firstStyleElement) {
  1277. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1278. } else {
  1279. this.doc.head.appendChild(styleElement);
  1280. }
  1281. let stylesheetContent = "";
  1282. this.cssVariables.forEach((content, indexResource) => {
  1283. this.cssVariables.delete(indexResource);
  1284. if (stylesheetContent) {
  1285. stylesheetContent += ";";
  1286. }
  1287. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1288. });
  1289. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1290. }
  1291. }
  1292. compressHTML() {
  1293. let size;
  1294. if (this.options.displayStats) {
  1295. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1296. }
  1297. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1298. if (this.options.displayStats) {
  1299. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1300. }
  1301. }
  1302. cleanupPage() {
  1303. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1304. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1305. if (metaCharset) {
  1306. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1307. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1308. this.doc.head.querySelector("meta[charset]").remove();
  1309. }
  1310. }
  1311. }
  1312. resetZoomLevel() {
  1313. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1314. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1315. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1316. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1317. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1318. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1319. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1320. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1321. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1322. this.doc.documentElement.style.removeProperty("-sf-transform");
  1323. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1324. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1325. }
  1326. async insertMAFFMetaData() {
  1327. const maffMetaData = await this.maffMetaDataPromise;
  1328. if (maffMetaData && maffMetaData.content) {
  1329. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1330. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1331. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1332. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1333. if (originalURLElement) {
  1334. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1335. }
  1336. if (archiveTimeElement) {
  1337. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1338. if (value) {
  1339. const date = new Date(value);
  1340. if (!isNaN(date.getTime())) {
  1341. this.options.saveDate = new Date(value);
  1342. }
  1343. }
  1344. }
  1345. }
  1346. }
  1347. async setDocInfo() {
  1348. const titleElement = this.doc.querySelector("title");
  1349. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1350. const authorElement = this.doc.querySelector("meta[name=author]");
  1351. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1352. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1353. const headingElement = this.doc.querySelector("h1");
  1354. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1355. this.options.info = {
  1356. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1357. lang: this.doc.documentElement.lang,
  1358. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1359. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1360. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1361. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1362. };
  1363. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1364. }
  1365. }
  1366. // ---------------
  1367. // ProcessorHelper
  1368. // ---------------
  1369. const DATA_URI_PREFIX = "data:";
  1370. const ABOUT_BLANK_URI = "about:blank";
  1371. const EMPTY_DATA_URI = "data:null;base64,";
  1372. const REGEXP_URL_HASH = /(#.+?)$/;
  1373. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1374. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1375. class ProcessorHelper {
  1376. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1377. const url = util.parseURL(options.saveUrl);
  1378. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1385. await evalDate(options.saveDate);
  1386. await evalDate(options.visitDate, "visit-");
  1387. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1388. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1397. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1398. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1400. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1401. for (const [name, value] of params) {
  1402. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1403. }
  1404. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1405. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1406. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1407. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1408. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1409. if (content) {
  1410. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1411. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1412. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1413. }
  1414. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1415. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1416. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1417. template = await evalTemplateVariable(template, "profile-name", () => options.profileName, dontReplaceSlash, options.filenameReplacementCharacter);
  1418. return template.trim();
  1419. function decode(value) {
  1420. try {
  1421. return decodeURI(value);
  1422. } catch (error) {
  1423. return value;
  1424. }
  1425. }
  1426. async function evalDate(date, prefix = "") {
  1427. if (date) {
  1428. template = await evalTemplateVariable(template, prefix + "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1429. template = await evalTemplateVariable(template, prefix + "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1430. template = await evalTemplateVariable(template, prefix + "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1431. template = await evalTemplateVariable(template, prefix + "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1432. template = await evalTemplateVariable(template, prefix + "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1433. template = await evalTemplateVariable(template, prefix + "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1434. template = await evalTemplateVariable(template, prefix + "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1435. template = await evalTemplateVariable(template, prefix + "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1436. template = await evalTemplateVariable(template, prefix + "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1437. template = await evalTemplateVariable(template, prefix + "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1438. template = await evalTemplateVariable(template, prefix + "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1439. template = await evalTemplateVariable(template, prefix + "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1440. template = await evalTemplateVariable(template, prefix + "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1441. template = await evalTemplateVariable(template, prefix + "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1442. template = await evalTemplateVariable(template, prefix + "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1443. template = await evalTemplateVariable(template, prefix + "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1444. template = await evalTemplateVariable(template, prefix + "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1445. template = await evalTemplateVariable(template, prefix + "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1446. template = await evalTemplateVariable(template, prefix + "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1447. template = await evalTemplateVariable(template, prefix + "time-ms", () => String(date.getTime()), dontReplaceSlash, options.filenameReplacementCharacter);
  1448. }
  1449. }
  1450. }
  1451. static setBackgroundImage(element, url, style) {
  1452. element.style.setProperty("background-blend-mode", "normal", "important");
  1453. element.style.setProperty("background-clip", "content-box", "important");
  1454. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1455. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1456. element.style.setProperty("background-image", url, "important");
  1457. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1458. element.style.setProperty("background-origin", "content-box", "important");
  1459. element.style.setProperty("background-repeat", "no-repeat", "important");
  1460. }
  1461. static processShortcutIcons(doc) {
  1462. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1463. if (!shortcutIcon) {
  1464. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1465. if (shortcutIcon) {
  1466. shortcutIcon.rel = "icon";
  1467. }
  1468. }
  1469. if (shortcutIcon) {
  1470. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1471. if (linkElement != shortcutIcon) {
  1472. linkElement.remove();
  1473. }
  1474. });
  1475. }
  1476. }
  1477. static removeSingleLineCssComments(stylesheet) {
  1478. const removedRules = [];
  1479. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1480. const ruleData = cssRule.data;
  1481. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1482. removedRules.push(cssRule);
  1483. }
  1484. }
  1485. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1486. }
  1487. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1488. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet);
  1489. const imports = getImportFunctions(stylesheetContent);
  1490. await Promise.all(imports.map(async cssImport => {
  1491. const match = matchImport(cssImport);
  1492. if (match) {
  1493. const regExpCssImport = getRegExp(cssImport);
  1494. let resourceURL = normalizeURL(match.resourceURL);
  1495. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1496. try {
  1497. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1498. } catch (error) {
  1499. // ignored
  1500. }
  1501. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1502. const content = await getStylesheetContent(resourceURL);
  1503. resourceURL = content.resourceURL;
  1504. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1505. let importedStylesheetContent = removeCssComments(content.data);
  1506. if (options.compressCSS) {
  1507. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1508. }
  1509. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1510. if (stylesheetContent.includes(cssImport)) {
  1511. const ancestorStyleSheets = new Set(importedStyleSheets);
  1512. ancestorStyleSheets.add(resourceURL);
  1513. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1514. workStylesheet.textContent = importedStylesheetContent;
  1515. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1516. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1517. } else {
  1518. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1519. }
  1520. }
  1521. } else {
  1522. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1523. }
  1524. } else {
  1525. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1526. }
  1527. }
  1528. }));
  1529. return stylesheetContent;
  1530. async function getStylesheetContent(resourceURL) {
  1531. const content = await util.getContent(resourceURL, {
  1532. maxResourceSize: options.maxResourceSize,
  1533. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1534. validateTextContentType: true,
  1535. frameId: options.frameId,
  1536. charset: options.charset,
  1537. resourceReferrer: options.resourceReferrer
  1538. });
  1539. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1540. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1541. return getStylesheetContent(resourceURL);
  1542. } else {
  1543. return content;
  1544. }
  1545. }
  1546. }
  1547. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet) {
  1548. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1549. urlFunctions.map(urlFunction => {
  1550. const originalResourceURL = matchURL(urlFunction);
  1551. let resourceURL = normalizeURL(originalResourceURL);
  1552. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1553. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1554. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1555. }
  1556. if (!testIgnoredPath(resourceURL)) {
  1557. if (!resourceURL || testValidPath(resourceURL)) {
  1558. let resolvedURL;
  1559. if (!originalResourceURL.startsWith("#")) {
  1560. try {
  1561. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1562. } catch (error) {
  1563. // ignored
  1564. }
  1565. }
  1566. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1567. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1568. }
  1569. } else {
  1570. let newUrlFunction;
  1571. if (originalResourceURL) {
  1572. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1573. } else {
  1574. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1575. }
  1576. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1577. }
  1578. }
  1579. });
  1580. return stylesheetContent;
  1581. }
  1582. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1583. resourceURL = normalizeURL(resourceURL);
  1584. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1585. const content = await util.getContent(resourceURL, {
  1586. maxResourceSize: options.maxResourceSize,
  1587. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1588. charset: options.charset,
  1589. frameId: options.frameId,
  1590. resourceReferrer: options.resourceReferrer,
  1591. validateTextContentType: true
  1592. });
  1593. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1594. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1595. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1596. }
  1597. resourceURL = content.resourceURL;
  1598. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1599. let stylesheetContent = removeCssComments(content.data);
  1600. if (options.compressCSS) {
  1601. stylesheetContent = util.compressCSS(stylesheetContent);
  1602. }
  1603. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1604. return stylesheetContent;
  1605. }
  1606. }
  1607. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1608. const promises = [];
  1609. const removedRules = [];
  1610. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1611. const ruleData = cssRule.data;
  1612. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1613. removedRules.push(cssRule);
  1614. } else if (ruleData.block && ruleData.block.children) {
  1615. if (ruleData.type == "Rule") {
  1616. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1617. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1618. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1619. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1620. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1621. }
  1622. }
  1623. }
  1624. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1625. await Promise.all(promises);
  1626. async function processFontFaceRule(ruleData, fontURLs) {
  1627. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1628. if (declaration.type == "Declaration" && declaration.value.children) {
  1629. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1630. await Promise.all(urlFunctions.map(async urlFunction => {
  1631. const originalResourceURL = matchURL(urlFunction);
  1632. const resourceURL = normalizeURL(originalResourceURL);
  1633. if (!testIgnoredPath(resourceURL)) {
  1634. if (testValidURL(resourceURL)) {
  1635. let { content } = await batchRequest.addURL(resourceURL, true, "font");
  1636. let resourceURLs = fontURLs.get(declaration);
  1637. if (!resourceURLs) {
  1638. resourceURLs = [];
  1639. fontURLs.set(declaration, resourceURLs);
  1640. }
  1641. resourceURLs.push(resourceURL);
  1642. replaceURLs(declaration, originalResourceURL, content);
  1643. }
  1644. }
  1645. }));
  1646. }
  1647. }));
  1648. function replaceURLs(declaration, oldURL, newURL) {
  1649. declaration.value.children.forEach(token => {
  1650. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1651. token.value.value = newURL;
  1652. }
  1653. });
  1654. }
  1655. }
  1656. }
  1657. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1658. await Promise.all(declarations.map(async declaration => {
  1659. if (declaration.value && !declaration.value.children && declaration.value.type == "Raw") {
  1660. try {
  1661. declaration.value = cssTree.parse(declaration.value.value, { context: "value" });
  1662. } catch (error) {
  1663. // ignored
  1664. }
  1665. }
  1666. if (declaration.type == "Declaration" && declaration.value.children) {
  1667. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1668. await Promise.all(urlFunctions.map(async urlFunction => {
  1669. const originalResourceURL = matchURL(urlFunction);
  1670. const resourceURL = normalizeURL(originalResourceURL);
  1671. if (!testIgnoredPath(resourceURL)) {
  1672. if (testValidURL(resourceURL)) {
  1673. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", true);
  1674. let variableDefined;
  1675. const tokens = [];
  1676. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1677. if (!originalResourceURL.startsWith("#")) {
  1678. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1679. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1680. tokens.push({ parent, token, value });
  1681. variableDefined = true;
  1682. } else {
  1683. token.data.value.value = content;
  1684. }
  1685. }
  1686. });
  1687. if (variableDefined) {
  1688. cssVariables.set(indexResource, content);
  1689. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1690. }
  1691. }
  1692. }
  1693. }));
  1694. }
  1695. }));
  1696. function findURLToken(url, children, callback, depth = 0) {
  1697. for (let token = children.head; token; token = token.next) {
  1698. if (token.data.children) {
  1699. findURLToken(url, token.data.children, callback, depth + 1);
  1700. }
  1701. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1702. callback(token, children, depth == 0);
  1703. }
  1704. }
  1705. }
  1706. }
  1707. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1708. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1709. let resourceURL = resourceElement.getAttribute(attributeName);
  1710. if (resourceURL != null) {
  1711. resourceURL = normalizeURL(resourceURL);
  1712. let originURL = resourceElement.dataset.singleFileOriginURL;
  1713. delete resourceElement.dataset.singleFileOriginURL;
  1714. if (!testIgnoredPath(resourceURL)) {
  1715. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1716. if (testValidPath(resourceURL)) {
  1717. try {
  1718. resourceURL = util.resolveURL(resourceURL, baseURI);
  1719. } catch (error) {
  1720. // ignored
  1721. }
  1722. if (testValidURL(resourceURL)) {
  1723. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", resourceElement.tagName == "IMG" && attributeName == "src");
  1724. if (originURL) {
  1725. if (content == EMPTY_DATA_URI) {
  1726. try {
  1727. originURL = util.resolveURL(originURL, baseURI);
  1728. } catch (error) {
  1729. // ignored
  1730. }
  1731. try {
  1732. resourceURL = originURL;
  1733. content = (await util.getContent(resourceURL, {
  1734. asBinary: true,
  1735. expectedType: "image",
  1736. maxResourceSize: options.maxResourceSize,
  1737. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1738. frameId: options.windowId,
  1739. resourceReferrer: options.resourceReferrer
  1740. })).data;
  1741. } catch (error) {
  1742. // ignored
  1743. }
  1744. }
  1745. }
  1746. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1747. resourceElement.remove();
  1748. } else {
  1749. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1750. if (!forbiddenPrefixFound) {
  1751. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1752. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1753. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1754. cssVariables.set(indexResource, content);
  1755. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1756. styles.set(resourceElement, declarationList);
  1757. } else {
  1758. resourceElement.setAttribute(attributeName, content);
  1759. }
  1760. } else {
  1761. resourceElement.setAttribute(attributeName, content);
  1762. }
  1763. }
  1764. }
  1765. }
  1766. }
  1767. }
  1768. }
  1769. }));
  1770. }
  1771. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1772. let attributeName = "xlink:href";
  1773. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1774. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1775. if (originalResourceURL == null) {
  1776. attributeName = "href";
  1777. originalResourceURL = resourceElement.getAttribute(attributeName);
  1778. }
  1779. let resourceURL = normalizeURL(originalResourceURL);
  1780. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1781. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1782. try {
  1783. resourceURL = util.resolveURL(resourceURL, baseURI);
  1784. } catch (error) {
  1785. // ignored
  1786. }
  1787. if (testValidURL(resourceURL)) {
  1788. const { content } = await batchRequest.addURL(resourceURL);
  1789. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1790. if (hashMatch && hashMatch[0]) {
  1791. let symbolElement;
  1792. try {
  1793. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1794. } catch (error) {
  1795. // ignored
  1796. }
  1797. if (symbolElement) {
  1798. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1799. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1800. }
  1801. } else {
  1802. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1803. }
  1804. }
  1805. } else if (resourceURL == options.url) {
  1806. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1807. }
  1808. }));
  1809. }
  1810. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1811. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1812. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1813. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1814. let resourceURL = normalizeURL(srcsetValue.url);
  1815. if (!testIgnoredPath(resourceURL)) {
  1816. if (testValidPath(resourceURL)) {
  1817. try {
  1818. resourceURL = util.resolveURL(resourceURL, baseURI);
  1819. } catch (error) {
  1820. // ignored
  1821. }
  1822. if (testValidURL(resourceURL)) {
  1823. const { content } = await batchRequest.addURL(resourceURL, true, "image");
  1824. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1825. if (forbiddenPrefixFound) {
  1826. return "";
  1827. }
  1828. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1829. } else {
  1830. return "";
  1831. }
  1832. } else {
  1833. return "";
  1834. }
  1835. } else {
  1836. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1837. }
  1838. }));
  1839. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1840. }));
  1841. }
  1842. static replaceImageSource(imgElement, variableName, options) {
  1843. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1844. if (attributeValue) {
  1845. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1846. if (imageData && imageData.replaceable) {
  1847. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1848. const backgroundStyle = {};
  1849. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1850. if (backgroundSize) {
  1851. backgroundStyle["background-size"] = imageData.objectFit;
  1852. }
  1853. if (imageData.objectPosition) {
  1854. backgroundStyle["background-position"] = imageData.objectPosition;
  1855. }
  1856. if (imageData.backgroundColor) {
  1857. backgroundStyle["background-color"] = imageData.backgroundColor;
  1858. }
  1859. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1860. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1861. return true;
  1862. }
  1863. }
  1864. }
  1865. }
  1866. // ----
  1867. // Util
  1868. // ----
  1869. const BLOB_URI_PREFIX = "blob:";
  1870. const HTTP_URI_PREFIX = /^https?:\/\//;
  1871. const FILE_URI_PREFIX = /^file:\/\//;
  1872. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1873. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1874. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1875. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1876. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1877. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1878. 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;
  1879. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1880. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1881. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1882. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1883. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1884. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1885. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1886. function getUpdatedResourceContent(resourceURL, content, options) {
  1887. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1888. options.updatedResources[resourceURL].retrieved = true;
  1889. return options.updatedResources[resourceURL].content;
  1890. } else {
  1891. return content.data || "";
  1892. }
  1893. }
  1894. function normalizeURL(url) {
  1895. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1896. return url;
  1897. } else {
  1898. return url.split("#")[0];
  1899. }
  1900. }
  1901. function getCSSValue(value) {
  1902. let result = "";
  1903. try {
  1904. result = cssTree.generate(value);
  1905. } catch (error) {
  1906. // ignored
  1907. }
  1908. return result;
  1909. }
  1910. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  1911. const stylesheetCharset = getCharset(stylesheetContent);
  1912. if (stylesheetCharset) {
  1913. return stylesheetCharset == charset.toLowerCase();
  1914. } else {
  1915. return true;
  1916. }
  1917. }
  1918. function getCharset(stylesheetContent) {
  1919. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  1920. if (match && match[1]) {
  1921. return match[1].toLowerCase().trim();
  1922. }
  1923. }
  1924. function getOnEventAttributeNames(doc) {
  1925. const element = doc.createElement("div");
  1926. const attributeNames = [];
  1927. for (const propertyName in element) {
  1928. if (propertyName.startsWith("on")) {
  1929. attributeNames.push(propertyName);
  1930. }
  1931. }
  1932. attributeNames.push("onunload");
  1933. return attributeNames;
  1934. }
  1935. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1936. let maxLength;
  1937. if (template) {
  1938. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1939. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1940. if (template.match(replaceRegExp)) {
  1941. const matchedLength = template.match(replaceRegExp)[0];
  1942. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1943. if (isNaN(maxLength) || maxLength <= 0) {
  1944. maxLength = null;
  1945. }
  1946. } else {
  1947. replaceRegExp = new RegExp(regExpVariable, "g");
  1948. }
  1949. if (template.match(replaceRegExp)) {
  1950. let value = await valueGetter();
  1951. if (!dontReplaceSlash) {
  1952. value = value.replace(/\/+/g, replacementCharacter);
  1953. }
  1954. if (maxLength) {
  1955. value = await util.truncateText(value, maxLength);
  1956. }
  1957. return template.replace(replaceRegExp, value);
  1958. }
  1959. }
  1960. return template;
  1961. }
  1962. function getLastSegment(url, replacementCharacter) {
  1963. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1964. if (!lastSegment) {
  1965. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1966. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1967. }
  1968. if (!lastSegment) {
  1969. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1970. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1971. }
  1972. if (!lastSegment) {
  1973. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1974. }
  1975. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1976. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1977. lastSegment = lastSegmentMatch[1];
  1978. }
  1979. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1980. return lastSegment;
  1981. }
  1982. function getRegExp(string) {
  1983. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1984. }
  1985. function getUrlFunctions(stylesheetContent, unique) {
  1986. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1987. if (unique) {
  1988. return [...new Set(result)];
  1989. } else {
  1990. return result;
  1991. }
  1992. }
  1993. function getImportFunctions(stylesheetContent) {
  1994. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1995. }
  1996. function findShortcutIcon(shortcutIcons) {
  1997. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1998. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1999. return shortcutIcons[0];
  2000. }
  2001. function matchURL(stylesheetContent) {
  2002. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  2003. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  2004. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  2005. return match && match[1];
  2006. }
  2007. function testIgnoredPath(resourceURL) {
  2008. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  2009. }
  2010. function testValidPath(resourceURL) {
  2011. return resourceURL && !resourceURL.match(EMPTY_URL);
  2012. }
  2013. function testValidURL(resourceURL) {
  2014. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  2015. }
  2016. function matchImport(stylesheetContent) {
  2017. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  2018. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  2019. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2020. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2021. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2022. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2023. if (match) {
  2024. const [, resourceURL, media] = match;
  2025. return { resourceURL, media };
  2026. }
  2027. }
  2028. function removeCssComments(stylesheetContent) {
  2029. try {
  2030. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2031. } catch (error) {
  2032. let start, end;
  2033. do {
  2034. start = stylesheetContent.indexOf("/*");
  2035. end = stylesheetContent.indexOf("*/", start + 2);
  2036. if (start != -1 && end != -1) {
  2037. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2038. }
  2039. } while (start != -1 && end != -1);
  2040. return stylesheetContent;
  2041. }
  2042. }
  2043. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2044. if (mediaQuery) {
  2045. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2046. } else {
  2047. return stylesheetContent;
  2048. }
  2049. }
  2050. function log(...args) {
  2051. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2052. }
  2053. // -----
  2054. // Stats
  2055. // -----
  2056. const STATS_DEFAULT_VALUES = {
  2057. discarded: {
  2058. "HTML bytes": 0,
  2059. "hidden elements": 0,
  2060. "HTML imports": 0,
  2061. scripts: 0,
  2062. objects: 0,
  2063. "audio sources": 0,
  2064. "video sources": 0,
  2065. frames: 0,
  2066. "CSS rules": 0,
  2067. canvas: 0,
  2068. stylesheets: 0,
  2069. resources: 0,
  2070. medias: 0
  2071. },
  2072. processed: {
  2073. "HTML bytes": 0,
  2074. "hidden elements": 0,
  2075. "HTML imports": 0,
  2076. scripts: 0,
  2077. objects: 0,
  2078. "audio sources": 0,
  2079. "video sources": 0,
  2080. frames: 0,
  2081. "CSS rules": 0,
  2082. canvas: 0,
  2083. stylesheets: 0,
  2084. resources: 0,
  2085. medias: 0
  2086. }
  2087. };
  2088. class Stats {
  2089. constructor(options) {
  2090. this.options = options;
  2091. if (options.displayStats) {
  2092. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2093. }
  2094. }
  2095. set(type, subType, value) {
  2096. if (this.options.displayStats) {
  2097. this.data[type][subType] = value;
  2098. }
  2099. }
  2100. add(type, subType, value) {
  2101. if (this.options.displayStats) {
  2102. this.data[type][subType] += value;
  2103. }
  2104. }
  2105. addAll(pageData) {
  2106. if (this.options.displayStats) {
  2107. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2108. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2109. }
  2110. }
  2111. }
  2112. export {
  2113. getClass
  2114. };