single-file-core.js 92 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. }
  34. async run() {
  35. if (this.options.userScriptEnabled) {
  36. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  37. }
  38. this.runner = new Runner(this.options, true);
  39. await this.runner.loadPage();
  40. await this.runner.initialize();
  41. if (this.options.userScriptEnabled) {
  42. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  43. }
  44. await this.runner.run();
  45. }
  46. cancel() {
  47. this.cancelled = true;
  48. if (this.runner) {
  49. this.runner.cancel();
  50. }
  51. }
  52. getPageData() {
  53. return this.runner.getPageData();
  54. }
  55. }
  56. // -------------
  57. // ProgressEvent
  58. // -------------
  59. const PAGE_LOADING = "page-loading";
  60. const PAGE_LOADED = "page-loaded";
  61. const RESOURCES_INITIALIZING = "resource-initializing";
  62. const RESOURCES_INITIALIZED = "resources-initialized";
  63. const RESOURCE_LOADED = "resource-loaded";
  64. const PAGE_ENDED = "page-ended";
  65. const STAGE_STARTED = "stage-started";
  66. const STAGE_ENDED = "stage-ended";
  67. const STAGE_TASK_STARTED = "stage-task-started";
  68. const STAGE_TASK_ENDED = "stage-task-ended";
  69. class ProgressEvent {
  70. constructor(type, detail) {
  71. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  72. }
  73. }
  74. // ------
  75. // Runner
  76. // ------
  77. const RESOLVE_URLS_STAGE = 0;
  78. const REPLACE_DATA_STAGE = 1;
  79. const REPLACE_DOCS_STAGE = 2;
  80. const POST_PROCESS_STAGE = 3;
  81. const STAGES = [{
  82. sequential: [
  83. { action: "preProcessPage" },
  84. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  85. { action: "replaceStyleContents" },
  86. { action: "resetCharsetMeta" },
  87. { option: "saveFavicon", action: "saveFavicon" },
  88. { action: "replaceCanvasElements" },
  89. { action: "insertFonts" },
  90. { action: "insertShadowRootContents" },
  91. { action: "setInputValues" },
  92. { option: "removeScripts", action: "removeScripts" },
  93. { option: "selected", action: "removeUnselectedElements" },
  94. { option: "removeVideoSrc", action: "insertVideoPosters" },
  95. { option: "removeFrames", action: "removeFrames" },
  96. { option: "removeVideoSrc", action: "removeVideoSources" },
  97. { option: "removeAudioSrc", action: "removeAudioSources" },
  98. { action: "removeDiscardedResources" },
  99. { option: "removeHiddenElements", action: "removeHiddenElements" },
  100. { action: "resolveHrefs" },
  101. { action: "resolveStyleAttributeURLs" }
  102. ],
  103. parallel: [
  104. { action: "resolveStylesheetURLs" },
  105. { option: "!removeFrames", action: "resolveFrameURLs" },
  106. { action: "resolveHtmlImportURLs" }
  107. ]
  108. }, {
  109. sequential: [
  110. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  111. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  112. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  113. ],
  114. parallel: [
  115. { action: "processStylesheets" },
  116. { action: "processStyleAttributes" },
  117. { action: "processPageResources" },
  118. { option: "!removeScripts", action: "processScripts" }
  119. ]
  120. }, {
  121. sequential: [
  122. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  123. ],
  124. parallel: [
  125. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  126. { option: "!removeFrames", action: "processFrames" },
  127. { option: "!removeImports", action: "processHtmlImports" },
  128. ]
  129. }, {
  130. sequential: [
  131. { action: "replaceStylesheets" },
  132. { action: "replaceStyleAttributes" },
  133. { action: "insertVariables" },
  134. { option: "compressHTML", action: "compressHTML" },
  135. { action: "cleanupPage" }
  136. ],
  137. parallel: [
  138. { option: "enableMaff", action: "insertMAFFMetaData" },
  139. { action: "setDocInfo" }
  140. ]
  141. }];
  142. class Runner {
  143. constructor(options, root) {
  144. const rootDocDefined = root && options.doc;
  145. this.root = root;
  146. this.options = options;
  147. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  148. const matchResourceReferrer = this.options.url.match(/^.*\//);
  149. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  150. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  151. this.options.rootDocument = root;
  152. this.options.updatedResources = this.options.updatedResources || {};
  153. this.options.fontTests = new Map();
  154. this.batchRequest = new BatchRequest();
  155. this.processor = new Processor(options, this.batchRequest);
  156. if (rootDocDefined) {
  157. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  158. this.options.canvases = docData.canvases;
  159. this.options.fonts = docData.fonts;
  160. this.options.stylesheets = docData.stylesheets;
  161. this.options.images = docData.images;
  162. this.options.posters = docData.posters;
  163. this.options.usedFonts = docData.usedFonts;
  164. this.options.shadowRoots = docData.shadowRoots;
  165. this.options.imports = docData.imports;
  166. this.options.referrer = docData.referrer;
  167. this.markedElements = docData.markedElements;
  168. }
  169. if (this.options.saveRawPage) {
  170. this.options.removeFrames = true;
  171. }
  172. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  173. this.onprogress = options.onprogress || (() => { });
  174. }
  175. async loadPage() {
  176. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  177. await this.processor.loadPage(this.options.content);
  178. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  179. }
  180. async initialize() {
  181. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  182. await this.executeStage(RESOLVE_URLS_STAGE);
  183. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  184. if (this.root && this.options.doc) {
  185. util.postProcessDoc(this.options.doc, this.markedElements);
  186. }
  187. }
  188. cancel() {
  189. this.cancelled = true;
  190. this.batchRequest.cancel();
  191. if (this.root) {
  192. if (this.options.frames) {
  193. this.options.frames.forEach(cancelRunner);
  194. }
  195. if (this.options.imports) {
  196. this.options.imports.forEach(cancelRunner);
  197. }
  198. }
  199. function cancelRunner(resourceData) {
  200. if (resourceData.runner) {
  201. resourceData.runner.cancel();
  202. }
  203. }
  204. }
  205. async run() {
  206. if (this.root) {
  207. this.processor.initialize(this.batchRequest);
  208. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  209. }
  210. await this.batchRequest.run(detail => {
  211. detail.pageURL = this.options.url;
  212. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  213. }, this.options);
  214. await this.pendingPromises;
  215. this.options.doc = null;
  216. this.options.win = null;
  217. await this.executeStage(REPLACE_DOCS_STAGE);
  218. await this.executeStage(POST_PROCESS_STAGE);
  219. this.processor.finalize();
  220. }
  221. getDocument() {
  222. return this.processor.doc;
  223. }
  224. getStyleSheets() {
  225. return this.processor.stylesheets;
  226. }
  227. getPageData() {
  228. if (this.root) {
  229. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  230. }
  231. return this.processor.getPageData();
  232. }
  233. async executeStage(step) {
  234. if (DEBUG) {
  235. log("**** STARTED STAGE", step, "****");
  236. }
  237. const frame = !this.root;
  238. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  239. STAGES[step].sequential.forEach(task => {
  240. let startTime;
  241. if (DEBUG) {
  242. startTime = Date.now();
  243. log(" -- STARTED task =", task.action);
  244. }
  245. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  246. if (!this.cancelled) {
  247. this.executeTask(task);
  248. }
  249. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  250. if (DEBUG) {
  251. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  252. }
  253. });
  254. let parallelTasksPromise;
  255. if (STAGES[step].parallel) {
  256. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  257. let startTime;
  258. if (DEBUG) {
  259. startTime = Date.now();
  260. log(" // STARTED task =", task.action);
  261. }
  262. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  263. if (!this.cancelled) {
  264. await this.executeTask(task);
  265. }
  266. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  267. if (DEBUG) {
  268. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  269. }
  270. }));
  271. } else {
  272. parallelTasksPromise = Promise.resolve();
  273. }
  274. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  275. if (DEBUG) {
  276. log("**** ENDED STAGE", step, "****");
  277. }
  278. return parallelTasksPromise;
  279. }
  280. executeTask(task) {
  281. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  282. return this.processor[task.action]();
  283. }
  284. }
  285. }
  286. // ------------
  287. // BatchRequest
  288. // ------------
  289. class BatchRequest {
  290. constructor() {
  291. this.requests = new Map();
  292. this.duplicates = new Map();
  293. }
  294. addURL(resourceURL, asBinary, expectedType, groupDuplicates) {
  295. return new Promise((resolve, reject) => {
  296. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType]);
  297. let resourceRequests = this.requests.get(requestKey);
  298. if (!resourceRequests) {
  299. resourceRequests = [];
  300. this.requests.set(requestKey, resourceRequests);
  301. }
  302. const callbacks = { resolve, reject };
  303. resourceRequests.push(callbacks);
  304. if (groupDuplicates) {
  305. let duplicateRequests = this.duplicates.get(requestKey);
  306. if (!duplicateRequests) {
  307. duplicateRequests = [];
  308. this.duplicates.set(requestKey, duplicateRequests);
  309. }
  310. duplicateRequests.push(callbacks);
  311. }
  312. });
  313. }
  314. getMaxResources() {
  315. return this.requests.size;
  316. }
  317. run(onloadListener, options) {
  318. const resourceURLs = [...this.requests.keys()];
  319. let indexResource = 0;
  320. return Promise.all(resourceURLs.map(async requestKey => {
  321. const [resourceURL, asBinary, expectedType] = JSON.parse(requestKey);
  322. const resourceRequests = this.requests.get(requestKey);
  323. try {
  324. const currentIndexResource = indexResource;
  325. indexResource = indexResource + 1;
  326. const content = await util.getContent(resourceURL, {
  327. asBinary,
  328. expectedType,
  329. maxResourceSize: options.maxResourceSize,
  330. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  331. frameId: options.windowId,
  332. resourceReferrer: options.resourceReferrer
  333. });
  334. onloadListener({ url: resourceURL });
  335. if (!this.cancelled) {
  336. resourceRequests.forEach(callbacks => {
  337. const duplicateCallbacks = this.duplicates.get(requestKey);
  338. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  339. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  340. });
  341. }
  342. } catch (error) {
  343. indexResource = indexResource + 1;
  344. onloadListener({ url: resourceURL });
  345. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  346. }
  347. this.requests.delete(requestKey);
  348. }));
  349. }
  350. cancel() {
  351. this.cancelled = true;
  352. const resourceURLs = [...this.requests.keys()];
  353. resourceURLs.forEach(requestKey => {
  354. const resourceRequests = this.requests.get(requestKey);
  355. resourceRequests.forEach(callbacks => callbacks.reject());
  356. this.requests.delete(requestKey);
  357. });
  358. }
  359. }
  360. // ---------
  361. // Processor
  362. // ---------
  363. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  364. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  365. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  366. const SCRIPT_TAG_FOUND = /<script/gi;
  367. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  368. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  369. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  370. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  371. const UTF8_CHARSET = "utf-8";
  372. class Processor {
  373. constructor(options, batchRequest) {
  374. this.options = options;
  375. this.stats = new Stats(options);
  376. this.baseURI = normalizeURL(options.baseURI || options.url);
  377. this.batchRequest = batchRequest;
  378. this.stylesheets = new Map();
  379. this.styles = new Map();
  380. this.cssVariables = new Map();
  381. this.fontTests = options.fontTests;
  382. }
  383. initialize() {
  384. this.options.saveDate = new Date();
  385. this.options.saveUrl = this.options.url;
  386. if (this.options.enableMaff) {
  387. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url));
  388. }
  389. this.maxResources = this.batchRequest.getMaxResources();
  390. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  391. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  392. }
  393. if (!this.options.removeImports && this.options.imports) {
  394. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  395. }
  396. this.stats.set("processed", "resources", this.maxResources);
  397. }
  398. async loadPage(pageContent, charset) {
  399. let content;
  400. if (!pageContent || this.options.saveRawPage) {
  401. content = await util.getContent(this.baseURI, {
  402. maxResourceSize: this.options.maxResourceSize,
  403. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  404. charset,
  405. frameId: this.options.windowId,
  406. resourceReferrer: this.options.resourceReferrer
  407. });
  408. pageContent = content.data;
  409. }
  410. this.doc = util.parseDocContent(pageContent, this.baseURI);
  411. if (this.options.saveRawPage) {
  412. let charset;
  413. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  414. const charsetDeclaration = element.content.split(";")[1];
  415. if (charsetDeclaration && !charset) {
  416. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  417. }
  418. });
  419. if (charset && charset != content.charset) {
  420. return this.loadPage(pageContent, charset);
  421. }
  422. }
  423. this.workStyleElement = this.doc.createElement("style");
  424. this.doc.body.appendChild(this.workStyleElement);
  425. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  426. }
  427. finalize() {
  428. if (this.workStyleElement.parentNode) {
  429. this.workStyleElement.remove();
  430. }
  431. }
  432. async getPageData() {
  433. util.postProcessDoc(this.doc);
  434. const url = util.parseURL(this.baseURI);
  435. if (this.options.insertSingleFileComment) {
  436. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  437. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  438. " \n url: " + this.options.saveUrl +
  439. " \n saved date: " + this.options.saveDate +
  440. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  441. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  442. }
  443. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  444. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  445. if (!canonicalLink) {
  446. canonicalLink = this.doc.createElement("link");
  447. canonicalLink.setAttribute("rel", "canonical");
  448. this.doc.head.appendChild(canonicalLink);
  449. }
  450. if (canonicalLink && !canonicalLink.href) {
  451. canonicalLink.href = this.options.saveUrl;
  452. }
  453. }
  454. if (this.options.insertMetaNoIndex) {
  455. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  456. if (!metaElement) {
  457. metaElement = this.doc.createElement("meta");
  458. metaElement.setAttribute("name", "robots");
  459. metaElement.setAttribute("content", "noindex");
  460. this.doc.head.appendChild(metaElement);
  461. }
  462. }
  463. let size;
  464. if (this.options.displayStats) {
  465. size = util.getContentSize(this.doc.documentElement.outerHTML);
  466. }
  467. const content = util.serialize(this.doc, this.options.compressHTML);
  468. if (this.options.displayStats) {
  469. const contentSize = util.getContentSize(content);
  470. this.stats.set("processed", "HTML bytes", contentSize);
  471. this.stats.add("discarded", "HTML bytes", size - contentSize);
  472. }
  473. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  474. const replacementCharacter = this.options.filenameReplacementCharacter;
  475. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  476. if (!this.options.backgroundSave) {
  477. filename = filename.replace(/\//g, replacementCharacter);
  478. }
  479. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  480. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  481. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  482. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  483. filename = filename + "…" + extension;
  484. }
  485. if (!filename) {
  486. filename = "Unnamed page";
  487. }
  488. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  489. const pageData = {
  490. stats: this.stats.data,
  491. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  492. filename,
  493. content
  494. };
  495. if (this.options.addProof) {
  496. pageData.hash = await util.digest("SHA-256", content);
  497. }
  498. if (this.options.retrieveLinks) {
  499. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  500. }
  501. return pageData;
  502. }
  503. preProcessPage() {
  504. if (this.options.win) {
  505. 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));
  506. }
  507. if (this.options.images && !this.options.saveRawPage) {
  508. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  509. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  510. if (attributeValue) {
  511. const imageData = this.options.images[Number(attributeValue)];
  512. if (imageData) {
  513. if (this.options.removeHiddenElements && (
  514. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  515. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  516. )) {
  517. imgElement.setAttribute("src", EMPTY_IMAGE);
  518. } else {
  519. if (imageData.currentSrc) {
  520. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  521. imgElement.setAttribute("src", imageData.currentSrc);
  522. }
  523. if (this.options.loadDeferredImages) {
  524. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  525. imageData.src = imgElement.dataset.src;
  526. imgElement.setAttribute("src", imgElement.dataset.src);
  527. imgElement.removeAttribute("data-src");
  528. }
  529. }
  530. }
  531. }
  532. }
  533. });
  534. if (this.options.loadDeferredImages) {
  535. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  536. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  537. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  538. imgElement.removeAttribute("data-srcset");
  539. }
  540. });
  541. }
  542. }
  543. }
  544. replaceStyleContents() {
  545. if (this.options.stylesheets) {
  546. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  547. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  548. if (attributeValue) {
  549. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  550. if (stylesheetContent) {
  551. styleElement.textContent = stylesheetContent;
  552. }
  553. }
  554. });
  555. }
  556. }
  557. removeUnselectedElements() {
  558. removeUnmarkedElements(this.doc.body);
  559. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  560. function removeUnmarkedElements(element) {
  561. let selectedElementFound = false;
  562. Array.from(element.childNodes).forEach(node => {
  563. if (node.nodeType == 1) {
  564. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  565. selectedElementFound = selectedElementFound || isSelectedElement;
  566. if (isSelectedElement) {
  567. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  568. removeUnmarkedElements(node);
  569. } else if (selectedElementFound) {
  570. removeNode(node);
  571. } else {
  572. hideNode(node);
  573. }
  574. }
  575. });
  576. }
  577. function removeNode(node) {
  578. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  579. node.remove();
  580. } else {
  581. hideNode(node);
  582. }
  583. }
  584. function hideNode(node) {
  585. if (canHideNode(node)) {
  586. node.style.setProperty("display", "none", "important");
  587. Array.from(node.childNodes).forEach(removeNode);
  588. }
  589. }
  590. function canHideNode(node) {
  591. if (node.nodeType == 1) {
  592. const tagName = node.tagName && node.tagName.toLowerCase();
  593. return (tagName != "svg" && tagName != "style" && tagName != "link");
  594. }
  595. }
  596. }
  597. insertVideoPosters() {
  598. if (this.options.posters) {
  599. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  600. let videoElement;
  601. if (element.tagName == "VIDEO") {
  602. videoElement = element;
  603. } else {
  604. videoElement = element.parentElement;
  605. }
  606. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  607. if (attributeValue) {
  608. const posterURL = this.options.posters[Number(attributeValue)];
  609. if (!videoElement.poster && posterURL) {
  610. videoElement.setAttribute("poster", posterURL);
  611. }
  612. }
  613. });
  614. }
  615. }
  616. removeFrames() {
  617. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  618. this.stats.set("discarded", "frames", frameElements.length);
  619. this.stats.set("processed", "frames", frameElements.length);
  620. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  621. }
  622. removeScripts() {
  623. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  624. this.doc.querySelectorAll("[href]").forEach(element => {
  625. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  626. element.setAttribute("href", "");
  627. }
  628. });
  629. this.doc.querySelectorAll("[src]").forEach(element => {
  630. if (element.src && element.src.match(/^\s*javascript:/)) {
  631. element.removeAttribute("src");
  632. }
  633. });
  634. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  635. this.stats.set("discarded", "scripts", scriptElements.length);
  636. this.stats.set("processed", "scripts", scriptElements.length);
  637. scriptElements.forEach(element => element.remove());
  638. }
  639. removeVideoSources() {
  640. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  641. this.stats.set("discarded", "video sources", videoSourceElements.length);
  642. this.stats.set("processed", "video sources", videoSourceElements.length);
  643. videoSourceElements.forEach(element => {
  644. if (element.tagName == "SOURCE") {
  645. element.remove();
  646. } else {
  647. videoSourceElements.forEach(element => element.removeAttribute("src"));
  648. }
  649. });
  650. }
  651. removeAudioSources() {
  652. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  653. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  654. this.stats.set("processed", "audio sources", audioSourceElements.length);
  655. audioSourceElements.forEach(element => {
  656. if (element.tagName == "SOURCE") {
  657. element.remove();
  658. } else {
  659. audioSourceElements.forEach(element => element.removeAttribute("src"));
  660. }
  661. });
  662. }
  663. removeDiscardedResources() {
  664. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  665. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  666. 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\"])");
  667. this.stats.set("discarded", "objects", objectElements.length);
  668. this.stats.set("processed", "objects", objectElements.length);
  669. objectElements.forEach(element => element.remove());
  670. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  671. replacedAttributeValue.forEach(element => {
  672. let regExp;
  673. if (this.options.removeScripts) {
  674. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  675. } else {
  676. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  677. }
  678. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  679. if (relValue.length) {
  680. element.setAttribute("rel", relValue);
  681. } else {
  682. element.remove();
  683. }
  684. });
  685. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  686. if (this.options.removeHiddenElements) {
  687. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  688. }
  689. if (!this.options.saveFavicon) {
  690. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  691. }
  692. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  693. }
  694. resetCharsetMeta() {
  695. let charset;
  696. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  697. const charsetDeclaration = element.content.split(";")[1];
  698. if (charsetDeclaration && !charset) {
  699. charset = charsetDeclaration.split("=")[1];
  700. if (charset) {
  701. this.charset = charset.trim().toLowerCase();
  702. }
  703. }
  704. element.remove();
  705. });
  706. const metaElement = this.doc.createElement("meta");
  707. metaElement.setAttribute("charset", UTF8_CHARSET);
  708. if (this.doc.head.firstChild) {
  709. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  710. } else {
  711. this.doc.head.appendChild(metaElement);
  712. }
  713. }
  714. setInputValues() {
  715. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  716. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  717. input.setAttribute("value", value || "");
  718. });
  719. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  720. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  721. if (value == "true") {
  722. input.setAttribute("checked", "");
  723. }
  724. });
  725. this.doc.querySelectorAll("textarea").forEach(textarea => {
  726. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  727. textarea.textContent = value || "";
  728. });
  729. this.doc.querySelectorAll("select").forEach(select => {
  730. select.querySelectorAll("option").forEach(option => {
  731. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  732. if (selected) {
  733. option.setAttribute("selected", "");
  734. }
  735. });
  736. });
  737. }
  738. saveFavicon() {
  739. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  740. if (!faviconElement) {
  741. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  742. }
  743. if (!faviconElement) {
  744. faviconElement = this.doc.createElement("link");
  745. faviconElement.setAttribute("type", "image/x-icon");
  746. faviconElement.setAttribute("rel", "shortcut icon");
  747. faviconElement.setAttribute("href", "/favicon.ico");
  748. }
  749. this.doc.head.appendChild(faviconElement);
  750. }
  751. replaceCanvasElements() {
  752. if (this.options.canvases) {
  753. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  754. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  755. if (attributeValue) {
  756. const canvasData = this.options.canvases[Number(attributeValue)];
  757. if (canvasData) {
  758. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  759. this.stats.add("processed", "canvas", 1);
  760. }
  761. }
  762. });
  763. }
  764. }
  765. insertFonts() {
  766. if (this.options.fonts && this.options.fonts.length) {
  767. let stylesheetContent = "";
  768. this.options.fonts.forEach(fontData => {
  769. if (fontData["font-family"] && fontData.src) {
  770. stylesheetContent += "@font-face{";
  771. let stylesContent = "";
  772. Object.keys(fontData).forEach(fontStyle => {
  773. if (stylesContent) {
  774. stylesContent += ";";
  775. }
  776. stylesContent += fontStyle + ":" + fontData[fontStyle];
  777. });
  778. stylesheetContent += stylesContent + "}";
  779. }
  780. });
  781. if (stylesheetContent) {
  782. const styleElement = this.doc.createElement("style");
  783. styleElement.textContent = stylesheetContent;
  784. const existingStyleElement = this.doc.querySelector("style");
  785. if (existingStyleElement) {
  786. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  787. } else {
  788. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  789. }
  790. }
  791. }
  792. }
  793. removeHiddenElements() {
  794. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  795. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  796. this.stats.set("discarded", "hidden elements", removedElements.length);
  797. this.stats.set("processed", "hidden elements", removedElements.length);
  798. if (hiddenElements.length) {
  799. const styleElement = this.doc.createElement("style");
  800. styleElement.textContent = ".sf-hidden{display:none!important;}";
  801. this.doc.head.appendChild(styleElement);
  802. hiddenElements.forEach(element => {
  803. if (element.style.getPropertyValue("display") != "none") {
  804. if (element.style.getPropertyPriority("display") == "important") {
  805. element.style.setProperty("display", "none", "important");
  806. } else {
  807. element.classList.add("sf-hidden");
  808. }
  809. }
  810. });
  811. }
  812. removedElements.forEach(element => element.remove());
  813. }
  814. resolveHrefs() {
  815. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  816. const href = element.getAttribute("href").trim();
  817. if (!testIgnoredPath(href)) {
  818. let resolvedURL;
  819. try {
  820. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  821. } catch (error) {
  822. // ignored
  823. }
  824. if (resolvedURL) {
  825. const url = normalizeURL(this.options.url);
  826. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  827. resolvedURL = resolvedURL.substring(url.length);
  828. }
  829. try {
  830. element.setAttribute("href", resolvedURL);
  831. } catch (error) {
  832. // ignored
  833. }
  834. }
  835. }
  836. });
  837. }
  838. resolveStyleAttributeURLs() {
  839. this.doc.querySelectorAll("[style]").forEach(element => {
  840. let styleContent = element.getAttribute("style");
  841. if (this.options.compressCSS) {
  842. styleContent = util.compressCSS(styleContent);
  843. }
  844. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement);
  845. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  846. this.styles.set(element, declarationList);
  847. });
  848. }
  849. async resolveStylesheetURLs() {
  850. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]")).map(async element => {
  851. const options = {
  852. maxResourceSize: this.options.maxResourceSize,
  853. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  854. url: this.options.url,
  855. charset: this.charset,
  856. compressCSS: this.options.compressCSS,
  857. updatedResources: this.options.updatedResources,
  858. rootDocument: this.options.rootDocument,
  859. frameId: this.options.windowId,
  860. resourceReferrer: this.options.resourceReferrer
  861. };
  862. let mediaText;
  863. if (element.media) {
  864. mediaText = element.media.toLowerCase();
  865. }
  866. const stylesheetInfo = { mediaText };
  867. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  868. stylesheetInfo.scoped = true;
  869. }
  870. if (element.tagName == "LINK" && element.charset) {
  871. options.charset = element.charset;
  872. }
  873. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  874. }));
  875. if (this.options.rootDocument) {
  876. 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]);
  877. await Promise.all(newResources.map(async resource => {
  878. resource.retrieved = true;
  879. const stylesheetInfo = {};
  880. const element = this.doc.createElement("style");
  881. this.doc.body.appendChild(element);
  882. element.textContent = resource.content;
  883. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, this.options, this.workStyleElement);
  884. }));
  885. }
  886. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  887. stylesheets.set(element, stylesheetInfo);
  888. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  889. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  890. options = Object.assign({}, options, { charset: getCharset(stylesheetContent) });
  891. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  892. }
  893. let stylesheet;
  894. try {
  895. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  896. } catch (error) {
  897. // ignored
  898. }
  899. if (stylesheet && stylesheet.children) {
  900. if (options.compressCSS) {
  901. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  902. }
  903. stylesheetInfo.stylesheet = stylesheet;
  904. } else {
  905. stylesheets.delete(element);
  906. }
  907. }
  908. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  909. let content;
  910. if (element.tagName == "LINK") {
  911. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  912. } else {
  913. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  914. }
  915. return content || "";
  916. }
  917. }
  918. async resolveFrameURLs() {
  919. if (!this.options.saveRawPage) {
  920. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  921. await Promise.all(frameElements.map(async frameElement => {
  922. if (frameElement.tagName == "OBJECT") {
  923. frameElement.setAttribute("data", "data:text/html,");
  924. } else {
  925. frameElement.removeAttribute("src");
  926. frameElement.removeAttribute("srcdoc");
  927. }
  928. Array.from(frameElement.childNodes).forEach(node => node.remove());
  929. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  930. if (this.options.frames && frameWindowId) {
  931. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  932. if (frameData) {
  933. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  934. }
  935. }
  936. }));
  937. }
  938. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  939. options.insertSingleFileComment = false;
  940. options.insertCanonicalLink = false;
  941. options.insertMetaNoIndex = false;
  942. options.saveFavicon = false;
  943. options.url = frameData.baseURI;
  944. options.windowId = frameWindowId;
  945. if (frameData.content) {
  946. options.content = frameData.content;
  947. options.canvases = frameData.canvases;
  948. options.fonts = frameData.fonts;
  949. options.stylesheets = frameData.stylesheets;
  950. options.images = frameData.images;
  951. options.posters = frameData.posters;
  952. options.usedFonts = frameData.usedFonts;
  953. options.shadowRoots = frameData.shadowRoots;
  954. options.imports = frameData.imports;
  955. frameData.runner = new Runner(options);
  956. frameData.frameElement = frameElement;
  957. await frameData.runner.loadPage();
  958. await frameData.runner.initialize();
  959. frameData.maxResources = batchRequest.getMaxResources();
  960. }
  961. }
  962. }
  963. insertShadowRootContents() {
  964. const doc = this.doc;
  965. const options = this.options;
  966. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  967. processElement(this.doc);
  968. if (this.options.removeScripts) {
  969. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  970. }
  971. const scriptElement = doc.createElement("script");
  972. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  973. 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)}}})}})()`;
  974. doc.body.appendChild(scriptElement);
  975. }
  976. function processElement(element) {
  977. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  978. shadowRootElements.forEach(element => {
  979. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  980. if (attributeValue) {
  981. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  982. if (shadowRootData) {
  983. const templateElement = doc.createElement("template");
  984. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  985. if (shadowRootData.delegatesFocus) {
  986. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  987. }
  988. if (shadowRootData.adoptedStyleSheets) {
  989. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  990. const styleElement = doc.createElement("style");
  991. styleElement.textContent = stylesheetContent;
  992. templateElement.appendChild(styleElement);
  993. });
  994. }
  995. const shadowDoc = util.parseDocContent(shadowRootData.content);
  996. if (shadowDoc.head) {
  997. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  998. if (metaCharset) {
  999. metaCharset.remove();
  1000. }
  1001. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1002. }
  1003. if (shadowDoc.body) {
  1004. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1005. }
  1006. processElement(templateElement);
  1007. if (element.firstChild) {
  1008. element.insertBefore(templateElement, element.firstChild);
  1009. } else {
  1010. element.appendChild(templateElement);
  1011. }
  1012. }
  1013. }
  1014. });
  1015. }
  1016. }
  1017. async resolveHtmlImportURLs() {
  1018. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1019. await Promise.all(linkElements.map(async linkElement => {
  1020. const resourceURL = linkElement.href;
  1021. linkElement.removeAttribute("href");
  1022. const options = Object.create(this.options);
  1023. options.insertSingleFileComment = false;
  1024. options.insertCanonicalLink = false;
  1025. options.insertMetaNoIndex = false;
  1026. options.saveFavicon = false;
  1027. options.removeUnusedStyles = false;
  1028. options.removeAlternativeMedias = false;
  1029. options.removeUnusedFonts = false;
  1030. options.removeHiddenElements = false;
  1031. options.url = resourceURL;
  1032. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1033. if (attributeValue) {
  1034. const importData = options.imports[Number(attributeValue)];
  1035. if (importData) {
  1036. options.content = importData.content;
  1037. importData.runner = new Runner(options);
  1038. await importData.runner.loadPage();
  1039. await importData.runner.initialize();
  1040. if (!options.removeImports) {
  1041. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1042. }
  1043. importData.runner.getStyleSheets().forEach(stylesheet => {
  1044. const importedStyleElement = this.doc.createElement("style");
  1045. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1046. this.stylesheets.set(importedStyleElement, stylesheet);
  1047. });
  1048. }
  1049. }
  1050. if (options.removeImports) {
  1051. linkElement.remove();
  1052. this.stats.add("discarded", "HTML imports", 1);
  1053. }
  1054. }));
  1055. }
  1056. removeUnusedStyles() {
  1057. if (!this.mediaAllInfo) {
  1058. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1059. }
  1060. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1061. this.stats.set("processed", "CSS rules", stats.processed);
  1062. this.stats.set("discarded", "CSS rules", stats.discarded);
  1063. }
  1064. removeUnusedFonts() {
  1065. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1066. }
  1067. removeAlternativeMedias() {
  1068. const stats = util.minifyMedias(this.stylesheets);
  1069. this.stats.set("processed", "medias", stats.processed);
  1070. this.stats.set("discarded", "medias", stats.discarded);
  1071. }
  1072. async processStylesheets() {
  1073. this.options.fontURLs = new Map();
  1074. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1075. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1076. ));
  1077. }
  1078. async processStyleAttributes() {
  1079. return Promise.all([...this.styles].map(([, declarationList]) =>
  1080. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1081. ));
  1082. }
  1083. async processPageResources() {
  1084. const processAttributeArgs = [
  1085. ["link[href][rel*=\"icon\"]", "href", false, true],
  1086. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1087. ["img[src], input[src][type=image]", "src", true],
  1088. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1089. ["video[poster]", "poster"],
  1090. ["*[background]", "background"],
  1091. ["image", "xlink:href"],
  1092. ["image", "href"]
  1093. ];
  1094. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1095. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1096. );
  1097. resourcePromises = resourcePromises.concat([
  1098. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1099. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1100. ]);
  1101. if (!this.options.removeAudioSrc) {
  1102. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1103. }
  1104. if (!this.options.removeVideoSrc) {
  1105. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1106. }
  1107. await Promise.all(resourcePromises);
  1108. if (this.options.saveFavicon) {
  1109. ProcessorHelper.processShortcutIcons(this.doc);
  1110. }
  1111. }
  1112. async processScripts() {
  1113. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1114. let resourceURL;
  1115. let scriptSrc;
  1116. if (element.tagName == "SCRIPT") {
  1117. scriptSrc = element.getAttribute("src");
  1118. element.removeAttribute("src");
  1119. } else {
  1120. scriptSrc = element.getAttribute("href");
  1121. element.removeAttribute("href");
  1122. }
  1123. element.removeAttribute("integrity");
  1124. element.textContent = "";
  1125. try {
  1126. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1127. } catch (error) {
  1128. // ignored
  1129. }
  1130. if (testValidURL(resourceURL)) {
  1131. this.stats.add("processed", "scripts", 1);
  1132. const content = await util.getContent(resourceURL, {
  1133. asBinary: true,
  1134. charset: this.charset != UTF8_CHARSET && this.charset,
  1135. maxResourceSize: this.options.maxResourceSize,
  1136. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1137. frameId: this.options.windowId,
  1138. resourceReferrer: this.options.resourceReferrer
  1139. });
  1140. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1141. if (element.tagName == "SCRIPT") {
  1142. element.setAttribute("src", content.data);
  1143. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1144. element.setAttribute("async", "");
  1145. }
  1146. } else {
  1147. const scriptElement = this.doc.createElement("script");
  1148. scriptElement.setAttribute("src", content.data);
  1149. scriptElement.setAttribute("async", "");
  1150. element.parentElement.replaceChild(scriptElement, element);
  1151. }
  1152. }
  1153. }));
  1154. }
  1155. removeAlternativeImages() {
  1156. util.removeAlternativeImages(this.doc);
  1157. }
  1158. async removeAlternativeFonts() {
  1159. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1160. }
  1161. async processFrames() {
  1162. if (this.options.frames) {
  1163. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1164. await Promise.all(frameElements.map(async frameElement => {
  1165. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1166. if (frameWindowId) {
  1167. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1168. if (frameData) {
  1169. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1170. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1171. this.stats.add("processed", "frames", 1);
  1172. await frameData.runner.run();
  1173. const pageData = await frameData.runner.getPageData();
  1174. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1175. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1176. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1177. sandbox += " allow-scripts allow-same-origin";
  1178. }
  1179. frameElement.setAttribute("sandbox", sandbox);
  1180. if (frameElement.tagName == "OBJECT") {
  1181. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1182. } else {
  1183. if (frameElement.tagName == "FRAME") {
  1184. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1185. } else {
  1186. frameElement.setAttribute("srcdoc", pageData.content);
  1187. frameElement.removeAttribute("src");
  1188. }
  1189. }
  1190. this.stats.addAll(pageData);
  1191. } else {
  1192. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1193. this.stats.add("discarded", "frames", 1);
  1194. }
  1195. }
  1196. }
  1197. }));
  1198. }
  1199. }
  1200. async processHtmlImports() {
  1201. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1202. await Promise.all(linkElements.map(async linkElement => {
  1203. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1204. if (attributeValue) {
  1205. const importData = this.options.imports[Number(attributeValue)];
  1206. if (importData.runner) {
  1207. this.stats.add("processed", "HTML imports", 1);
  1208. await importData.runner.run();
  1209. const pageData = await importData.runner.getPageData();
  1210. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1211. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1212. this.stats.addAll(pageData);
  1213. } else {
  1214. this.stats.add("discarded", "HTML imports", 1);
  1215. }
  1216. }
  1217. }));
  1218. }
  1219. replaceStylesheets() {
  1220. this.doc.querySelectorAll("style").forEach(styleElement => {
  1221. const stylesheetInfo = this.stylesheets.get(styleElement);
  1222. if (stylesheetInfo) {
  1223. this.stylesheets.delete(styleElement);
  1224. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1225. styleElement.textContent = stylesheetContent;
  1226. if (stylesheetInfo.mediaText) {
  1227. styleElement.media = stylesheetInfo.mediaText;
  1228. }
  1229. } else {
  1230. styleElement.remove();
  1231. }
  1232. });
  1233. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1234. const stylesheetInfo = this.stylesheets.get(linkElement);
  1235. if (stylesheetInfo) {
  1236. this.stylesheets.delete(linkElement);
  1237. const styleElement = this.doc.createElement("style");
  1238. if (stylesheetInfo.mediaText) {
  1239. styleElement.media = stylesheetInfo.mediaText;
  1240. }
  1241. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1242. styleElement.textContent = stylesheetContent;
  1243. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1244. } else {
  1245. linkElement.remove();
  1246. }
  1247. });
  1248. }
  1249. replaceStyleAttributes() {
  1250. this.doc.querySelectorAll("[style]").forEach(element => {
  1251. const declarations = this.styles.get(element);
  1252. if (declarations) {
  1253. this.styles.delete(element);
  1254. let styleContent = cssTree.generate(declarations);
  1255. element.setAttribute("style", styleContent);
  1256. } else {
  1257. element.setAttribute("style", "");
  1258. }
  1259. });
  1260. }
  1261. insertVariables() {
  1262. if (this.cssVariables.size) {
  1263. const styleElement = this.doc.createElement("style");
  1264. const firstStyleElement = this.doc.head.querySelector("style");
  1265. if (firstStyleElement) {
  1266. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1267. } else {
  1268. this.doc.head.appendChild(styleElement);
  1269. }
  1270. let stylesheetContent = "";
  1271. this.cssVariables.forEach((content, indexResource) => {
  1272. this.cssVariables.delete(indexResource);
  1273. if (stylesheetContent) {
  1274. stylesheetContent += ";";
  1275. }
  1276. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1277. });
  1278. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1279. }
  1280. }
  1281. compressHTML() {
  1282. let size;
  1283. if (this.options.displayStats) {
  1284. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1285. }
  1286. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1287. if (this.options.displayStats) {
  1288. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1289. }
  1290. }
  1291. cleanupPage() {
  1292. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1293. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1294. if (metaCharset) {
  1295. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1296. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1297. this.doc.head.querySelector("meta[charset]").remove();
  1298. }
  1299. }
  1300. }
  1301. resetZoomLevel() {
  1302. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1303. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1304. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1305. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1306. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1307. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1308. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1309. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1310. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1311. this.doc.documentElement.style.removeProperty("-sf-transform");
  1312. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1313. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1314. }
  1315. async insertMAFFMetaData() {
  1316. const maffMetaData = await this.maffMetaDataPromise;
  1317. if (maffMetaData && maffMetaData.content) {
  1318. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1319. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1320. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1321. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1322. if (originalURLElement) {
  1323. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1324. }
  1325. if (archiveTimeElement) {
  1326. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1327. if (value) {
  1328. const date = new Date(value);
  1329. if (!isNaN(date.getTime())) {
  1330. this.options.saveDate = new Date(value);
  1331. }
  1332. }
  1333. }
  1334. }
  1335. }
  1336. async setDocInfo() {
  1337. const titleElement = this.doc.querySelector("title");
  1338. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1339. const authorElement = this.doc.querySelector("meta[name=author]");
  1340. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1341. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1342. const headingElement = this.doc.querySelector("h1");
  1343. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1344. this.options.info = {
  1345. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1346. lang: this.doc.documentElement.lang,
  1347. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1348. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1349. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1350. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1351. };
  1352. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1353. }
  1354. }
  1355. // ---------------
  1356. // ProcessorHelper
  1357. // ---------------
  1358. const DATA_URI_PREFIX = "data:";
  1359. const ABOUT_BLANK_URI = "about:blank";
  1360. const EMPTY_DATA_URI = "data:null;base64,";
  1361. const REGEXP_URL_HASH = /(#.+?)$/;
  1362. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1363. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1364. class ProcessorHelper {
  1365. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1366. const date = options.saveDate;
  1367. const url = util.parseURL(options.saveUrl);
  1368. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1370. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1371. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1376. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1377. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1387. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1388. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1397. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1398. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1401. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1402. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1403. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1404. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1405. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1406. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1407. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1408. for (const [name, value] of params) {
  1409. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1410. }
  1411. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1412. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1413. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1414. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1415. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1416. if (content) {
  1417. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1418. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1419. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1420. }
  1421. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1422. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1423. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1424. return template.trim();
  1425. function decode(value) {
  1426. try {
  1427. return decodeURI(value);
  1428. } catch (error) {
  1429. return value;
  1430. }
  1431. }
  1432. }
  1433. static setBackgroundImage(element, url, style) {
  1434. element.style.setProperty("background-blend-mode", "normal", "important");
  1435. element.style.setProperty("background-clip", "content-box", "important");
  1436. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1437. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1438. element.style.setProperty("background-image", url, "important");
  1439. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1440. element.style.setProperty("background-origin", "content-box", "important");
  1441. element.style.setProperty("background-repeat", "no-repeat", "important");
  1442. }
  1443. static processShortcutIcons(doc) {
  1444. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1445. if (!shortcutIcon) {
  1446. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1447. if (shortcutIcon) {
  1448. shortcutIcon.rel = "icon";
  1449. }
  1450. }
  1451. if (shortcutIcon) {
  1452. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1453. if (linkElement != shortcutIcon) {
  1454. linkElement.remove();
  1455. }
  1456. });
  1457. }
  1458. }
  1459. static removeSingleLineCssComments(stylesheet) {
  1460. const removedRules = [];
  1461. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1462. const ruleData = cssRule.data;
  1463. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1464. removedRules.push(cssRule);
  1465. }
  1466. }
  1467. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1468. }
  1469. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1470. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet);
  1471. const imports = getImportFunctions(stylesheetContent);
  1472. await Promise.all(imports.map(async cssImport => {
  1473. const match = matchImport(cssImport);
  1474. if (match) {
  1475. const regExpCssImport = getRegExp(cssImport);
  1476. let resourceURL = normalizeURL(match.resourceURL);
  1477. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1478. try {
  1479. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1480. } catch (error) {
  1481. // ignored
  1482. }
  1483. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1484. const content = await getStylesheetContent(resourceURL);
  1485. resourceURL = content.resourceURL;
  1486. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1487. let importedStylesheetContent = removeCssComments(content.data);
  1488. if (options.compressCSS) {
  1489. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1490. }
  1491. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1492. if (stylesheetContent.includes(cssImport)) {
  1493. const ancestorStyleSheets = new Set(importedStyleSheets);
  1494. ancestorStyleSheets.add(resourceURL);
  1495. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1496. workStylesheet.textContent = importedStylesheetContent;
  1497. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1498. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1499. } else {
  1500. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1501. }
  1502. }
  1503. } else {
  1504. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1505. }
  1506. } else {
  1507. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1508. }
  1509. }
  1510. }));
  1511. return stylesheetContent;
  1512. async function getStylesheetContent(resourceURL) {
  1513. const content = await util.getContent(resourceURL, {
  1514. maxResourceSize: options.maxResourceSize,
  1515. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1516. validateTextContentType: true,
  1517. frameId: options.frameId,
  1518. charset: options.charset,
  1519. resourceReferrer: options.resourceReferrer
  1520. });
  1521. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1522. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1523. return getStylesheetContent(resourceURL);
  1524. } else {
  1525. return content;
  1526. }
  1527. }
  1528. }
  1529. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet) {
  1530. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1531. urlFunctions.map(urlFunction => {
  1532. const originalResourceURL = matchURL(urlFunction);
  1533. let resourceURL = normalizeURL(originalResourceURL);
  1534. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1535. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1536. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1537. }
  1538. if (!testIgnoredPath(resourceURL)) {
  1539. if (!resourceURL || testValidPath(resourceURL)) {
  1540. let resolvedURL;
  1541. if (!originalResourceURL.startsWith("#")) {
  1542. try {
  1543. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1544. } catch (error) {
  1545. // ignored
  1546. }
  1547. }
  1548. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1549. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1550. }
  1551. } else {
  1552. let newUrlFunction;
  1553. if (originalResourceURL) {
  1554. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1555. } else {
  1556. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1557. }
  1558. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1559. }
  1560. }
  1561. });
  1562. return stylesheetContent;
  1563. }
  1564. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1565. resourceURL = normalizeURL(resourceURL);
  1566. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1567. const content = await util.getContent(resourceURL, {
  1568. maxResourceSize: options.maxResourceSize,
  1569. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1570. charset: options.charset,
  1571. frameId: options.frameId,
  1572. resourceReferrer: options.resourceReferrer,
  1573. validateTextContentType: true
  1574. });
  1575. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1576. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1577. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1578. }
  1579. resourceURL = content.resourceURL;
  1580. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1581. let stylesheetContent = removeCssComments(content.data);
  1582. if (options.compressCSS) {
  1583. stylesheetContent = util.compressCSS(stylesheetContent);
  1584. }
  1585. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1586. return stylesheetContent;
  1587. }
  1588. }
  1589. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1590. const promises = [];
  1591. const removedRules = [];
  1592. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1593. const ruleData = cssRule.data;
  1594. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1595. removedRules.push(cssRule);
  1596. } else if (ruleData.block && ruleData.block.children) {
  1597. if (ruleData.type == "Rule") {
  1598. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1599. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1600. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1601. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1602. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1603. }
  1604. }
  1605. }
  1606. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1607. await Promise.all(promises);
  1608. async function processFontFaceRule(ruleData, fontURLs) {
  1609. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1610. if (declaration.type == "Declaration" && declaration.value.children) {
  1611. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1612. await Promise.all(urlFunctions.map(async urlFunction => {
  1613. const originalResourceURL = matchURL(urlFunction);
  1614. const resourceURL = normalizeURL(originalResourceURL);
  1615. if (!testIgnoredPath(resourceURL)) {
  1616. if (testValidURL(resourceURL)) {
  1617. let { content } = await batchRequest.addURL(resourceURL, true, "font");
  1618. let resourceURLs = fontURLs.get(declaration);
  1619. if (!resourceURLs) {
  1620. resourceURLs = [];
  1621. fontURLs.set(declaration, resourceURLs);
  1622. }
  1623. resourceURLs.push(resourceURL);
  1624. replaceURLs(declaration, originalResourceURL, content);
  1625. }
  1626. }
  1627. }));
  1628. }
  1629. }));
  1630. function replaceURLs(declaration, oldURL, newURL) {
  1631. declaration.value.children.forEach(token => {
  1632. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1633. token.value.value = newURL;
  1634. }
  1635. });
  1636. }
  1637. }
  1638. }
  1639. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1640. await Promise.all(declarations.map(async declaration => {
  1641. let children;
  1642. if (declaration.value) {
  1643. if (declaration.value.children) {
  1644. children = declaration.value.children;
  1645. } else if (declaration.value.type == "Raw") {
  1646. try {
  1647. children = cssTree.parse(declaration.value.value, { context: "value" }).children;
  1648. } catch (error) {
  1649. // ignored
  1650. }
  1651. }
  1652. }
  1653. if (declaration.type == "Declaration" && children) {
  1654. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1655. await Promise.all(urlFunctions.map(async urlFunction => {
  1656. const originalResourceURL = matchURL(urlFunction);
  1657. const resourceURL = normalizeURL(originalResourceURL);
  1658. if (!testIgnoredPath(resourceURL)) {
  1659. if (testValidURL(resourceURL)) {
  1660. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", true);
  1661. let variableDefined;
  1662. const tokens = [];
  1663. findURLToken(originalResourceURL, children, (token, parent, rootFunction) => {
  1664. if (!originalResourceURL.startsWith("#")) {
  1665. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1666. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1667. tokens.push({ parent, token, value });
  1668. variableDefined = true;
  1669. } else {
  1670. token.data.value.value = content;
  1671. }
  1672. }
  1673. });
  1674. if (variableDefined) {
  1675. cssVariables.set(indexResource, content);
  1676. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1677. }
  1678. }
  1679. }
  1680. }));
  1681. }
  1682. }));
  1683. function findURLToken(url, children, callback, depth = 0) {
  1684. for (let token = children.head; token; token = token.next) {
  1685. if (token.data.children) {
  1686. findURLToken(url, token.data.children, callback, depth + 1);
  1687. }
  1688. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1689. callback(token, children, depth == 0);
  1690. }
  1691. }
  1692. }
  1693. }
  1694. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1695. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1696. let resourceURL = resourceElement.getAttribute(attributeName);
  1697. if (resourceURL != null) {
  1698. resourceURL = normalizeURL(resourceURL);
  1699. let originURL = resourceElement.dataset.singleFileOriginURL;
  1700. delete resourceElement.dataset.singleFileOriginURL;
  1701. if (!testIgnoredPath(resourceURL)) {
  1702. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1703. if (testValidPath(resourceURL)) {
  1704. try {
  1705. resourceURL = util.resolveURL(resourceURL, baseURI);
  1706. } catch (error) {
  1707. // ignored
  1708. }
  1709. if (testValidURL(resourceURL)) {
  1710. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", resourceElement.tagName == "IMG" && attributeName == "src");
  1711. if (originURL) {
  1712. if (content == EMPTY_DATA_URI) {
  1713. try {
  1714. originURL = util.resolveURL(originURL, baseURI);
  1715. } catch (error) {
  1716. // ignored
  1717. }
  1718. try {
  1719. resourceURL = originURL;
  1720. content = (await util.getContent(resourceURL, {
  1721. asBinary: true,
  1722. expectedType: "image",
  1723. maxResourceSize: options.maxResourceSize,
  1724. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1725. frameId: options.windowId,
  1726. resourceReferrer: options.resourceReferrer
  1727. })).data;
  1728. } catch (error) {
  1729. // ignored
  1730. }
  1731. }
  1732. }
  1733. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1734. resourceElement.remove();
  1735. } else {
  1736. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1737. if (!forbiddenPrefixFound) {
  1738. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1739. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1740. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1741. cssVariables.set(indexResource, content);
  1742. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1743. styles.set(resourceElement, declarationList);
  1744. } else {
  1745. resourceElement.setAttribute(attributeName, content);
  1746. }
  1747. } else {
  1748. resourceElement.setAttribute(attributeName, content);
  1749. }
  1750. }
  1751. }
  1752. }
  1753. }
  1754. }
  1755. }
  1756. }));
  1757. }
  1758. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1759. let attributeName = "xlink:href";
  1760. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1761. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1762. if (originalResourceURL == null) {
  1763. attributeName = "href";
  1764. originalResourceURL = resourceElement.getAttribute(attributeName);
  1765. }
  1766. let resourceURL = normalizeURL(originalResourceURL);
  1767. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1768. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1769. try {
  1770. resourceURL = util.resolveURL(resourceURL, baseURI);
  1771. } catch (error) {
  1772. // ignored
  1773. }
  1774. if (testValidURL(resourceURL)) {
  1775. const { content } = await batchRequest.addURL(resourceURL);
  1776. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1777. if (hashMatch && hashMatch[0]) {
  1778. let symbolElement;
  1779. try {
  1780. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1781. } catch (error) {
  1782. // ignored
  1783. }
  1784. if (symbolElement) {
  1785. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1786. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1787. }
  1788. } else {
  1789. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1790. }
  1791. }
  1792. } else if (resourceURL == options.url) {
  1793. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1794. }
  1795. }));
  1796. }
  1797. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1798. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1799. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1800. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1801. let resourceURL = normalizeURL(srcsetValue.url);
  1802. if (!testIgnoredPath(resourceURL)) {
  1803. if (testValidPath(resourceURL)) {
  1804. try {
  1805. resourceURL = util.resolveURL(resourceURL, baseURI);
  1806. } catch (error) {
  1807. // ignored
  1808. }
  1809. if (testValidURL(resourceURL)) {
  1810. const { content } = await batchRequest.addURL(resourceURL, true, "image");
  1811. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1812. if (forbiddenPrefixFound) {
  1813. return "";
  1814. }
  1815. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1816. } else {
  1817. return "";
  1818. }
  1819. } else {
  1820. return "";
  1821. }
  1822. } else {
  1823. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1824. }
  1825. }));
  1826. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1827. }));
  1828. }
  1829. static replaceImageSource(imgElement, variableName, options) {
  1830. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1831. if (attributeValue) {
  1832. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1833. if (imageData && imageData.replaceable) {
  1834. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1835. const backgroundStyle = {};
  1836. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1837. if (backgroundSize) {
  1838. backgroundStyle["background-size"] = imageData.objectFit;
  1839. }
  1840. if (imageData.objectPosition) {
  1841. backgroundStyle["background-position"] = imageData.objectPosition;
  1842. }
  1843. if (imageData.backgroundColor) {
  1844. backgroundStyle["background-color"] = imageData.backgroundColor;
  1845. }
  1846. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1847. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1848. return true;
  1849. }
  1850. }
  1851. }
  1852. }
  1853. // ----
  1854. // Util
  1855. // ----
  1856. const BLOB_URI_PREFIX = "blob:";
  1857. const HTTP_URI_PREFIX = /^https?:\/\//;
  1858. const FILE_URI_PREFIX = /^file:\/\//;
  1859. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1860. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1861. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1862. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1863. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1864. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1865. 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;
  1866. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1867. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1868. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1869. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1870. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1871. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1872. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1873. function getUpdatedResourceContent(resourceURL, content, options) {
  1874. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1875. options.updatedResources[resourceURL].retrieved = true;
  1876. return options.updatedResources[resourceURL].content;
  1877. } else {
  1878. return content.data || "";
  1879. }
  1880. }
  1881. function normalizeURL(url) {
  1882. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1883. return url;
  1884. } else {
  1885. return url.split("#")[0];
  1886. }
  1887. }
  1888. function getCSSValue(value) {
  1889. let result = "";
  1890. try {
  1891. result = cssTree.generate(value);
  1892. } catch (error) {
  1893. // ignored
  1894. }
  1895. return result;
  1896. }
  1897. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  1898. const stylesheetCharset = getCharset(stylesheetContent);
  1899. if (stylesheetCharset) {
  1900. return stylesheetCharset == charset.toLowerCase();
  1901. } else {
  1902. return true;
  1903. }
  1904. }
  1905. function getCharset(stylesheetContent) {
  1906. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  1907. if (match && match[1]) {
  1908. return match[1].toLowerCase().trim();
  1909. }
  1910. }
  1911. function getOnEventAttributeNames(doc) {
  1912. const element = doc.createElement("div");
  1913. const attributeNames = [];
  1914. for (const propertyName in element) {
  1915. if (propertyName.startsWith("on")) {
  1916. attributeNames.push(propertyName);
  1917. }
  1918. }
  1919. attributeNames.push("onunload");
  1920. return attributeNames;
  1921. }
  1922. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1923. let maxLength;
  1924. if (template) {
  1925. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1926. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1927. if (template.match(replaceRegExp)) {
  1928. const matchedLength = template.match(replaceRegExp)[0];
  1929. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1930. if (isNaN(maxLength) || maxLength <= 0) {
  1931. maxLength = null;
  1932. }
  1933. } else {
  1934. replaceRegExp = new RegExp(regExpVariable, "g");
  1935. }
  1936. if (template.match(replaceRegExp)) {
  1937. let value = await valueGetter();
  1938. if (!dontReplaceSlash) {
  1939. value = value.replace(/\/+/g, replacementCharacter);
  1940. }
  1941. if (maxLength) {
  1942. value = await util.truncateText(value, maxLength);
  1943. }
  1944. return template.replace(replaceRegExp, value);
  1945. }
  1946. }
  1947. return template;
  1948. }
  1949. function getLastSegment(url, replacementCharacter) {
  1950. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1951. if (!lastSegment) {
  1952. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1953. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1954. }
  1955. if (!lastSegment) {
  1956. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1957. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1958. }
  1959. if (!lastSegment) {
  1960. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1961. }
  1962. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1963. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1964. lastSegment = lastSegmentMatch[1];
  1965. }
  1966. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1967. return lastSegment;
  1968. }
  1969. function getRegExp(string) {
  1970. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1971. }
  1972. function getUrlFunctions(stylesheetContent, unique) {
  1973. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1974. if (unique) {
  1975. return [...new Set(result)];
  1976. } else {
  1977. return result;
  1978. }
  1979. }
  1980. function getImportFunctions(stylesheetContent) {
  1981. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1982. }
  1983. function findShortcutIcon(shortcutIcons) {
  1984. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1985. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1986. return shortcutIcons[0];
  1987. }
  1988. function matchURL(stylesheetContent) {
  1989. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1990. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1991. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1992. return match && match[1];
  1993. }
  1994. function testIgnoredPath(resourceURL) {
  1995. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1996. }
  1997. function testValidPath(resourceURL) {
  1998. return resourceURL && !resourceURL.match(EMPTY_URL);
  1999. }
  2000. function testValidURL(resourceURL) {
  2001. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  2002. }
  2003. function matchImport(stylesheetContent) {
  2004. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  2005. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  2006. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2007. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2008. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2009. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2010. if (match) {
  2011. const [, resourceURL, media] = match;
  2012. return { resourceURL, media };
  2013. }
  2014. }
  2015. function removeCssComments(stylesheetContent) {
  2016. try {
  2017. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2018. } catch (error) {
  2019. let start, end;
  2020. do {
  2021. start = stylesheetContent.indexOf("/*");
  2022. end = stylesheetContent.indexOf("*/", start + 2);
  2023. if (start != -1 && end != -1) {
  2024. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2025. }
  2026. } while (start != -1 && end != -1);
  2027. return stylesheetContent;
  2028. }
  2029. }
  2030. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2031. if (mediaQuery) {
  2032. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2033. } else {
  2034. return stylesheetContent;
  2035. }
  2036. }
  2037. function log(...args) {
  2038. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2039. }
  2040. // -----
  2041. // Stats
  2042. // -----
  2043. const STATS_DEFAULT_VALUES = {
  2044. discarded: {
  2045. "HTML bytes": 0,
  2046. "hidden elements": 0,
  2047. "HTML imports": 0,
  2048. scripts: 0,
  2049. objects: 0,
  2050. "audio sources": 0,
  2051. "video sources": 0,
  2052. frames: 0,
  2053. "CSS rules": 0,
  2054. canvas: 0,
  2055. stylesheets: 0,
  2056. resources: 0,
  2057. medias: 0
  2058. },
  2059. processed: {
  2060. "HTML bytes": 0,
  2061. "hidden elements": 0,
  2062. "HTML imports": 0,
  2063. scripts: 0,
  2064. objects: 0,
  2065. "audio sources": 0,
  2066. "video sources": 0,
  2067. frames: 0,
  2068. "CSS rules": 0,
  2069. canvas: 0,
  2070. stylesheets: 0,
  2071. resources: 0,
  2072. medias: 0
  2073. }
  2074. };
  2075. class Stats {
  2076. constructor(options) {
  2077. this.options = options;
  2078. if (options.displayStats) {
  2079. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2080. }
  2081. }
  2082. set(type, subType, value) {
  2083. if (this.options.displayStats) {
  2084. this.data[type][subType] = value;
  2085. }
  2086. }
  2087. add(type, subType, value) {
  2088. if (this.options.displayStats) {
  2089. this.data[type][subType] += value;
  2090. }
  2091. }
  2092. addAll(pageData) {
  2093. if (this.options.displayStats) {
  2094. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2095. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2096. }
  2097. }
  2098. }
  2099. return { getClass };
  2100. })();