single-file-core.js 92 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. /* global globalThis */
  24. const DEBUG = false;
  25. const Set = globalThis.Set;
  26. const Map = globalThis.Map;
  27. let util, cssTree;
  28. function getClass(...args) {
  29. [util, cssTree] = args;
  30. return SingleFileClass;
  31. }
  32. class SingleFileClass {
  33. constructor(options) {
  34. this.options = options;
  35. }
  36. async run() {
  37. const waitForUserScript = globalThis._singleFile_waitForUserScript;
  38. if (this.options.userScriptEnabled && waitForUserScript) {
  39. await waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  40. }
  41. this.runner = new Runner(this.options, true);
  42. await this.runner.loadPage();
  43. await this.runner.initialize();
  44. if (this.options.userScriptEnabled && waitForUserScript) {
  45. await waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  46. }
  47. await this.runner.run();
  48. }
  49. cancel() {
  50. this.cancelled = true;
  51. if (this.runner) {
  52. this.runner.cancel();
  53. }
  54. }
  55. getPageData() {
  56. return this.runner.getPageData();
  57. }
  58. }
  59. // -------------
  60. // ProgressEvent
  61. // -------------
  62. const PAGE_LOADING = "page-loading";
  63. const PAGE_LOADED = "page-loaded";
  64. const RESOURCES_INITIALIZING = "resource-initializing";
  65. const RESOURCES_INITIALIZED = "resources-initialized";
  66. const RESOURCE_LOADED = "resource-loaded";
  67. const PAGE_ENDED = "page-ended";
  68. const STAGE_STARTED = "stage-started";
  69. const STAGE_ENDED = "stage-ended";
  70. const STAGE_TASK_STARTED = "stage-task-started";
  71. const STAGE_TASK_ENDED = "stage-task-ended";
  72. class ProgressEvent {
  73. constructor(type, detail) {
  74. 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 };
  75. }
  76. }
  77. // ------
  78. // Runner
  79. // ------
  80. const RESOLVE_URLS_STAGE = 0;
  81. const REPLACE_DATA_STAGE = 1;
  82. const REPLACE_DOCS_STAGE = 2;
  83. const POST_PROCESS_STAGE = 3;
  84. const STAGES = [{
  85. sequential: [
  86. { action: "preProcessPage" },
  87. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  88. { action: "replaceStyleContents" },
  89. { action: "resetCharsetMeta" },
  90. { option: "saveFavicon", action: "saveFavicon" },
  91. { action: "replaceCanvasElements" },
  92. { action: "insertFonts" },
  93. { action: "insertShadowRootContents" },
  94. { action: "setInputValues" },
  95. { option: "removeScripts", action: "removeScripts" },
  96. { option: "selected", action: "removeUnselectedElements" },
  97. { option: "removeVideoSrc", action: "insertVideoPosters" },
  98. { option: "removeFrames", action: "removeFrames" },
  99. { option: "removeVideoSrc", action: "removeVideoSources" },
  100. { option: "removeAudioSrc", action: "removeAudioSources" },
  101. { action: "removeDiscardedResources" },
  102. { option: "removeHiddenElements", action: "removeHiddenElements" },
  103. { action: "resolveHrefs" },
  104. { action: "resolveStyleAttributeURLs" }
  105. ],
  106. parallel: [
  107. { action: "resolveStylesheetURLs" },
  108. { option: "!removeFrames", action: "resolveFrameURLs" },
  109. { action: "resolveHtmlImportURLs" }
  110. ]
  111. }, {
  112. sequential: [
  113. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  114. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  115. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  116. ],
  117. parallel: [
  118. { action: "processStylesheets" },
  119. { action: "processStyleAttributes" },
  120. { action: "processPageResources" },
  121. { option: "!removeScripts", action: "processScripts" }
  122. ]
  123. }, {
  124. sequential: [
  125. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  126. ],
  127. parallel: [
  128. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  129. { option: "!removeFrames", action: "processFrames" },
  130. { option: "!removeImports", action: "processHtmlImports" },
  131. ]
  132. }, {
  133. sequential: [
  134. { action: "replaceStylesheets" },
  135. { action: "replaceStyleAttributes" },
  136. { action: "insertVariables" },
  137. { option: "compressHTML", action: "compressHTML" },
  138. { action: "cleanupPage" }
  139. ],
  140. parallel: [
  141. { option: "enableMaff", action: "insertMAFFMetaData" },
  142. { action: "setDocInfo" }
  143. ]
  144. }];
  145. class Runner {
  146. constructor(options, root) {
  147. const rootDocDefined = root && options.doc;
  148. this.root = root;
  149. this.options = options;
  150. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  151. const matchResourceReferrer = this.options.url.match(/^.*\//);
  152. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  153. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  154. this.options.rootDocument = root;
  155. this.options.updatedResources = this.options.updatedResources || {};
  156. this.options.fontTests = new Map();
  157. this.batchRequest = new BatchRequest();
  158. this.processor = new Processor(options, this.batchRequest);
  159. if (rootDocDefined) {
  160. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  161. this.options.canvases = docData.canvases;
  162. this.options.fonts = docData.fonts;
  163. this.options.stylesheets = docData.stylesheets;
  164. this.options.images = docData.images;
  165. this.options.posters = docData.posters;
  166. this.options.usedFonts = docData.usedFonts;
  167. this.options.shadowRoots = docData.shadowRoots;
  168. this.options.imports = docData.imports;
  169. this.options.referrer = docData.referrer;
  170. this.markedElements = docData.markedElements;
  171. }
  172. if (this.options.saveRawPage) {
  173. this.options.removeFrames = true;
  174. }
  175. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  176. this.onprogress = options.onprogress || (() => { });
  177. }
  178. async loadPage() {
  179. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  180. await this.processor.loadPage(this.options.content);
  181. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  182. }
  183. async initialize() {
  184. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  185. await this.executeStage(RESOLVE_URLS_STAGE);
  186. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  187. if (this.root && this.options.doc) {
  188. util.postProcessDoc(this.options.doc, this.markedElements);
  189. }
  190. }
  191. cancel() {
  192. this.cancelled = true;
  193. this.batchRequest.cancel();
  194. if (this.root) {
  195. if (this.options.frames) {
  196. this.options.frames.forEach(cancelRunner);
  197. }
  198. if (this.options.imports) {
  199. this.options.imports.forEach(cancelRunner);
  200. }
  201. }
  202. function cancelRunner(resourceData) {
  203. if (resourceData.runner) {
  204. resourceData.runner.cancel();
  205. }
  206. }
  207. }
  208. async run() {
  209. if (this.root) {
  210. this.processor.initialize(this.batchRequest);
  211. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  212. }
  213. await this.batchRequest.run(detail => {
  214. detail.pageURL = this.options.url;
  215. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  216. }, this.options);
  217. await this.pendingPromises;
  218. this.options.doc = null;
  219. this.options.win = null;
  220. await this.executeStage(REPLACE_DOCS_STAGE);
  221. await this.executeStage(POST_PROCESS_STAGE);
  222. this.processor.finalize();
  223. }
  224. getDocument() {
  225. return this.processor.doc;
  226. }
  227. getStyleSheets() {
  228. return this.processor.stylesheets;
  229. }
  230. getPageData() {
  231. if (this.root) {
  232. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  233. }
  234. return this.processor.getPageData();
  235. }
  236. async executeStage(step) {
  237. if (DEBUG) {
  238. log("**** STARTED STAGE", step, "****");
  239. }
  240. const frame = !this.root;
  241. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  242. STAGES[step].sequential.forEach(task => {
  243. let startTime;
  244. if (DEBUG) {
  245. startTime = Date.now();
  246. log(" -- STARTED task =", task.action);
  247. }
  248. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  249. if (!this.cancelled) {
  250. this.executeTask(task);
  251. }
  252. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  253. if (DEBUG) {
  254. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  255. }
  256. });
  257. let parallelTasksPromise;
  258. if (STAGES[step].parallel) {
  259. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  260. let startTime;
  261. if (DEBUG) {
  262. startTime = Date.now();
  263. log(" // STARTED task =", task.action);
  264. }
  265. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  266. if (!this.cancelled) {
  267. await this.executeTask(task);
  268. }
  269. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  270. if (DEBUG) {
  271. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  272. }
  273. }));
  274. } else {
  275. parallelTasksPromise = Promise.resolve();
  276. }
  277. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  278. if (DEBUG) {
  279. log("**** ENDED STAGE", step, "****");
  280. }
  281. return parallelTasksPromise;
  282. }
  283. executeTask(task) {
  284. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  285. return this.processor[task.action]();
  286. }
  287. }
  288. }
  289. // ------------
  290. // BatchRequest
  291. // ------------
  292. class BatchRequest {
  293. constructor() {
  294. this.requests = new Map();
  295. this.duplicates = new Map();
  296. }
  297. addURL(resourceURL, asBinary, expectedType, groupDuplicates) {
  298. return new Promise((resolve, reject) => {
  299. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType]);
  300. let resourceRequests = this.requests.get(requestKey);
  301. if (!resourceRequests) {
  302. resourceRequests = [];
  303. this.requests.set(requestKey, resourceRequests);
  304. }
  305. const callbacks = { resolve, reject };
  306. resourceRequests.push(callbacks);
  307. if (groupDuplicates) {
  308. let duplicateRequests = this.duplicates.get(requestKey);
  309. if (!duplicateRequests) {
  310. duplicateRequests = [];
  311. this.duplicates.set(requestKey, duplicateRequests);
  312. }
  313. duplicateRequests.push(callbacks);
  314. }
  315. });
  316. }
  317. getMaxResources() {
  318. return this.requests.size;
  319. }
  320. run(onloadListener, options) {
  321. const resourceURLs = [...this.requests.keys()];
  322. let indexResource = 0;
  323. return Promise.all(resourceURLs.map(async requestKey => {
  324. const [resourceURL, asBinary, expectedType] = JSON.parse(requestKey);
  325. const resourceRequests = this.requests.get(requestKey);
  326. try {
  327. const currentIndexResource = indexResource;
  328. indexResource = indexResource + 1;
  329. const content = await util.getContent(resourceURL, {
  330. asBinary,
  331. expectedType,
  332. maxResourceSize: options.maxResourceSize,
  333. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  334. frameId: options.windowId,
  335. resourceReferrer: options.resourceReferrer
  336. });
  337. onloadListener({ url: resourceURL });
  338. if (!this.cancelled) {
  339. resourceRequests.forEach(callbacks => {
  340. const duplicateCallbacks = this.duplicates.get(requestKey);
  341. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  342. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  343. });
  344. }
  345. } catch (error) {
  346. indexResource = indexResource + 1;
  347. onloadListener({ url: resourceURL });
  348. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  349. }
  350. this.requests.delete(requestKey);
  351. }));
  352. }
  353. cancel() {
  354. this.cancelled = true;
  355. const resourceURLs = [...this.requests.keys()];
  356. resourceURLs.forEach(requestKey => {
  357. const resourceRequests = this.requests.get(requestKey);
  358. resourceRequests.forEach(callbacks => callbacks.reject());
  359. this.requests.delete(requestKey);
  360. });
  361. }
  362. }
  363. // ---------
  364. // Processor
  365. // ---------
  366. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  367. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  368. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  369. const SCRIPT_TAG_FOUND = /<script/gi;
  370. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  371. const CANVAS_TAG_FOUND = /<canvas/gi;
  372. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  373. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  374. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  375. const UTF8_CHARSET = "utf-8";
  376. class Processor {
  377. constructor(options, batchRequest) {
  378. this.options = options;
  379. this.stats = new Stats(options);
  380. this.baseURI = normalizeURL(options.baseURI || options.url);
  381. this.batchRequest = batchRequest;
  382. this.stylesheets = new Map();
  383. this.styles = new Map();
  384. this.cssVariables = new Map();
  385. this.fontTests = options.fontTests;
  386. }
  387. initialize() {
  388. this.options.saveDate = new Date();
  389. this.options.saveUrl = this.options.url;
  390. if (this.options.enableMaff) {
  391. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url));
  392. }
  393. this.maxResources = this.batchRequest.getMaxResources();
  394. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  395. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  396. }
  397. if (!this.options.removeImports && this.options.imports) {
  398. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  399. }
  400. this.stats.set("processed", "resources", this.maxResources);
  401. }
  402. async loadPage(pageContent, charset) {
  403. let content;
  404. if (!pageContent || this.options.saveRawPage) {
  405. content = await util.getContent(this.baseURI, {
  406. maxResourceSize: this.options.maxResourceSize,
  407. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  408. charset,
  409. frameId: this.options.windowId,
  410. resourceReferrer: this.options.resourceReferrer
  411. });
  412. pageContent = content.data;
  413. }
  414. this.doc = util.parseDocContent(pageContent, this.baseURI);
  415. if (this.options.saveRawPage) {
  416. let charset;
  417. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  418. const charsetDeclaration = element.content.split(";")[1];
  419. if (charsetDeclaration && !charset) {
  420. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  421. }
  422. });
  423. if (charset && charset.toLowerCase() != content.charset.toLowerCase()) {
  424. return this.loadPage(pageContent, charset);
  425. }
  426. }
  427. this.workStyleElement = this.doc.createElement("style");
  428. this.doc.body.appendChild(this.workStyleElement);
  429. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  430. }
  431. finalize() {
  432. if (this.workStyleElement.parentNode) {
  433. this.workStyleElement.remove();
  434. }
  435. }
  436. async getPageData() {
  437. util.postProcessDoc(this.doc);
  438. const url = util.parseURL(this.baseURI);
  439. if (this.options.insertSingleFileComment) {
  440. const firstComment = this.doc.documentElement.firstChild;
  441. let infobarURL = this.options.saveUrl, infobarSaveDate = this.options.saveDate;
  442. if (firstComment.nodeType == 8 && (firstComment.textContent.includes(util.COMMENT_HEADER_LEGACY) || firstComment.textContent.includes(util.COMMENT_HEADER))) {
  443. const info = this.doc.documentElement.firstChild.textContent.split("\n");
  444. const [, , url, saveDate] = info;
  445. infobarURL = url.split("url: ")[1];
  446. infobarSaveDate = saveDate.split("saved date: ")[1];
  447. firstComment.remove();
  448. }
  449. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  450. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  451. " \n url: " + infobarURL +
  452. " \n saved date: " + infobarSaveDate +
  453. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  454. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  455. }
  456. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  457. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  458. if (!canonicalLink) {
  459. canonicalLink = this.doc.createElement("link");
  460. canonicalLink.setAttribute("rel", "canonical");
  461. this.doc.head.appendChild(canonicalLink);
  462. }
  463. if (canonicalLink && !canonicalLink.href) {
  464. canonicalLink.href = this.options.saveUrl;
  465. }
  466. }
  467. if (this.options.insertMetaCSP) {
  468. const metaTag = this.doc.createElement("meta");
  469. metaTag.httpEquiv = "content-security-policy";
  470. metaTag.content = "default-src 'none'; font-src 'self' data:; img-src 'self' data:; style-src 'unsafe-inline'; media-src 'self' data:; script-src 'unsafe-inline' data:;";
  471. this.doc.head.appendChild(metaTag);
  472. }
  473. if (this.options.insertMetaNoIndex) {
  474. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  475. if (!metaElement) {
  476. metaElement = this.doc.createElement("meta");
  477. metaElement.setAttribute("name", "robots");
  478. metaElement.setAttribute("content", "noindex");
  479. this.doc.head.appendChild(metaElement);
  480. }
  481. }
  482. let size;
  483. if (this.options.displayStats) {
  484. size = util.getContentSize(this.doc.documentElement.outerHTML);
  485. }
  486. const content = util.serialize(this.doc, this.options.compressHTML);
  487. if (this.options.displayStats) {
  488. const contentSize = util.getContentSize(content);
  489. this.stats.set("processed", "HTML bytes", contentSize);
  490. this.stats.add("discarded", "HTML bytes", size - contentSize);
  491. }
  492. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  493. const replacementCharacter = this.options.filenameReplacementCharacter;
  494. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  495. if (!this.options.backgroundSave) {
  496. filename = filename.replace(/\//g, replacementCharacter);
  497. }
  498. if (!this.options.saveToGDrive && !this.options.saveToGitHub && !this.options.saveWithCompanion && util.getContentSize(filename) > this.options.filenameMaxLength) {
  499. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  500. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  501. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  502. filename = filename + "…" + extension;
  503. }
  504. if (!filename) {
  505. filename = "Unnamed page";
  506. }
  507. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  508. const pageData = {
  509. stats: this.stats.data,
  510. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  511. filename,
  512. content
  513. };
  514. if (this.options.addProof) {
  515. pageData.hash = await util.digest("SHA-256", content);
  516. }
  517. if (this.options.retrieveLinks) {
  518. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  519. }
  520. return pageData;
  521. }
  522. preProcessPage() {
  523. if (this.options.win) {
  524. 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));
  525. }
  526. if (this.options.images && !this.options.saveRawPage) {
  527. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  528. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  529. if (attributeValue) {
  530. const imageData = this.options.images[Number(attributeValue)];
  531. if (imageData) {
  532. if (this.options.removeHiddenElements && (
  533. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  534. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  535. )) {
  536. imgElement.setAttribute("src", EMPTY_IMAGE);
  537. } else {
  538. if (imageData.currentSrc) {
  539. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  540. imgElement.setAttribute("src", imageData.currentSrc);
  541. }
  542. if (this.options.loadDeferredImages) {
  543. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  544. imageData.src = imgElement.dataset.src;
  545. imgElement.setAttribute("src", imgElement.dataset.src);
  546. imgElement.removeAttribute("data-src");
  547. }
  548. }
  549. }
  550. }
  551. }
  552. });
  553. if (this.options.loadDeferredImages) {
  554. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  555. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  556. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  557. imgElement.removeAttribute("data-srcset");
  558. }
  559. });
  560. }
  561. }
  562. }
  563. replaceStyleContents() {
  564. if (this.options.stylesheets) {
  565. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  566. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  567. if (attributeValue) {
  568. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  569. if (stylesheetContent) {
  570. styleElement.textContent = stylesheetContent;
  571. }
  572. }
  573. });
  574. }
  575. }
  576. removeUnselectedElements() {
  577. removeUnmarkedElements(this.doc.body);
  578. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  579. function removeUnmarkedElements(element) {
  580. let selectedElementFound = false;
  581. Array.from(element.childNodes).forEach(node => {
  582. if (node.nodeType == 1) {
  583. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  584. selectedElementFound = selectedElementFound || isSelectedElement;
  585. if (isSelectedElement) {
  586. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  587. removeUnmarkedElements(node);
  588. } else if (selectedElementFound) {
  589. removeNode(node);
  590. } else {
  591. hideNode(node);
  592. }
  593. }
  594. });
  595. }
  596. function removeNode(node) {
  597. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  598. node.remove();
  599. } else {
  600. hideNode(node);
  601. }
  602. }
  603. function hideNode(node) {
  604. if (canHideNode(node)) {
  605. node.style.setProperty("display", "none", "important");
  606. Array.from(node.childNodes).forEach(removeNode);
  607. }
  608. }
  609. function canHideNode(node) {
  610. if (node.nodeType == 1) {
  611. const tagName = node.tagName && node.tagName.toLowerCase();
  612. return (tagName != "svg" && tagName != "style" && tagName != "link");
  613. }
  614. }
  615. }
  616. insertVideoPosters() {
  617. if (this.options.posters) {
  618. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  619. let videoElement;
  620. if (element.tagName == "VIDEO") {
  621. videoElement = element;
  622. } else {
  623. videoElement = element.parentElement;
  624. }
  625. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  626. if (attributeValue) {
  627. const posterURL = this.options.posters[Number(attributeValue)];
  628. if (!videoElement.poster && posterURL) {
  629. videoElement.setAttribute("poster", posterURL);
  630. }
  631. }
  632. });
  633. }
  634. }
  635. removeFrames() {
  636. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  637. this.stats.set("discarded", "frames", frameElements.length);
  638. this.stats.set("processed", "frames", frameElements.length);
  639. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  640. }
  641. removeScripts() {
  642. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  643. this.doc.querySelectorAll("[href]").forEach(element => {
  644. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  645. element.setAttribute("href", "");
  646. }
  647. });
  648. this.doc.querySelectorAll("[src]").forEach(element => {
  649. if (element.src && element.src.match(/^\s*javascript:/)) {
  650. element.removeAttribute("src");
  651. }
  652. });
  653. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  654. this.stats.set("discarded", "scripts", scriptElements.length);
  655. this.stats.set("processed", "scripts", scriptElements.length);
  656. scriptElements.forEach(element => element.remove());
  657. }
  658. removeVideoSources() {
  659. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  660. this.stats.set("discarded", "video sources", videoSourceElements.length);
  661. this.stats.set("processed", "video sources", videoSourceElements.length);
  662. videoSourceElements.forEach(element => {
  663. if (element.tagName == "SOURCE") {
  664. element.remove();
  665. } else {
  666. videoSourceElements.forEach(element => element.removeAttribute("src"));
  667. }
  668. });
  669. }
  670. removeAudioSources() {
  671. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  672. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  673. this.stats.set("processed", "audio sources", audioSourceElements.length);
  674. audioSourceElements.forEach(element => {
  675. if (element.tagName == "SOURCE") {
  676. element.remove();
  677. } else {
  678. audioSourceElements.forEach(element => element.removeAttribute("src"));
  679. }
  680. });
  681. }
  682. removeDiscardedResources() {
  683. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  684. this.doc.querySelectorAll("button button").forEach(element => element.remove());
  685. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  686. 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\"])");
  687. this.stats.set("discarded", "objects", objectElements.length);
  688. this.stats.set("processed", "objects", objectElements.length);
  689. objectElements.forEach(element => element.remove());
  690. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  691. replacedAttributeValue.forEach(element => {
  692. let regExp;
  693. if (this.options.removeScripts) {
  694. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  695. } else {
  696. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  697. }
  698. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  699. if (relValue.length) {
  700. element.setAttribute("rel", relValue);
  701. } else {
  702. element.remove();
  703. }
  704. });
  705. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  706. if (this.options.removeHiddenElements) {
  707. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  708. }
  709. if (!this.options.saveFavicon) {
  710. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  711. }
  712. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  713. }
  714. resetCharsetMeta() {
  715. let charset;
  716. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  717. const charsetDeclaration = element.content.split(";")[1];
  718. if (charsetDeclaration && !charset) {
  719. charset = charsetDeclaration.split("=")[1];
  720. if (charset) {
  721. this.charset = charset.trim().toLowerCase();
  722. }
  723. }
  724. element.remove();
  725. });
  726. const metaElement = this.doc.createElement("meta");
  727. metaElement.setAttribute("charset", UTF8_CHARSET);
  728. if (this.doc.head.firstChild) {
  729. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  730. } else {
  731. this.doc.head.appendChild(metaElement);
  732. }
  733. }
  734. setInputValues() {
  735. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  736. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  737. input.setAttribute("value", value || "");
  738. });
  739. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  740. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  741. if (value == "true") {
  742. input.setAttribute("checked", "");
  743. }
  744. });
  745. this.doc.querySelectorAll("textarea").forEach(textarea => {
  746. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  747. textarea.textContent = value || "";
  748. });
  749. this.doc.querySelectorAll("select").forEach(select => {
  750. select.querySelectorAll("option").forEach(option => {
  751. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  752. if (selected) {
  753. option.setAttribute("selected", "");
  754. }
  755. });
  756. });
  757. }
  758. saveFavicon() {
  759. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  760. if (!faviconElement) {
  761. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  762. }
  763. if (!faviconElement) {
  764. faviconElement = this.doc.createElement("link");
  765. faviconElement.setAttribute("type", "image/x-icon");
  766. faviconElement.setAttribute("rel", "shortcut icon");
  767. faviconElement.setAttribute("href", "/favicon.ico");
  768. }
  769. this.doc.head.appendChild(faviconElement);
  770. }
  771. replaceCanvasElements() {
  772. if (this.options.canvases) {
  773. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  774. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  775. if (attributeValue) {
  776. const canvasData = this.options.canvases[Number(attributeValue)];
  777. if (canvasData) {
  778. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  779. this.stats.add("processed", "canvas", 1);
  780. }
  781. }
  782. });
  783. }
  784. }
  785. insertFonts() {
  786. if (this.options.fonts && this.options.fonts.length) {
  787. let stylesheetContent = "";
  788. this.options.fonts.forEach(fontData => {
  789. if (fontData["font-family"] && fontData.src) {
  790. stylesheetContent += "@font-face{";
  791. let stylesContent = "";
  792. Object.keys(fontData).forEach(fontStyle => {
  793. if (stylesContent) {
  794. stylesContent += ";";
  795. }
  796. stylesContent += fontStyle + ":" + fontData[fontStyle];
  797. });
  798. stylesheetContent += stylesContent + "}";
  799. }
  800. });
  801. if (stylesheetContent) {
  802. const styleElement = this.doc.createElement("style");
  803. styleElement.textContent = stylesheetContent;
  804. const existingStyleElement = this.doc.querySelector("style");
  805. if (existingStyleElement) {
  806. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  807. } else {
  808. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  809. }
  810. }
  811. }
  812. }
  813. removeHiddenElements() {
  814. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  815. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  816. this.stats.set("discarded", "hidden elements", removedElements.length);
  817. this.stats.set("processed", "hidden elements", removedElements.length);
  818. if (hiddenElements.length) {
  819. const styleElement = this.doc.createElement("style");
  820. styleElement.textContent = ".sf-hidden{display:none!important;}";
  821. this.doc.head.appendChild(styleElement);
  822. hiddenElements.forEach(element => {
  823. if (element.style.getPropertyValue("display") != "none") {
  824. if (element.style.getPropertyPriority("display") == "important") {
  825. element.style.setProperty("display", "none", "important");
  826. } else {
  827. element.classList.add("sf-hidden");
  828. }
  829. }
  830. });
  831. }
  832. removedElements.forEach(element => element.remove());
  833. }
  834. resolveHrefs() {
  835. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  836. const href = element.getAttribute("href").trim();
  837. if (!testIgnoredPath(href)) {
  838. let resolvedURL;
  839. try {
  840. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  841. } catch (error) {
  842. // ignored
  843. }
  844. if (resolvedURL) {
  845. const url = normalizeURL(this.options.url);
  846. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  847. resolvedURL = resolvedURL.substring(url.length);
  848. }
  849. try {
  850. element.setAttribute("href", resolvedURL);
  851. } catch (error) {
  852. // ignored
  853. }
  854. }
  855. }
  856. });
  857. }
  858. resolveStyleAttributeURLs() {
  859. this.doc.querySelectorAll("[style]").forEach(element => {
  860. let styleContent = element.getAttribute("style");
  861. if (this.options.compressCSS) {
  862. styleContent = util.compressCSS(styleContent);
  863. }
  864. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement);
  865. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  866. this.styles.set(element, declarationList);
  867. });
  868. }
  869. async resolveStylesheetURLs() {
  870. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]")).map(async element => {
  871. const options = {
  872. maxResourceSize: this.options.maxResourceSize,
  873. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  874. url: this.options.url,
  875. charset: this.charset,
  876. compressCSS: this.options.compressCSS,
  877. updatedResources: this.options.updatedResources,
  878. rootDocument: this.options.rootDocument,
  879. frameId: this.options.windowId,
  880. resourceReferrer: this.options.resourceReferrer
  881. };
  882. let mediaText;
  883. if (element.media) {
  884. mediaText = element.media.toLowerCase();
  885. }
  886. const stylesheetInfo = { mediaText };
  887. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  888. stylesheetInfo.scoped = true;
  889. }
  890. if (element.tagName == "LINK" && element.charset) {
  891. options.charset = element.charset;
  892. }
  893. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  894. }));
  895. if (this.options.rootDocument) {
  896. 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]);
  897. await Promise.all(newResources.map(async resource => {
  898. resource.retrieved = true;
  899. const stylesheetInfo = {};
  900. const element = this.doc.createElement("style");
  901. this.doc.body.appendChild(element);
  902. element.textContent = resource.content;
  903. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, this.options, this.workStyleElement);
  904. }));
  905. }
  906. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  907. stylesheets.set(element, stylesheetInfo);
  908. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  909. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  910. options = Object.assign({}, options, { charset: getCharset(stylesheetContent) });
  911. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  912. }
  913. let stylesheet;
  914. try {
  915. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  916. } catch (error) {
  917. // ignored
  918. }
  919. if (stylesheet && stylesheet.children) {
  920. if (options.compressCSS) {
  921. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  922. }
  923. stylesheetInfo.stylesheet = stylesheet;
  924. } else {
  925. stylesheets.delete(element);
  926. }
  927. }
  928. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  929. let content;
  930. if (element.tagName == "LINK") {
  931. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  932. } else {
  933. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  934. }
  935. return content || "";
  936. }
  937. }
  938. async resolveFrameURLs() {
  939. if (!this.options.saveRawPage) {
  940. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  941. await Promise.all(frameElements.map(async frameElement => {
  942. if (frameElement.tagName == "OBJECT") {
  943. frameElement.setAttribute("data", "data:text/html,");
  944. } else {
  945. frameElement.removeAttribute("src");
  946. frameElement.removeAttribute("srcdoc");
  947. }
  948. Array.from(frameElement.childNodes).forEach(node => node.remove());
  949. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  950. if (this.options.frames && frameWindowId) {
  951. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  952. if (frameData) {
  953. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  954. }
  955. }
  956. }));
  957. }
  958. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  959. options.insertSingleFileComment = false;
  960. options.insertCanonicalLink = false;
  961. options.insertMetaNoIndex = false;
  962. options.saveFavicon = false;
  963. options.url = frameData.baseURI;
  964. options.windowId = frameWindowId;
  965. if (frameData.content) {
  966. options.content = frameData.content;
  967. options.canvases = frameData.canvases;
  968. options.fonts = frameData.fonts;
  969. options.stylesheets = frameData.stylesheets;
  970. options.images = frameData.images;
  971. options.posters = frameData.posters;
  972. options.usedFonts = frameData.usedFonts;
  973. options.shadowRoots = frameData.shadowRoots;
  974. options.imports = frameData.imports;
  975. frameData.runner = new Runner(options);
  976. frameData.frameElement = frameElement;
  977. await frameData.runner.loadPage();
  978. await frameData.runner.initialize();
  979. frameData.maxResources = batchRequest.getMaxResources();
  980. }
  981. }
  982. }
  983. insertShadowRootContents() {
  984. const doc = this.doc;
  985. const options = this.options;
  986. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  987. processElement(this.doc);
  988. if (this.options.removeScripts) {
  989. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  990. }
  991. const scriptElement = doc.createElement("script");
  992. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  993. 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)}}})}})()`;
  994. doc.body.appendChild(scriptElement);
  995. }
  996. function processElement(element) {
  997. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  998. shadowRootElements.forEach(element => {
  999. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  1000. if (attributeValue) {
  1001. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  1002. if (shadowRootData) {
  1003. const templateElement = doc.createElement("template");
  1004. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  1005. if (shadowRootData.delegatesFocus) {
  1006. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  1007. }
  1008. if (shadowRootData.adoptedStyleSheets) {
  1009. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  1010. const styleElement = doc.createElement("style");
  1011. styleElement.textContent = stylesheetContent;
  1012. templateElement.appendChild(styleElement);
  1013. });
  1014. }
  1015. const shadowDoc = util.parseDocContent(shadowRootData.content);
  1016. if (shadowDoc.head) {
  1017. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  1018. if (metaCharset) {
  1019. metaCharset.remove();
  1020. }
  1021. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1022. }
  1023. if (shadowDoc.body) {
  1024. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1025. }
  1026. processElement(templateElement);
  1027. if (element.firstChild) {
  1028. element.insertBefore(templateElement, element.firstChild);
  1029. } else {
  1030. element.appendChild(templateElement);
  1031. }
  1032. }
  1033. }
  1034. });
  1035. }
  1036. }
  1037. async resolveHtmlImportURLs() {
  1038. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1039. await Promise.all(linkElements.map(async linkElement => {
  1040. const resourceURL = linkElement.href;
  1041. linkElement.removeAttribute("href");
  1042. const options = Object.create(this.options);
  1043. options.insertSingleFileComment = false;
  1044. options.insertCanonicalLink = false;
  1045. options.insertMetaNoIndex = false;
  1046. options.saveFavicon = false;
  1047. options.removeUnusedStyles = false;
  1048. options.removeAlternativeMedias = false;
  1049. options.removeUnusedFonts = false;
  1050. options.removeHiddenElements = false;
  1051. options.url = resourceURL;
  1052. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1053. if (attributeValue) {
  1054. const importData = options.imports[Number(attributeValue)];
  1055. if (importData) {
  1056. options.content = importData.content;
  1057. importData.runner = new Runner(options);
  1058. await importData.runner.loadPage();
  1059. await importData.runner.initialize();
  1060. if (!options.removeImports) {
  1061. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1062. }
  1063. importData.runner.getStyleSheets().forEach(stylesheet => {
  1064. const importedStyleElement = this.doc.createElement("style");
  1065. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1066. this.stylesheets.set(importedStyleElement, stylesheet);
  1067. });
  1068. }
  1069. }
  1070. if (options.removeImports) {
  1071. linkElement.remove();
  1072. this.stats.add("discarded", "HTML imports", 1);
  1073. }
  1074. }));
  1075. }
  1076. removeUnusedStyles() {
  1077. if (!this.mediaAllInfo) {
  1078. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1079. }
  1080. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1081. this.stats.set("processed", "CSS rules", stats.processed);
  1082. this.stats.set("discarded", "CSS rules", stats.discarded);
  1083. }
  1084. removeUnusedFonts() {
  1085. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1086. }
  1087. removeAlternativeMedias() {
  1088. const stats = util.minifyMedias(this.stylesheets);
  1089. this.stats.set("processed", "medias", stats.processed);
  1090. this.stats.set("discarded", "medias", stats.discarded);
  1091. }
  1092. async processStylesheets() {
  1093. this.options.fontURLs = new Map();
  1094. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1095. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1096. ));
  1097. }
  1098. async processStyleAttributes() {
  1099. return Promise.all([...this.styles].map(([, declarationList]) =>
  1100. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1101. ));
  1102. }
  1103. async processPageResources() {
  1104. const processAttributeArgs = [
  1105. ["link[href][rel*=\"icon\"]", "href", false, true],
  1106. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1107. ["img[src], input[src][type=image]", "src", true],
  1108. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1109. ["video[poster]", "poster"],
  1110. ["*[background]", "background"],
  1111. ["image", "xlink:href"],
  1112. ["image", "href"]
  1113. ];
  1114. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1115. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1116. );
  1117. resourcePromises = resourcePromises.concat([
  1118. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.doc, this.baseURI, this.options, this.batchRequest),
  1119. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1120. ]);
  1121. if (!this.options.removeAudioSrc) {
  1122. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1123. }
  1124. if (!this.options.removeVideoSrc) {
  1125. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1126. }
  1127. await Promise.all(resourcePromises);
  1128. if (this.options.saveFavicon) {
  1129. ProcessorHelper.processShortcutIcons(this.doc);
  1130. }
  1131. }
  1132. async processScripts() {
  1133. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1134. let resourceURL;
  1135. let scriptSrc;
  1136. if (element.tagName == "SCRIPT") {
  1137. scriptSrc = element.getAttribute("src");
  1138. } else {
  1139. scriptSrc = element.getAttribute("href");
  1140. }
  1141. element.removeAttribute("integrity");
  1142. element.textContent = "";
  1143. try {
  1144. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1145. } catch (error) {
  1146. // ignored
  1147. }
  1148. if (testValidURL(resourceURL)) {
  1149. if (element.tagName == "SCRIPT") {
  1150. element.removeAttribute("src");
  1151. } else {
  1152. element.removeAttribute("href");
  1153. }
  1154. this.stats.add("processed", "scripts", 1);
  1155. const content = await util.getContent(resourceURL, {
  1156. asBinary: true,
  1157. charset: this.charset != UTF8_CHARSET && this.charset,
  1158. maxResourceSize: this.options.maxResourceSize,
  1159. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1160. frameId: this.options.windowId,
  1161. resourceReferrer: this.options.resourceReferrer
  1162. });
  1163. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1164. if (element.tagName == "SCRIPT") {
  1165. element.setAttribute("src", content.data);
  1166. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1167. element.setAttribute("async", "");
  1168. }
  1169. } else {
  1170. const scriptElement = this.doc.createElement("script");
  1171. scriptElement.setAttribute("src", content.data);
  1172. scriptElement.setAttribute("async", "");
  1173. element.parentElement.replaceChild(scriptElement, element);
  1174. }
  1175. }
  1176. }));
  1177. }
  1178. removeAlternativeImages() {
  1179. util.removeAlternativeImages(this.doc);
  1180. }
  1181. async removeAlternativeFonts() {
  1182. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1183. }
  1184. async processFrames() {
  1185. if (this.options.frames) {
  1186. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1187. await Promise.all(frameElements.map(async frameElement => {
  1188. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1189. if (frameWindowId) {
  1190. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1191. if (frameData) {
  1192. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1193. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1194. this.stats.add("processed", "frames", 1);
  1195. await frameData.runner.run();
  1196. const pageData = await frameData.runner.getPageData();
  1197. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1198. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1199. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(CANVAS_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1200. sandbox += " allow-scripts allow-same-origin";
  1201. }
  1202. frameElement.setAttribute("sandbox", sandbox);
  1203. if (frameElement.tagName == "OBJECT") {
  1204. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1205. } else {
  1206. if (frameElement.tagName == "FRAME") {
  1207. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1208. } else {
  1209. frameElement.setAttribute("srcdoc", pageData.content);
  1210. frameElement.removeAttribute("src");
  1211. }
  1212. }
  1213. this.stats.addAll(pageData);
  1214. } else {
  1215. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1216. this.stats.add("discarded", "frames", 1);
  1217. }
  1218. }
  1219. }
  1220. }));
  1221. }
  1222. }
  1223. async processHtmlImports() {
  1224. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1225. await Promise.all(linkElements.map(async linkElement => {
  1226. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1227. if (attributeValue) {
  1228. const importData = this.options.imports[Number(attributeValue)];
  1229. if (importData.runner) {
  1230. this.stats.add("processed", "HTML imports", 1);
  1231. await importData.runner.run();
  1232. const pageData = await importData.runner.getPageData();
  1233. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1234. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1235. this.stats.addAll(pageData);
  1236. } else {
  1237. this.stats.add("discarded", "HTML imports", 1);
  1238. }
  1239. }
  1240. }));
  1241. }
  1242. replaceStylesheets() {
  1243. this.doc.querySelectorAll("style").forEach(styleElement => {
  1244. const stylesheetInfo = this.stylesheets.get(styleElement);
  1245. if (stylesheetInfo) {
  1246. this.stylesheets.delete(styleElement);
  1247. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1248. styleElement.textContent = stylesheetContent;
  1249. if (stylesheetInfo.mediaText) {
  1250. styleElement.media = stylesheetInfo.mediaText;
  1251. }
  1252. } else {
  1253. styleElement.remove();
  1254. }
  1255. });
  1256. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1257. const stylesheetInfo = this.stylesheets.get(linkElement);
  1258. if (stylesheetInfo) {
  1259. this.stylesheets.delete(linkElement);
  1260. const styleElement = this.doc.createElement("style");
  1261. if (stylesheetInfo.mediaText) {
  1262. styleElement.media = stylesheetInfo.mediaText;
  1263. }
  1264. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1265. styleElement.textContent = stylesheetContent;
  1266. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1267. } else {
  1268. linkElement.remove();
  1269. }
  1270. });
  1271. }
  1272. replaceStyleAttributes() {
  1273. this.doc.querySelectorAll("[style]").forEach(element => {
  1274. const declarations = this.styles.get(element);
  1275. if (declarations) {
  1276. this.styles.delete(element);
  1277. let styleContent = cssTree.generate(declarations);
  1278. element.setAttribute("style", styleContent);
  1279. } else {
  1280. element.setAttribute("style", "");
  1281. }
  1282. });
  1283. }
  1284. insertVariables() {
  1285. if (this.cssVariables.size) {
  1286. const styleElement = this.doc.createElement("style");
  1287. const firstStyleElement = this.doc.head.querySelector("style");
  1288. if (firstStyleElement) {
  1289. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1290. } else {
  1291. this.doc.head.appendChild(styleElement);
  1292. }
  1293. let stylesheetContent = "";
  1294. this.cssVariables.forEach((content, indexResource) => {
  1295. this.cssVariables.delete(indexResource);
  1296. if (stylesheetContent) {
  1297. stylesheetContent += ";";
  1298. }
  1299. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1300. });
  1301. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1302. }
  1303. }
  1304. compressHTML() {
  1305. let size;
  1306. if (this.options.displayStats) {
  1307. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1308. }
  1309. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1310. if (this.options.displayStats) {
  1311. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1312. }
  1313. }
  1314. cleanupPage() {
  1315. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1316. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1317. if (metaCharset) {
  1318. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1319. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1320. this.doc.head.querySelector("meta[charset]").remove();
  1321. }
  1322. }
  1323. }
  1324. resetZoomLevel() {
  1325. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1326. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1327. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1328. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1329. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1330. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1331. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1332. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1333. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1334. this.doc.documentElement.style.removeProperty("-sf-transform");
  1335. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1336. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1337. }
  1338. async insertMAFFMetaData() {
  1339. const maffMetaData = await this.maffMetaDataPromise;
  1340. if (maffMetaData && maffMetaData.content) {
  1341. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1342. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1343. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1344. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1345. if (originalURLElement) {
  1346. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1347. }
  1348. if (archiveTimeElement) {
  1349. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1350. if (value) {
  1351. const date = new Date(value);
  1352. if (!isNaN(date.getTime())) {
  1353. this.options.saveDate = new Date(value);
  1354. }
  1355. }
  1356. }
  1357. }
  1358. }
  1359. async setDocInfo() {
  1360. const titleElement = this.doc.querySelector("title");
  1361. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1362. const authorElement = this.doc.querySelector("meta[name=author]");
  1363. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1364. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1365. const headingElement = this.doc.querySelector("h1");
  1366. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1367. this.options.info = {
  1368. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1369. lang: this.doc.documentElement.lang,
  1370. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1371. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1372. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1373. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1374. };
  1375. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1376. }
  1377. }
  1378. // ---------------
  1379. // ProcessorHelper
  1380. // ---------------
  1381. const DATA_URI_PREFIX = "data:";
  1382. const ABOUT_BLANK_URI = "about:blank";
  1383. const EMPTY_DATA_URI = "data:null;base64,";
  1384. const REGEXP_URL_HASH = /(#.+?)$/;
  1385. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1386. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1387. class ProcessorHelper {
  1388. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1389. const url = util.parseURL(options.saveUrl);
  1390. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1397. await evalDate(options.saveDate);
  1398. await evalDate(options.visitDate, "visit-");
  1399. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1401. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1402. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1403. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1404. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1405. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1406. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1407. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1408. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1409. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1410. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1411. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1412. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1413. for (const [name, value] of params) {
  1414. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1415. }
  1416. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1417. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1418. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1419. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1420. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1421. if (content) {
  1422. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1423. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1424. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1425. }
  1426. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1427. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1428. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1429. template = await evalTemplateVariable(template, "profile-name", () => options.profileName, dontReplaceSlash, options.filenameReplacementCharacter);
  1430. return template.trim();
  1431. function decode(value) {
  1432. try {
  1433. return decodeURI(value);
  1434. } catch (error) {
  1435. return value;
  1436. }
  1437. }
  1438. async function evalDate(date, prefix = "") {
  1439. if (date) {
  1440. template = await evalTemplateVariable(template, prefix + "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1441. template = await evalTemplateVariable(template, prefix + "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1442. template = await evalTemplateVariable(template, prefix + "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1443. template = await evalTemplateVariable(template, prefix + "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1444. template = await evalTemplateVariable(template, prefix + "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1445. template = await evalTemplateVariable(template, prefix + "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1446. template = await evalTemplateVariable(template, prefix + "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1447. template = await evalTemplateVariable(template, prefix + "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1448. template = await evalTemplateVariable(template, prefix + "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1449. template = await evalTemplateVariable(template, prefix + "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1450. template = await evalTemplateVariable(template, prefix + "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1451. template = await evalTemplateVariable(template, prefix + "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1452. template = await evalTemplateVariable(template, prefix + "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1453. template = await evalTemplateVariable(template, prefix + "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1454. template = await evalTemplateVariable(template, prefix + "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1455. template = await evalTemplateVariable(template, prefix + "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1456. template = await evalTemplateVariable(template, prefix + "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1457. template = await evalTemplateVariable(template, prefix + "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1458. template = await evalTemplateVariable(template, prefix + "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1459. template = await evalTemplateVariable(template, prefix + "time-ms", () => String(date.getTime()), dontReplaceSlash, options.filenameReplacementCharacter);
  1460. }
  1461. }
  1462. }
  1463. static setBackgroundImage(element, url, style) {
  1464. element.style.setProperty("background-blend-mode", "normal", "important");
  1465. element.style.setProperty("background-clip", "content-box", "important");
  1466. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1467. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1468. element.style.setProperty("background-image", url, "important");
  1469. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1470. element.style.setProperty("background-origin", "content-box", "important");
  1471. element.style.setProperty("background-repeat", "no-repeat", "important");
  1472. }
  1473. static processShortcutIcons(doc) {
  1474. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1475. if (!shortcutIcon) {
  1476. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1477. if (shortcutIcon) {
  1478. shortcutIcon.rel = "icon";
  1479. }
  1480. }
  1481. if (shortcutIcon) {
  1482. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1483. if (linkElement != shortcutIcon) {
  1484. linkElement.remove();
  1485. }
  1486. });
  1487. }
  1488. }
  1489. static removeSingleLineCssComments(stylesheet) {
  1490. const removedRules = [];
  1491. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1492. const ruleData = cssRule.data;
  1493. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1494. removedRules.push(cssRule);
  1495. }
  1496. }
  1497. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1498. }
  1499. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1500. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet);
  1501. const imports = getImportFunctions(stylesheetContent);
  1502. await Promise.all(imports.map(async cssImport => {
  1503. const match = matchImport(cssImport);
  1504. if (match) {
  1505. const regExpCssImport = getRegExp(cssImport);
  1506. let resourceURL = normalizeURL(match.resourceURL);
  1507. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1508. try {
  1509. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1510. } catch (error) {
  1511. // ignored
  1512. }
  1513. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1514. const content = await getStylesheetContent(resourceURL);
  1515. resourceURL = content.resourceURL;
  1516. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1517. if (content.data && content.data.match(/^<!doctype /i)) {
  1518. content.data = "";
  1519. }
  1520. let importedStylesheetContent = removeCssComments(content.data);
  1521. if (options.compressCSS) {
  1522. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1523. }
  1524. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1525. if (stylesheetContent.includes(cssImport)) {
  1526. const ancestorStyleSheets = new Set(importedStyleSheets);
  1527. ancestorStyleSheets.add(resourceURL);
  1528. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1529. workStylesheet.textContent = importedStylesheetContent;
  1530. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1531. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1532. } else {
  1533. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1534. }
  1535. }
  1536. } else {
  1537. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1538. }
  1539. } else {
  1540. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1541. }
  1542. }
  1543. }));
  1544. return stylesheetContent;
  1545. async function getStylesheetContent(resourceURL) {
  1546. const content = await util.getContent(resourceURL, {
  1547. maxResourceSize: options.maxResourceSize,
  1548. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1549. validateTextContentType: true,
  1550. frameId: options.frameId,
  1551. charset: options.charset,
  1552. resourceReferrer: options.resourceReferrer
  1553. });
  1554. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1555. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1556. return util.getContent(resourceURL, {
  1557. maxResourceSize: options.maxResourceSize,
  1558. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1559. validateTextContentType: true,
  1560. frameId: options.frameId,
  1561. charset: options.charset,
  1562. resourceReferrer: options.resourceReferrer
  1563. });
  1564. } else {
  1565. return content;
  1566. }
  1567. }
  1568. }
  1569. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet) {
  1570. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1571. urlFunctions.map(urlFunction => {
  1572. const originalResourceURL = matchURL(urlFunction);
  1573. let resourceURL = normalizeURL(originalResourceURL);
  1574. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1575. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1576. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1577. }
  1578. if (!testIgnoredPath(resourceURL)) {
  1579. if (!resourceURL || testValidPath(resourceURL)) {
  1580. let resolvedURL;
  1581. if (!originalResourceURL.startsWith("#")) {
  1582. try {
  1583. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1584. } catch (error) {
  1585. // ignored
  1586. }
  1587. }
  1588. if (testValidURL(resolvedURL) && originalResourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1589. try {
  1590. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1591. } catch (error) {
  1592. // ignored
  1593. }
  1594. }
  1595. } else {
  1596. let newUrlFunction;
  1597. if (originalResourceURL) {
  1598. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1599. } else {
  1600. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1601. }
  1602. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1603. }
  1604. }
  1605. });
  1606. return stylesheetContent;
  1607. }
  1608. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1609. resourceURL = normalizeURL(resourceURL);
  1610. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1611. const content = await util.getContent(resourceURL, {
  1612. maxResourceSize: options.maxResourceSize,
  1613. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1614. charset: options.charset,
  1615. frameId: options.frameId,
  1616. resourceReferrer: options.resourceReferrer,
  1617. validateTextContentType: true
  1618. });
  1619. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1620. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1621. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1622. }
  1623. resourceURL = content.resourceURL;
  1624. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1625. if (content.data && content.data.match(/^<!doctype /i)) {
  1626. content.data = "";
  1627. }
  1628. let stylesheetContent = removeCssComments(content.data);
  1629. if (options.compressCSS) {
  1630. stylesheetContent = util.compressCSS(stylesheetContent);
  1631. }
  1632. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1633. return stylesheetContent;
  1634. }
  1635. }
  1636. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1637. const promises = [];
  1638. const removedRules = [];
  1639. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1640. const ruleData = cssRule.data;
  1641. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1642. removedRules.push(cssRule);
  1643. } else if (ruleData.block && ruleData.block.children) {
  1644. if (ruleData.type == "Rule") {
  1645. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1646. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1647. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1648. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1649. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1650. }
  1651. }
  1652. }
  1653. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1654. await Promise.all(promises);
  1655. async function processFontFaceRule(ruleData, fontURLs) {
  1656. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1657. if (declaration.type == "Declaration" && declaration.value.children) {
  1658. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1659. await Promise.all(urlFunctions.map(async urlFunction => {
  1660. const originalResourceURL = matchURL(urlFunction);
  1661. const resourceURL = normalizeURL(originalResourceURL);
  1662. if (!testIgnoredPath(resourceURL)) {
  1663. if (testValidURL(resourceURL)) {
  1664. let { content } = await batchRequest.addURL(resourceURL, true, "font");
  1665. let resourceURLs = fontURLs.get(declaration);
  1666. if (!resourceURLs) {
  1667. resourceURLs = [];
  1668. fontURLs.set(declaration, resourceURLs);
  1669. }
  1670. resourceURLs.push(resourceURL);
  1671. replaceURLs(declaration, originalResourceURL, content);
  1672. }
  1673. }
  1674. }));
  1675. }
  1676. }));
  1677. function replaceURLs(declaration, oldURL, newURL) {
  1678. declaration.value.children.forEach(token => {
  1679. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1680. token.value.value = newURL;
  1681. }
  1682. });
  1683. }
  1684. }
  1685. }
  1686. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1687. await Promise.all(declarations.map(async declaration => {
  1688. if (declaration.value && !declaration.value.children && declaration.value.type == "Raw") {
  1689. try {
  1690. declaration.value = cssTree.parse(declaration.value.value, { context: "value" });
  1691. } catch (error) {
  1692. // ignored
  1693. }
  1694. }
  1695. if (declaration.type == "Declaration" && declaration.value.children) {
  1696. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1697. await Promise.all(urlFunctions.map(async urlFunction => {
  1698. const originalResourceURL = matchURL(urlFunction);
  1699. const resourceURL = normalizeURL(originalResourceURL);
  1700. if (!testIgnoredPath(resourceURL)) {
  1701. if (testValidURL(resourceURL)) {
  1702. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", true);
  1703. let variableDefined;
  1704. const tokens = [];
  1705. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1706. if (!originalResourceURL.startsWith("#")) {
  1707. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1708. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1709. tokens.push({ parent, token, value });
  1710. variableDefined = true;
  1711. } else {
  1712. token.data.value.value = content;
  1713. }
  1714. }
  1715. });
  1716. if (variableDefined) {
  1717. cssVariables.set(indexResource, content);
  1718. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1719. }
  1720. }
  1721. }
  1722. }));
  1723. }
  1724. }));
  1725. function findURLToken(url, children, callback, depth = 0) {
  1726. for (let token = children.head; token; token = token.next) {
  1727. if (token.data.children) {
  1728. findURLToken(url, token.data.children, callback, depth + 1);
  1729. }
  1730. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1731. callback(token, children, depth == 0);
  1732. }
  1733. }
  1734. }
  1735. }
  1736. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1737. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1738. let resourceURL = resourceElement.getAttribute(attributeName);
  1739. if (resourceURL != null) {
  1740. resourceURL = normalizeURL(resourceURL);
  1741. let originURL = resourceElement.dataset.singleFileOriginURL;
  1742. delete resourceElement.dataset.singleFileOriginURL;
  1743. if (!testIgnoredPath(resourceURL)) {
  1744. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1745. if (testValidPath(resourceURL)) {
  1746. try {
  1747. resourceURL = util.resolveURL(resourceURL, baseURI);
  1748. } catch (error) {
  1749. // ignored
  1750. }
  1751. if (testValidURL(resourceURL)) {
  1752. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", resourceElement.tagName == "IMG" && attributeName == "src");
  1753. if (originURL) {
  1754. if (content == EMPTY_DATA_URI) {
  1755. try {
  1756. originURL = util.resolveURL(originURL, baseURI);
  1757. } catch (error) {
  1758. // ignored
  1759. }
  1760. try {
  1761. resourceURL = originURL;
  1762. content = (await util.getContent(resourceURL, {
  1763. asBinary: true,
  1764. expectedType: "image",
  1765. maxResourceSize: options.maxResourceSize,
  1766. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1767. frameId: options.windowId,
  1768. resourceReferrer: options.resourceReferrer
  1769. })).data;
  1770. } catch (error) {
  1771. // ignored
  1772. }
  1773. }
  1774. }
  1775. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1776. resourceElement.remove();
  1777. } else {
  1778. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1779. if (!forbiddenPrefixFound) {
  1780. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1781. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1782. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1783. cssVariables.set(indexResource, content);
  1784. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1785. styles.set(resourceElement, declarationList);
  1786. } else {
  1787. resourceElement.setAttribute(attributeName, content);
  1788. }
  1789. } else {
  1790. resourceElement.setAttribute(attributeName, content);
  1791. }
  1792. }
  1793. }
  1794. }
  1795. }
  1796. }
  1797. }
  1798. }));
  1799. }
  1800. static async processXLinks(resourceElements, doc, baseURI, options, batchRequest) {
  1801. let attributeName = "xlink:href";
  1802. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1803. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1804. if (originalResourceURL == null) {
  1805. attributeName = "href";
  1806. originalResourceURL = resourceElement.getAttribute(attributeName);
  1807. }
  1808. let resourceURL = normalizeURL(originalResourceURL);
  1809. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1810. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1811. try {
  1812. resourceURL = util.resolveURL(resourceURL, baseURI);
  1813. } catch (error) {
  1814. // ignored
  1815. }
  1816. if (testValidURL(resourceURL)) {
  1817. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1818. if (hashMatch && hashMatch[0]) {
  1819. if (originalResourceURL.startsWith(baseURI + "#")) {
  1820. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1821. } else {
  1822. const response = await batchRequest.addURL(resourceURL);
  1823. const svgDoc = util.parseSVGContent(response.content);
  1824. try {
  1825. const spriteElement = svgDoc.querySelector(hashMatch[0]);
  1826. if (spriteElement.tagName.toLowerCase() == "symbol") {
  1827. resourceElement.after(...spriteElement.childNodes);
  1828. resourceElement.remove();
  1829. } else {
  1830. resourceElement.parentElement.replaceChild(spriteElement, resourceElement);
  1831. }
  1832. } catch (error) {
  1833. // ignored
  1834. }
  1835. }
  1836. } else {
  1837. const content = await batchRequest.addURL(resourceURL);
  1838. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1839. }
  1840. }
  1841. } else if (resourceURL == options.url) {
  1842. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1843. }
  1844. }));
  1845. }
  1846. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1847. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1848. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1849. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1850. let resourceURL = normalizeURL(srcsetValue.url);
  1851. if (!testIgnoredPath(resourceURL)) {
  1852. if (testValidPath(resourceURL)) {
  1853. try {
  1854. resourceURL = util.resolveURL(resourceURL, baseURI);
  1855. } catch (error) {
  1856. // ignored
  1857. }
  1858. if (testValidURL(resourceURL)) {
  1859. const { content } = await batchRequest.addURL(resourceURL, true, "image");
  1860. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1861. if (forbiddenPrefixFound) {
  1862. return "";
  1863. }
  1864. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1865. } else {
  1866. return "";
  1867. }
  1868. } else {
  1869. return "";
  1870. }
  1871. } else {
  1872. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1873. }
  1874. }));
  1875. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1876. }));
  1877. }
  1878. static replaceImageSource(imgElement, variableName, options) {
  1879. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1880. if (attributeValue) {
  1881. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1882. if (imageData && imageData.replaceable) {
  1883. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1884. const backgroundStyle = {};
  1885. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1886. if (backgroundSize) {
  1887. backgroundStyle["background-size"] = imageData.objectFit;
  1888. }
  1889. if (imageData.objectPosition) {
  1890. backgroundStyle["background-position"] = imageData.objectPosition;
  1891. }
  1892. if (imageData.backgroundColor) {
  1893. backgroundStyle["background-color"] = imageData.backgroundColor;
  1894. }
  1895. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1896. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1897. return true;
  1898. }
  1899. }
  1900. }
  1901. }
  1902. // ----
  1903. // Util
  1904. // ----
  1905. const BLOB_URI_PREFIX = "blob:";
  1906. const HTTP_URI_PREFIX = /^https?:\/\//;
  1907. const FILE_URI_PREFIX = /^file:\/\//;
  1908. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1909. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1910. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1911. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1912. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1913. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1914. 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;
  1915. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1916. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1917. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1918. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1919. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1920. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1921. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1922. function getUpdatedResourceContent(resourceURL, content, options) {
  1923. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1924. options.updatedResources[resourceURL].retrieved = true;
  1925. return options.updatedResources[resourceURL].content;
  1926. } else {
  1927. return content.data || "";
  1928. }
  1929. }
  1930. function normalizeURL(url) {
  1931. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1932. return url;
  1933. } else {
  1934. return url.split("#")[0];
  1935. }
  1936. }
  1937. function getCSSValue(value) {
  1938. let result = "";
  1939. try {
  1940. result = cssTree.generate(value);
  1941. } catch (error) {
  1942. // ignored
  1943. }
  1944. return result;
  1945. }
  1946. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  1947. const stylesheetCharset = getCharset(stylesheetContent);
  1948. if (stylesheetCharset) {
  1949. return stylesheetCharset == charset.toLowerCase();
  1950. } else {
  1951. return true;
  1952. }
  1953. }
  1954. function getCharset(stylesheetContent) {
  1955. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  1956. if (match && match[1]) {
  1957. return match[1].toLowerCase().trim();
  1958. }
  1959. }
  1960. function getOnEventAttributeNames(doc) {
  1961. const element = doc.createElement("div");
  1962. const attributeNames = [];
  1963. for (const propertyName in element) {
  1964. if (propertyName.startsWith("on")) {
  1965. attributeNames.push(propertyName);
  1966. }
  1967. }
  1968. attributeNames.push("onunload");
  1969. return attributeNames;
  1970. }
  1971. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1972. let maxLength;
  1973. if (template) {
  1974. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1975. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1976. if (template.match(replaceRegExp)) {
  1977. const matchedLength = template.match(replaceRegExp)[0];
  1978. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1979. if (isNaN(maxLength) || maxLength <= 0) {
  1980. maxLength = null;
  1981. }
  1982. } else {
  1983. replaceRegExp = new RegExp(regExpVariable, "g");
  1984. }
  1985. if (template.match(replaceRegExp)) {
  1986. let value = await valueGetter();
  1987. if (!dontReplaceSlash) {
  1988. value = value.replace(/\/+/g, replacementCharacter);
  1989. }
  1990. if (maxLength) {
  1991. value = await util.truncateText(value, maxLength);
  1992. }
  1993. return template.replace(replaceRegExp, value);
  1994. }
  1995. }
  1996. return template;
  1997. }
  1998. function getLastSegment(url, replacementCharacter) {
  1999. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2000. if (!lastSegment) {
  2001. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  2002. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2003. }
  2004. if (!lastSegment) {
  2005. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  2006. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2007. }
  2008. if (!lastSegment) {
  2009. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  2010. }
  2011. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  2012. if (lastSegmentMatch && lastSegmentMatch[1]) {
  2013. lastSegment = lastSegmentMatch[1];
  2014. }
  2015. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  2016. return lastSegment;
  2017. }
  2018. function getRegExp(string) {
  2019. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  2020. }
  2021. function getUrlFunctions(stylesheetContent, unique) {
  2022. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  2023. if (unique) {
  2024. return [...new Set(result)];
  2025. } else {
  2026. return result;
  2027. }
  2028. }
  2029. function getImportFunctions(stylesheetContent) {
  2030. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  2031. }
  2032. function findShortcutIcon(shortcutIcons) {
  2033. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  2034. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  2035. return shortcutIcons[0];
  2036. }
  2037. function matchURL(stylesheetContent) {
  2038. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  2039. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  2040. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  2041. return match && match[1];
  2042. }
  2043. function testIgnoredPath(resourceURL) {
  2044. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  2045. }
  2046. function testValidPath(resourceURL) {
  2047. return resourceURL && !resourceURL.match(EMPTY_URL);
  2048. }
  2049. function testValidURL(resourceURL) {
  2050. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  2051. }
  2052. function matchImport(stylesheetContent) {
  2053. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  2054. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  2055. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2056. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2057. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2058. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2059. if (match) {
  2060. const [, resourceURL, media] = match;
  2061. return { resourceURL, media };
  2062. }
  2063. }
  2064. function removeCssComments(stylesheetContent) {
  2065. try {
  2066. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2067. } catch (error) {
  2068. let start, end;
  2069. do {
  2070. start = stylesheetContent.indexOf("/*");
  2071. end = stylesheetContent.indexOf("*/", start + 2);
  2072. if (start != -1 && end != -1) {
  2073. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2074. }
  2075. } while (start != -1 && end != -1);
  2076. return stylesheetContent;
  2077. }
  2078. }
  2079. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2080. if (mediaQuery) {
  2081. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2082. } else {
  2083. return stylesheetContent;
  2084. }
  2085. }
  2086. function log(...args) {
  2087. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2088. }
  2089. // -----
  2090. // Stats
  2091. // -----
  2092. const STATS_DEFAULT_VALUES = {
  2093. discarded: {
  2094. "HTML bytes": 0,
  2095. "hidden elements": 0,
  2096. "HTML imports": 0,
  2097. scripts: 0,
  2098. objects: 0,
  2099. "audio sources": 0,
  2100. "video sources": 0,
  2101. frames: 0,
  2102. "CSS rules": 0,
  2103. canvas: 0,
  2104. stylesheets: 0,
  2105. resources: 0,
  2106. medias: 0
  2107. },
  2108. processed: {
  2109. "HTML bytes": 0,
  2110. "hidden elements": 0,
  2111. "HTML imports": 0,
  2112. scripts: 0,
  2113. objects: 0,
  2114. "audio sources": 0,
  2115. "video sources": 0,
  2116. frames: 0,
  2117. "CSS rules": 0,
  2118. canvas: 0,
  2119. stylesheets: 0,
  2120. resources: 0,
  2121. medias: 0
  2122. }
  2123. };
  2124. class Stats {
  2125. constructor(options) {
  2126. this.options = options;
  2127. if (options.displayStats) {
  2128. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2129. }
  2130. }
  2131. set(type, subType, value) {
  2132. if (this.options.displayStats) {
  2133. this.data[type][subType] = value;
  2134. }
  2135. }
  2136. add(type, subType, value) {
  2137. if (this.options.displayStats) {
  2138. this.data[type][subType] += value;
  2139. }
  2140. }
  2141. addAll(pageData) {
  2142. if (this.options.displayStats) {
  2143. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2144. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2145. }
  2146. }
  2147. }
  2148. export {
  2149. getClass
  2150. };