single-file-core.js 83 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030
  1. /*
  2. * Copyright 2010-2019 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree, sessionId = 0;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. options.insertFaviconLink = true;
  34. if (options.sessionId === undefined) {
  35. options.sessionId = sessionId;
  36. sessionId++;
  37. }
  38. }
  39. async run() {
  40. if (this.options.userScriptEnabled) {
  41. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  42. }
  43. this.runner = new Runner(this.options, true);
  44. await this.runner.loadPage();
  45. await this.runner.initialize();
  46. if (this.options.userScriptEnabled) {
  47. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  48. }
  49. await this.runner.run();
  50. }
  51. cancel() {
  52. this.cancelled = true;
  53. if (this.runner) {
  54. this.runner.cancel();
  55. }
  56. }
  57. async getPageData() {
  58. return this.runner.getPageData();
  59. }
  60. }
  61. // -------------
  62. // ProgressEvent
  63. // -------------
  64. const PAGE_LOADING = "page-loading";
  65. const PAGE_LOADED = "page-loaded";
  66. const RESOURCES_INITIALIZING = "resource-initializing";
  67. const RESOURCES_INITIALIZED = "resources-initialized";
  68. const RESOURCE_LOADED = "resource-loaded";
  69. const PAGE_ENDED = "page-ended";
  70. const STAGE_STARTED = "stage-started";
  71. const STAGE_ENDED = "stage-ended";
  72. const STAGE_TASK_STARTED = "stage-task-started";
  73. const STAGE_TASK_ENDED = "stage-task-ended";
  74. class ProgressEvent {
  75. constructor(type, detail) {
  76. 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 };
  77. }
  78. }
  79. // ------
  80. // Runner
  81. // ------
  82. const RESOLVE_URLS_STAGE = 0;
  83. const REPLACE_DATA_STAGE = 1;
  84. const REPLACE_DOCS_STAGE = 2;
  85. const POST_PROCESS_STAGE = 3;
  86. const STAGES = [{
  87. sequential: [
  88. { action: "preProcessPage" },
  89. { action: "replaceStyleContents" },
  90. { action: "resetCharsetMeta" },
  91. { option: "insertFaviconLink", action: "insertFaviconLink" },
  92. { action: "replaceCanvasElements" },
  93. { action: "insertFonts" },
  94. { action: "insertShadowRootContents" },
  95. { action: "setInputValues" },
  96. { option: "removeScripts", action: "removeScripts" },
  97. { option: "selected", action: "removeUnselectedElements" },
  98. { option: "removeVideoSrc", action: "insertVideoPosters" },
  99. { option: "removeFrames", action: "removeFrames" },
  100. { option: "removeVideoSrc", action: "removeVideoSources" },
  101. { option: "removeAudioSrc", action: "removeAudioSources" },
  102. { action: "removeDiscardedResources" },
  103. { option: "removeHiddenElements", action: "removeHiddenElements" },
  104. { action: "resolveHrefs" },
  105. { action: "resolveStyleAttributeURLs" }
  106. ],
  107. parallel: [
  108. { action: "resolveStylesheetURLs" },
  109. { option: "!removeFrames", action: "resolveFrameURLs" },
  110. { action: "resolveHtmlImportURLs" }
  111. ]
  112. }, {
  113. sequential: [
  114. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  115. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  116. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  117. ],
  118. parallel: [
  119. { action: "processStylesheets" },
  120. { action: "processStyleAttributes" },
  121. { action: "processPageResources" },
  122. { option: "!removeScripts", action: "processScripts" }
  123. ]
  124. }, {
  125. sequential: [
  126. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  127. ],
  128. parallel: [
  129. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  130. { option: "!removeFrames", action: "processFrames" },
  131. { option: "!removeImports", action: "processHtmlImports" },
  132. ]
  133. }, {
  134. sequential: [
  135. { action: "replaceStylesheets" },
  136. { action: "replaceStyleAttributes" },
  137. { action: "insertVariables" },
  138. { option: "compressHTML", action: "compressHTML" },
  139. { action: "cleanupPage" }
  140. ],
  141. parallel: [
  142. { option: "enableMaff", action: "insertMAFFMetaData" },
  143. { action: "setDocInfo" }
  144. ]
  145. }];
  146. class Runner {
  147. constructor(options, root) {
  148. this.root = root;
  149. this.options = options;
  150. this.options.url = this.options.url || (this.options.doc && this.options.doc.location.href);
  151. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  152. this.options.rootDocument = root;
  153. this.options.updatedResources = this.options.updatedResources || {};
  154. this.batchRequest = new BatchRequest();
  155. this.processor = new Processor(options, this.batchRequest);
  156. if (this.options.doc) {
  157. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  158. this.options.canvases = docData.canvases;
  159. this.options.fonts = docData.fonts;
  160. this.options.stylesheets = docData.stylesheets;
  161. this.options.images = docData.images;
  162. this.options.posters = docData.posters;
  163. this.options.usedFonts = docData.usedFonts;
  164. this.options.shadowRoots = docData.shadowRoots;
  165. this.options.imports = docData.imports;
  166. this.options.referrer = docData.referrer;
  167. }
  168. if (this.options.saveRawPage) {
  169. this.options.removeFrames = true;
  170. }
  171. this.options.content = this.options.content || (this.options.doc ? util.serialize(this.options.doc) : null);
  172. this.onprogress = options.onprogress || (() => { });
  173. }
  174. async loadPage() {
  175. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  176. await this.processor.loadPage(this.options.content);
  177. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  178. }
  179. async initialize() {
  180. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  181. await this.executeStage(RESOLVE_URLS_STAGE);
  182. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  183. if (this.options.doc) {
  184. util.postProcessDoc(this.options.doc, this.markedElements);
  185. }
  186. this.options.doc = null;
  187. this.options.win = null;
  188. }
  189. cancel() {
  190. this.cancelled = true;
  191. this.batchRequest.cancel();
  192. if (this.root) {
  193. if (this.options.frames) {
  194. this.options.frames.forEach(cancelRunner);
  195. }
  196. if (this.options.imports) {
  197. this.options.imports.forEach(cancelRunner);
  198. }
  199. }
  200. function cancelRunner(resourceData) {
  201. if (resourceData.runner) {
  202. resourceData.runner.cancel();
  203. }
  204. }
  205. }
  206. async run() {
  207. if (this.root) {
  208. this.processor.initialize(this.batchRequest);
  209. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  210. }
  211. await this.batchRequest.run(detail => {
  212. detail.pageURL = this.options.url;
  213. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  214. }, this.options);
  215. await this.pendingPromises;
  216. await this.executeStage(REPLACE_DOCS_STAGE);
  217. await this.executeStage(POST_PROCESS_STAGE);
  218. this.processor.finalize();
  219. }
  220. getDocument() {
  221. return this.processor.doc;
  222. }
  223. getStyleSheets() {
  224. return this.processor.stylesheets;
  225. }
  226. async getPageData() {
  227. if (this.root) {
  228. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  229. }
  230. return this.processor.getPageData();
  231. }
  232. async executeStage(step) {
  233. if (DEBUG) {
  234. log("**** STARTED STAGE", step, "****");
  235. }
  236. const frame = !this.root;
  237. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  238. STAGES[step].sequential.forEach(task => {
  239. let startTime;
  240. if (DEBUG) {
  241. startTime = Date.now();
  242. log(" -- STARTED task =", task.action);
  243. }
  244. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  245. if (!this.cancelled) {
  246. this.executeTask(task);
  247. }
  248. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  249. if (DEBUG) {
  250. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  251. }
  252. });
  253. let parallelTasksPromise;
  254. if (STAGES[step].parallel) {
  255. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  256. let startTime;
  257. if (DEBUG) {
  258. startTime = Date.now();
  259. log(" // STARTED task =", task.action);
  260. }
  261. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  262. if (!this.cancelled) {
  263. await this.executeTask(task);
  264. }
  265. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  266. if (DEBUG) {
  267. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  268. }
  269. }));
  270. } else {
  271. parallelTasksPromise = Promise.resolve();
  272. }
  273. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  274. if (DEBUG) {
  275. log("**** ENDED STAGE", step, "****");
  276. }
  277. return parallelTasksPromise;
  278. }
  279. executeTask(task) {
  280. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  281. return this.processor[task.action]();
  282. }
  283. }
  284. }
  285. // ------------
  286. // BatchRequest
  287. // ------------
  288. class BatchRequest {
  289. constructor() {
  290. this.requests = new Map();
  291. this.duplicates = new Map();
  292. }
  293. async addURL(resourceURL, asBinary, groupDuplicates) {
  294. return new Promise((resolve, reject) => {
  295. const requestKey = JSON.stringify([resourceURL, asBinary]);
  296. let resourceRequests = this.requests.get(requestKey);
  297. if (!resourceRequests) {
  298. resourceRequests = [];
  299. this.requests.set(requestKey, resourceRequests);
  300. }
  301. const callbacks = { resolve, reject };
  302. resourceRequests.push(callbacks);
  303. if (groupDuplicates) {
  304. let duplicateRequests = this.duplicates.get(requestKey);
  305. if (!duplicateRequests) {
  306. duplicateRequests = [];
  307. this.duplicates.set(requestKey, duplicateRequests);
  308. }
  309. duplicateRequests.push(callbacks);
  310. }
  311. });
  312. }
  313. getMaxResources() {
  314. return this.requests.size;
  315. }
  316. async run(onloadListener, options) {
  317. const resourceURLs = [...this.requests.keys()];
  318. let indexResource = 0;
  319. return Promise.all(resourceURLs.map(async requestKey => {
  320. const [resourceURL, asBinary] = JSON.parse(requestKey);
  321. const resourceRequests = this.requests.get(requestKey);
  322. try {
  323. const content = await util.getContent(resourceURL, {
  324. asBinary,
  325. maxResourceSize: options.maxResourceSize,
  326. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  327. frameId: options.windowId
  328. });
  329. indexResource = indexResource + 1;
  330. onloadListener({ url: resourceURL });
  331. if (!this.cancelled) {
  332. resourceRequests.forEach(callbacks => {
  333. const duplicateCallbacks = this.duplicates.get(requestKey);
  334. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  335. callbacks.resolve({ content: content.data, indexResource, duplicate });
  336. });
  337. }
  338. } catch (error) {
  339. indexResource = indexResource + 1;
  340. onloadListener({ url: resourceURL });
  341. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  342. }
  343. this.requests.delete(requestKey);
  344. }));
  345. }
  346. cancel() {
  347. this.cancelled = true;
  348. const resourceURLs = [...this.requests.keys()];
  349. resourceURLs.forEach(requestKey => {
  350. const resourceRequests = this.requests.get(requestKey);
  351. resourceRequests.forEach(callbacks => callbacks.reject());
  352. this.requests.delete(requestKey);
  353. });
  354. }
  355. }
  356. // ---------
  357. // Processor
  358. // ---------
  359. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  360. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  361. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  362. const SCRIPT_TAG_FOUND = /<script/gi;
  363. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  364. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  365. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  366. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  367. class Processor {
  368. constructor(options, batchRequest) {
  369. this.options = options;
  370. this.stats = new Stats(options);
  371. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  372. this.batchRequest = batchRequest;
  373. this.stylesheets = new Map();
  374. this.styles = new Map();
  375. this.cssVariables = new Map();
  376. }
  377. initialize() {
  378. this.options.saveDate = new Date();
  379. this.options.saveUrl = this.options.url;
  380. if (this.options.enableMaff) {
  381. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  382. }
  383. this.maxResources = this.batchRequest.getMaxResources();
  384. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  385. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  386. }
  387. if (!this.options.removeImports && this.options.imports) {
  388. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  389. }
  390. this.stats.set("processed", "resources", this.maxResources);
  391. }
  392. async loadPage(pageContent, charset) {
  393. let content;
  394. if (!pageContent || this.options.saveRawPage) {
  395. content = await util.getContent(this.baseURI, {
  396. maxResourceSize: this.options.maxResourceSize,
  397. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  398. charset,
  399. frameId: this.options.windowId
  400. });
  401. pageContent = content.data;
  402. }
  403. this.doc = util.parseDocContent(pageContent, this.baseURI);
  404. if (this.options.saveRawPage) {
  405. let charset;
  406. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  407. const charsetDeclaration = element.content.split(";")[1];
  408. if (charsetDeclaration && !charset) {
  409. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  410. }
  411. });
  412. if (charset && charset != content.charset) {
  413. return this.loadPage(pageContent, charset);
  414. }
  415. }
  416. this.workStyleElement = this.doc.createElement("style");
  417. this.doc.body.appendChild(this.workStyleElement);
  418. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  419. }
  420. finalize() {
  421. if (this.workStyleElement.parentNode) {
  422. this.workStyleElement.remove();
  423. }
  424. }
  425. async getPageData() {
  426. util.postProcessDoc(this.doc);
  427. const url = util.parseURL(this.baseURI);
  428. if (this.options.insertSingleFileComment) {
  429. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  430. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  431. " \n url: " + this.options.saveUrl +
  432. " \n saved date: " + this.options.saveDate +
  433. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  434. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  435. }
  436. let size;
  437. if (this.options.displayStats) {
  438. size = util.getContentSize(this.doc.documentElement.outerHTML);
  439. }
  440. const content = util.serialize(this.doc, this.options.compressHTML);
  441. if (this.options.displayStats) {
  442. const contentSize = util.getContentSize(content);
  443. this.stats.set("processed", "HTML bytes", contentSize);
  444. this.stats.add("discarded", "HTML bytes", size - contentSize);
  445. }
  446. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  447. const replacementCharacter = this.options.filenameReplacementCharacter;
  448. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  449. if (!this.options.backgroundSave) {
  450. filename = filename.replace(/\//g, replacementCharacter);
  451. }
  452. if (util.getContentSize(filename) > this.options.filenameMaxLength) {
  453. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  454. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  455. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  456. filename = filename + "…" + extension;
  457. }
  458. if (!filename) {
  459. filename = "Unnamed page";
  460. }
  461. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  462. return {
  463. stats: this.stats.data,
  464. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  465. filename,
  466. content
  467. };
  468. }
  469. preProcessPage() {
  470. if (this.options.win) {
  471. 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));
  472. }
  473. if (this.options.images && !this.options.saveRawPage) {
  474. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  475. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  476. if (attributeValue) {
  477. const imageData = this.options.images[Number(attributeValue)];
  478. if (imageData) {
  479. if (this.options.removeHiddenElements && imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) {
  480. imgElement.setAttribute("src", EMPTY_IMAGE);
  481. } else if (imageData.currentSrc) {
  482. imgElement.setAttribute("src", imageData.currentSrc);
  483. }
  484. if (this.options.loadDeferredImages) {
  485. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  486. imageData.src = imgElement.dataset.src;
  487. imgElement.setAttribute("src", imgElement.dataset.src);
  488. imgElement.removeAttribute("data-src");
  489. }
  490. }
  491. }
  492. }
  493. });
  494. if (this.options.loadDeferredImages) {
  495. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  496. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  497. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  498. imgElement.removeAttribute("data-srcset");
  499. }
  500. });
  501. }
  502. }
  503. }
  504. replaceStyleContents() {
  505. if (this.options.stylesheets) {
  506. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  507. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  508. if (attributeValue) {
  509. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  510. if (stylesheetContent) {
  511. styleElement.textContent = stylesheetContent;
  512. }
  513. }
  514. });
  515. }
  516. }
  517. removeUnselectedElements() {
  518. removeUnmarkedElements(this.doc.body);
  519. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  520. function removeUnmarkedElements(element) {
  521. let selectedElementFound = false;
  522. Array.from(element.childNodes).forEach(node => {
  523. if (node.nodeType == 1) {
  524. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  525. selectedElementFound = selectedElementFound || isSelectedElement;
  526. if (isSelectedElement) {
  527. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  528. removeUnmarkedElements(node);
  529. } else if (selectedElementFound) {
  530. removeNode(node);
  531. } else {
  532. hideNode(node);
  533. }
  534. }
  535. });
  536. }
  537. function removeNode(node) {
  538. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  539. node.remove();
  540. } else {
  541. hideNode(node);
  542. }
  543. }
  544. function hideNode(node) {
  545. if (canHideNode(node)) {
  546. node.style.setProperty("display", "none", "important");
  547. Array.from(node.childNodes).forEach(removeNode);
  548. }
  549. }
  550. function canHideNode(node) {
  551. if (node.nodeType == 1) {
  552. const tagName = node.tagName && node.tagName.toLowerCase();
  553. return (tagName != "svg" && tagName != "style" && tagName != "link");
  554. }
  555. }
  556. }
  557. insertVideoPosters() {
  558. if (this.options.posters) {
  559. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  560. let videoElement;
  561. if (element.tagName == "VIDEO") {
  562. videoElement = element;
  563. } else {
  564. videoElement = element.parentElement;
  565. }
  566. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  567. if (attributeValue) {
  568. const posterURL = this.options.posters[Number(attributeValue)];
  569. if (!videoElement.poster && posterURL) {
  570. videoElement.setAttribute("poster", posterURL);
  571. }
  572. }
  573. });
  574. }
  575. }
  576. removeFrames() {
  577. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  578. this.stats.set("discarded", "frames", frameElements.length);
  579. this.stats.set("processed", "frames", frameElements.length);
  580. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  581. }
  582. removeScripts() {
  583. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  584. this.doc.querySelectorAll("[href]").forEach(element => {
  585. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  586. element.setAttribute("href", "");
  587. }
  588. });
  589. this.doc.querySelectorAll("[src]").forEach(element => {
  590. if (element.src && element.src.match(/^\s*javascript:/)) {
  591. element.removeAttribute("src");
  592. }
  593. });
  594. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  595. this.stats.set("discarded", "scripts", scriptElements.length);
  596. this.stats.set("processed", "scripts", scriptElements.length);
  597. scriptElements.forEach(element => element.remove());
  598. }
  599. removeVideoSources() {
  600. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  601. this.stats.set("discarded", "video sources", videoSourceElements.length);
  602. this.stats.set("processed", "video sources", videoSourceElements.length);
  603. videoSourceElements.forEach(element => {
  604. if (element.tagName == "SOURCE") {
  605. element.remove();
  606. } else {
  607. videoSourceElements.forEach(element => element.removeAttribute("src"));
  608. }
  609. });
  610. }
  611. removeAudioSources() {
  612. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  613. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  614. this.stats.set("processed", "audio sources", audioSourceElements.length);
  615. audioSourceElements.forEach(element => {
  616. if (element.tagName == "SOURCE") {
  617. element.remove();
  618. } else {
  619. audioSourceElements.forEach(element => element.removeAttribute("src"));
  620. }
  621. });
  622. }
  623. removeDiscardedResources() {
  624. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  625. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  626. 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\"])");
  627. this.stats.set("discarded", "objects", objectElements.length);
  628. this.stats.set("processed", "objects", objectElements.length);
  629. objectElements.forEach(element => element.remove());
  630. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  631. replacedAttributeValue.forEach(element => {
  632. let regExp;
  633. if (this.options.removeScripts) {
  634. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  635. } else {
  636. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  637. }
  638. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  639. if (relValue.length) {
  640. element.setAttribute("rel", relValue);
  641. } else {
  642. element.remove();
  643. }
  644. });
  645. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  646. if (this.options.compressHTML) {
  647. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  648. }
  649. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  650. }
  651. resetCharsetMeta() {
  652. let charset;
  653. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  654. const charsetDeclaration = element.content.split(";")[1];
  655. if (charsetDeclaration && !charset) {
  656. charset = charsetDeclaration.split("=")[1];
  657. if (charset) {
  658. this.charset = charset.trim().toLowerCase();
  659. }
  660. }
  661. element.remove();
  662. });
  663. const metaElement = this.doc.createElement("meta");
  664. metaElement.setAttribute("charset", "utf-8");
  665. if (this.doc.head.firstChild) {
  666. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  667. } else {
  668. this.doc.head.appendChild(metaElement);
  669. }
  670. }
  671. setInputValues() {
  672. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  673. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  674. input.setAttribute("value", value || "");
  675. });
  676. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  677. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  678. if (value == "true") {
  679. input.setAttribute("checked", "");
  680. }
  681. });
  682. this.doc.querySelectorAll("textarea").forEach(textarea => {
  683. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  684. textarea.textContent = value || "";
  685. });
  686. this.doc.querySelectorAll("select").forEach(select => {
  687. select.querySelectorAll("option").forEach(option => {
  688. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  689. if (selected) {
  690. option.setAttribute("selected", "");
  691. }
  692. });
  693. });
  694. }
  695. insertFaviconLink() {
  696. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  697. if (!faviconElement) {
  698. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  699. }
  700. if (!faviconElement) {
  701. faviconElement = this.doc.createElement("link");
  702. faviconElement.setAttribute("type", "image/x-icon");
  703. faviconElement.setAttribute("rel", "shortcut icon");
  704. faviconElement.setAttribute("href", "/favicon.ico");
  705. }
  706. this.doc.head.appendChild(faviconElement);
  707. }
  708. replaceCanvasElements() {
  709. if (this.options.canvases) {
  710. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  711. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  712. if (attributeValue) {
  713. const canvasData = this.options.canvases[Number(attributeValue)];
  714. if (canvasData) {
  715. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  716. this.stats.add("processed", "canvas", 1);
  717. }
  718. }
  719. });
  720. }
  721. }
  722. insertFonts() {
  723. if (this.options.fonts && this.options.fonts.length) {
  724. let stylesheetContent = "";
  725. this.options.fonts.forEach(fontData => {
  726. if (fontData["font-family"] && fontData.src) {
  727. stylesheetContent += "@font-face{";
  728. let stylesContent = "";
  729. Object.keys(fontData).forEach(fontStyle => {
  730. if (stylesContent) {
  731. stylesContent += ";";
  732. }
  733. stylesContent += fontStyle + ":" + fontData[fontStyle];
  734. });
  735. stylesheetContent += stylesContent + "}";
  736. }
  737. });
  738. if (stylesheetContent) {
  739. const styleElement = this.doc.createElement("style");
  740. styleElement.textContent = stylesheetContent;
  741. const existingStyleElement = this.doc.querySelector("style");
  742. if (existingStyleElement) {
  743. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  744. } else {
  745. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  746. }
  747. }
  748. }
  749. }
  750. removeHiddenElements() {
  751. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  752. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  753. this.stats.set("discarded", "hidden elements", removedElements.length);
  754. this.stats.set("processed", "hidden elements", removedElements.length);
  755. if (hiddenElements.length) {
  756. const styleElement = this.doc.createElement("style");
  757. styleElement.textContent = ".sf-hidden{display:none!important;}";
  758. this.doc.head.appendChild(styleElement);
  759. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  760. }
  761. removedElements.forEach(element => element.remove());
  762. }
  763. resolveHrefs() {
  764. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  765. const href = element.getAttribute("href").trim();
  766. if (!Util.testIgnoredPath(href)) {
  767. let resolvedURL;
  768. try {
  769. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  770. } catch (error) {
  771. // ignored
  772. }
  773. if (resolvedURL) {
  774. const url = Util.normalizeURL(this.options.url);
  775. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  776. resolvedURL = resolvedURL.substring(url.length);
  777. }
  778. try {
  779. element.setAttribute("href", resolvedURL);
  780. } catch (error) {
  781. // ignored
  782. }
  783. }
  784. }
  785. });
  786. }
  787. resolveStyleAttributeURLs() {
  788. this.doc.querySelectorAll("[style]").forEach(element => {
  789. let styleContent = element.getAttribute("style");
  790. if (this.options.compressCSS) {
  791. styleContent = util.compressCSS(styleContent);
  792. }
  793. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  794. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  795. this.styles.set(element, declarationList);
  796. });
  797. }
  798. async resolveStylesheetURLs() {
  799. const options = {
  800. maxResourceSize: this.options.maxResourceSize,
  801. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  802. url: this.options.url,
  803. charset: this.charset,
  804. compressCSS: this.options.compressCSS,
  805. updatedResources: this.options.updatedResources,
  806. rootDocument: this.options.rootDocument,
  807. frameId: this.options.windowId
  808. };
  809. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  810. .map(async element => {
  811. let mediaText;
  812. if (element.media) {
  813. mediaText = element.media.toLowerCase();
  814. }
  815. const stylesheetInfo = { mediaText };
  816. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  817. stylesheetInfo.scoped = true;
  818. }
  819. if (element.tagName.toLowerCase() == "link") {
  820. if (element.charset) {
  821. options.charset = element.charset;
  822. }
  823. }
  824. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  825. }));
  826. if (options.rootDocument) {
  827. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  828. await Promise.all(newResources.map(async resource => {
  829. resource.retrieved = true;
  830. const stylesheetInfo = {};
  831. const element = this.doc.createElement("style");
  832. this.doc.body.appendChild(element);
  833. element.textContent = resource.content;
  834. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  835. }));
  836. }
  837. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  838. stylesheets.set(element, stylesheetInfo);
  839. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  840. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  841. if (match && match[1] && match[1] != options.charset) {
  842. options.charset = match[1];
  843. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  844. }
  845. let stylesheet;
  846. try {
  847. stylesheet = cssTree.parse(Util.removeCssComments(stylesheetContent));
  848. } catch (error) {
  849. // ignored
  850. }
  851. if (stylesheet && stylesheet.children) {
  852. if (options.compressCSS) {
  853. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  854. }
  855. stylesheetInfo.stylesheet = stylesheet;
  856. } else {
  857. stylesheets.delete(element);
  858. }
  859. }
  860. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  861. let content;
  862. if (element.tagName.toLowerCase() == "link") {
  863. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  864. } else {
  865. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  866. }
  867. return content || "";
  868. }
  869. }
  870. async resolveFrameURLs() {
  871. if (!this.options.saveRawPage && this.options.frames) {
  872. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  873. await Promise.all(frameElements.map(async frameElement => {
  874. if (frameElement.tagName == "OBJECT") {
  875. frameElement.setAttribute("data", "data:text/html,");
  876. } else {
  877. frameElement.removeAttribute("src");
  878. frameElement.removeAttribute("srcdoc");
  879. }
  880. Array.from(frameElement.childNodes).forEach(node => node.remove());
  881. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  882. if (frameWindowId) {
  883. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  884. if (frameData) {
  885. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  886. }
  887. }
  888. }));
  889. }
  890. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  891. options.insertSingleFileComment = false;
  892. options.insertFaviconLink = false;
  893. options.doc = null;
  894. options.win = null;
  895. options.url = frameData.baseURI;
  896. options.windowId = frameWindowId;
  897. if (frameData.content) {
  898. options.content = frameData.content;
  899. options.canvases = frameData.canvases;
  900. options.fonts = frameData.fonts;
  901. options.stylesheets = frameData.stylesheets;
  902. options.images = frameData.images;
  903. options.posters = frameData.posters;
  904. options.usedFonts = frameData.usedFonts;
  905. options.shadowRoots = frameData.shadowRoots;
  906. options.imports = frameData.imports;
  907. frameData.runner = new Runner(options);
  908. frameData.frameElement = frameElement;
  909. await frameData.runner.loadPage();
  910. await frameData.runner.initialize();
  911. frameData.maxResources = batchRequest.getMaxResources();
  912. }
  913. }
  914. }
  915. insertShadowRootContents() {
  916. const doc = this.doc;
  917. const options = this.options;
  918. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  919. processElement(this.doc);
  920. if (this.options.removeScripts) {
  921. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  922. }
  923. const scriptElement = doc.createElement("script");
  924. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  925. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{if (!element.parentElement.shadowRoot) {const shadowRoot=element.parentElement.attachShadow({mode:element.getAttribute("${SHADOW_MODE_ATTRIBUTE_NAME}"),delegatesFocus:Boolean(element.getAttribute("${SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME}"))});shadowRoot.innerHTML=element.innerHTML;element.remove();processNode(shadowRoot)}})}})()`;
  926. doc.body.appendChild(scriptElement);
  927. }
  928. function processElement(element) {
  929. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  930. shadowRootElements.forEach(element => {
  931. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  932. if (attributeValue) {
  933. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  934. if (shadowRootData) {
  935. const templateElement = doc.createElement("template");
  936. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  937. if (shadowRootData.delegatesFocus) {
  938. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  939. }
  940. if (shadowRootData.adoptedStyleSheets) {
  941. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  942. const styleElement = doc.createElement("style");
  943. styleElement.textContent = stylesheetContent;
  944. templateElement.appendChild(styleElement);
  945. });
  946. }
  947. const shadowDoc = util.parseDocContent(shadowRootData.content);
  948. if (shadowDoc.head) {
  949. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  950. if (metaCharset) {
  951. metaCharset.remove();
  952. }
  953. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  954. }
  955. if (shadowDoc.body) {
  956. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  957. }
  958. processElement(templateElement);
  959. if (element.firstChild) {
  960. element.insertBefore(templateElement, element.firstChild);
  961. } else {
  962. element.appendChild(templateElement);
  963. }
  964. }
  965. }
  966. });
  967. }
  968. }
  969. async resolveHtmlImportURLs() {
  970. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  971. await Promise.all(linkElements.map(async linkElement => {
  972. const resourceURL = linkElement.href;
  973. linkElement.removeAttribute("href");
  974. const options = Object.create(this.options);
  975. options.insertSingleFileComment = false;
  976. options.insertFaviconLink = false;
  977. options.removeUnusedStyles = false;
  978. options.removeAlternativeMedias = false;
  979. options.removeUnusedFonts = false;
  980. options.removeHiddenElements = false;
  981. options.doc = null;
  982. options.win = null;
  983. options.url = resourceURL;
  984. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  985. if (attributeValue) {
  986. const importData = options.imports[Number(attributeValue)];
  987. if (importData) {
  988. options.content = importData.content;
  989. importData.runner = new Runner(options);
  990. await importData.runner.loadPage();
  991. await importData.runner.initialize();
  992. if (!options.removeImports) {
  993. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  994. }
  995. importData.runner.getStyleSheets().forEach(stylesheet => {
  996. const importedStyleElement = this.doc.createElement("style");
  997. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  998. this.stylesheets.set(importedStyleElement, stylesheet);
  999. });
  1000. }
  1001. }
  1002. if (options.removeImports) {
  1003. linkElement.remove();
  1004. this.stats.add("discarded", "HTML imports", 1);
  1005. }
  1006. }));
  1007. }
  1008. removeUnusedStyles() {
  1009. if (!this.mediaAllInfo) {
  1010. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1011. }
  1012. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1013. this.stats.set("processed", "CSS rules", stats.processed);
  1014. this.stats.set("discarded", "CSS rules", stats.discarded);
  1015. }
  1016. removeUnusedFonts() {
  1017. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1018. }
  1019. removeAlternativeMedias() {
  1020. const stats = util.minifyMedias(this.stylesheets);
  1021. this.stats.set("processed", "medias", stats.processed);
  1022. this.stats.set("discarded", "medias", stats.discarded);
  1023. }
  1024. async processStylesheets() {
  1025. await Promise.all(Array.from(this.stylesheets).map(([, stylesheetInfo]) =>
  1026. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1027. ));
  1028. }
  1029. async processStyleAttributes() {
  1030. return Promise.all(Array.from(this.styles).map(([, declarationList]) =>
  1031. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1032. ));
  1033. }
  1034. async processPageResources() {
  1035. const processAttributeArgs = [
  1036. ["link[href][rel*=\"icon\"]", "href", false, true],
  1037. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1038. ["img[src], input[src][type=image]", "src", true],
  1039. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1040. ["video[poster]", "poster"],
  1041. ["*[background]", "background"],
  1042. ["image", "xlink:href"],
  1043. ["image", "href"]
  1044. ];
  1045. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1046. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1047. );
  1048. resourcePromises = resourcePromises.concat([
  1049. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1050. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1051. ]);
  1052. if (!this.options.removeAudioSrc) {
  1053. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1054. }
  1055. if (!this.options.removeVideoSrc) {
  1056. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1057. }
  1058. await Promise.all(resourcePromises);
  1059. ProcessorHelper.processShortcutIcons(this.doc);
  1060. }
  1061. async processScripts() {
  1062. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1063. let resourceURL;
  1064. let scriptSrc;
  1065. if (element.tagName == "SCRIPT") {
  1066. scriptSrc = element.getAttribute("src");
  1067. element.removeAttribute("src");
  1068. } else {
  1069. scriptSrc = element.getAttribute("href");
  1070. element.removeAttribute("href");
  1071. }
  1072. element.removeAttribute("integrity");
  1073. element.textContent = "";
  1074. try {
  1075. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1076. } catch (error) {
  1077. // ignored
  1078. }
  1079. if (Util.testValidURL(resourceURL)) {
  1080. this.stats.add("processed", "scripts", 1);
  1081. const content = await util.getContent(resourceURL, {
  1082. asBinary: true,
  1083. maxResourceSize: this.options.maxResourceSize,
  1084. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1085. frameId: this.options.windowId
  1086. });
  1087. content.data = Util.getUpdatedResourceContent(resourceURL, content, this.options);
  1088. if (element.tagName == "SCRIPT") {
  1089. element.setAttribute("src", content.data);
  1090. } else {
  1091. const scriptElement = this.doc.createElement("script");
  1092. scriptElement.setAttribute("src", content.data);
  1093. element.parentElement.replaceChild(scriptElement, element);
  1094. }
  1095. }
  1096. }));
  1097. }
  1098. removeAlternativeImages() {
  1099. util.removeAlternativeImages(this.doc);
  1100. }
  1101. async removeAlternativeFonts() {
  1102. await util.removeAlternativeFonts(this.doc, this.stylesheets);
  1103. }
  1104. async processFrames() {
  1105. if (this.options.frames) {
  1106. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1107. await Promise.all(frameElements.map(async frameElement => {
  1108. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1109. if (frameWindowId) {
  1110. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1111. if (frameData) {
  1112. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1113. if (frameData.runner) {
  1114. this.stats.add("processed", "frames", 1);
  1115. await frameData.runner.run();
  1116. const pageData = await frameData.runner.getPageData();
  1117. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1118. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1119. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1120. sandbox += " allow-scripts allow-same-origin";
  1121. }
  1122. frameElement.setAttribute("sandbox", sandbox);
  1123. if (frameElement.tagName == "OBJECT") {
  1124. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1125. } else {
  1126. if (frameElement.tagName == "FRAME") {
  1127. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1128. } else {
  1129. frameElement.setAttribute("srcdoc", pageData.content);
  1130. frameElement.removeAttribute("src");
  1131. }
  1132. }
  1133. this.stats.addAll(pageData);
  1134. } else {
  1135. this.stats.add("discarded", "frames", 1);
  1136. }
  1137. }
  1138. }
  1139. }));
  1140. }
  1141. }
  1142. async processHtmlImports() {
  1143. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1144. await Promise.all(linkElements.map(async linkElement => {
  1145. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1146. if (attributeValue) {
  1147. const importData = this.options.imports[Number(attributeValue)];
  1148. if (importData.runner) {
  1149. this.stats.add("processed", "HTML imports", 1);
  1150. await importData.runner.run();
  1151. const pageData = await importData.runner.getPageData();
  1152. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1153. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1154. this.stats.addAll(pageData);
  1155. } else {
  1156. this.stats.add("discarded", "HTML imports", 1);
  1157. }
  1158. }
  1159. }));
  1160. }
  1161. replaceStylesheets() {
  1162. this.doc.querySelectorAll("style").forEach(styleElement => {
  1163. const stylesheetInfo = this.stylesheets.get(styleElement);
  1164. if (stylesheetInfo) {
  1165. this.stylesheets.delete(styleElement);
  1166. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1167. styleElement.textContent = stylesheetContent;
  1168. if (stylesheetInfo.mediaText) {
  1169. styleElement.media = stylesheetInfo.mediaText;
  1170. }
  1171. } else {
  1172. styleElement.remove();
  1173. }
  1174. });
  1175. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1176. const stylesheetInfo = this.stylesheets.get(linkElement);
  1177. if (stylesheetInfo) {
  1178. this.stylesheets.delete(linkElement);
  1179. const styleElement = this.doc.createElement("style");
  1180. if (stylesheetInfo.mediaText) {
  1181. styleElement.media = stylesheetInfo.mediaText;
  1182. }
  1183. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1184. styleElement.textContent = stylesheetContent;
  1185. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1186. } else {
  1187. linkElement.remove();
  1188. }
  1189. });
  1190. }
  1191. replaceStyleAttributes() {
  1192. this.doc.querySelectorAll("[style]").forEach(async element => {
  1193. const declarations = this.styles.get(element);
  1194. if (declarations) {
  1195. this.styles.delete(element);
  1196. let styleContent = cssTree.generate(declarations);
  1197. element.setAttribute("style", styleContent);
  1198. } else {
  1199. element.setAttribute("style", "");
  1200. }
  1201. });
  1202. }
  1203. insertVariables() {
  1204. if (this.cssVariables.size) {
  1205. const styleElement = this.doc.createElement("style");
  1206. const firstStyleElement = this.doc.head.querySelector("style");
  1207. if (firstStyleElement) {
  1208. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1209. } else {
  1210. this.doc.head.appendChild(styleElement);
  1211. }
  1212. let stylesheetContent = "";
  1213. this.cssVariables.forEach((content, indexResource) => {
  1214. this.cssVariables.delete(indexResource);
  1215. if (stylesheetContent) {
  1216. stylesheetContent += ";";
  1217. }
  1218. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1219. });
  1220. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1221. }
  1222. }
  1223. compressHTML() {
  1224. let size;
  1225. if (this.options.displayStats) {
  1226. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1227. }
  1228. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1229. if (this.options.displayStats) {
  1230. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1231. }
  1232. }
  1233. cleanupPage() {
  1234. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1235. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1236. if (metaCharset) {
  1237. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1238. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1239. this.doc.head.querySelector("meta[charset]").remove();
  1240. }
  1241. }
  1242. }
  1243. async insertMAFFMetaData() {
  1244. const maffMetaData = await this.maffMetaDataPromise;
  1245. if (maffMetaData && maffMetaData.content) {
  1246. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1247. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1248. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1249. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1250. if (originalURLElement) {
  1251. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1252. }
  1253. if (archiveTimeElement) {
  1254. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1255. if (value) {
  1256. const date = new Date(value);
  1257. if (!isNaN(date.getTime())) {
  1258. this.options.saveDate = new Date(value);
  1259. }
  1260. }
  1261. }
  1262. }
  1263. }
  1264. async setDocInfo() {
  1265. const titleElement = this.doc.querySelector("title");
  1266. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1267. const authorElement = this.doc.querySelector("meta[name=author]");
  1268. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1269. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1270. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1271. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1272. this.options.info = {
  1273. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1274. lang: this.doc.documentElement.lang,
  1275. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1276. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1277. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : ""
  1278. };
  1279. }
  1280. }
  1281. // ---------------
  1282. // ProcessorHelper
  1283. // ---------------
  1284. const DATA_URI_PREFIX = "data:";
  1285. const ABOUT_BLANK_URI = "about:blank";
  1286. const EMPTY_DATA_URI = "data:base64,";
  1287. const REGEXP_URL_HASH = /(#.+?)$/;
  1288. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1289. const VAR_MAX_SIZE = 1024 * 1024 * 2;
  1290. class ProcessorHelper {
  1291. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1292. const date = options.saveDate;
  1293. const url = util.parseURL(options.saveUrl);
  1294. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1295. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1296. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1297. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1298. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1299. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1300. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1301. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1302. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1303. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1304. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1305. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1306. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1307. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1308. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1309. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1310. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1311. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1312. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1313. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1314. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1315. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1316. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1317. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1318. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1319. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1320. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1321. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1322. template = await Util.evalTemplateVariable(template, "url-href", () => decodeURI(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1323. template = await Util.evalTemplateVariable(template, "url-referrer", () => decodeURI(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1324. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1325. template = await Util.evalTemplateVariable(template, "url-pathname", () => decodeURI(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1326. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1327. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1328. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1329. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1330. for (const [name, value] of params) {
  1331. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1332. }
  1333. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1334. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1335. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1336. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1337. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decodeURI(Util.getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1338. if (content) {
  1339. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1340. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1341. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1342. }
  1343. return template.trim();
  1344. }
  1345. static setBackgroundImage(element, url, style) {
  1346. element.style.setProperty("background-blend-mode", "normal", "important");
  1347. element.style.setProperty("background-clip", "content-box", "important");
  1348. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1349. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1350. element.style.setProperty("background-image", url, "important");
  1351. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1352. element.style.setProperty("background-origin", "content-box", "important");
  1353. element.style.setProperty("background-repeat", "no-repeat", "important");
  1354. }
  1355. static processShortcutIcons(doc) {
  1356. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1357. if (!shortcutIcon) {
  1358. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1359. if (shortcutIcon) {
  1360. shortcutIcon.rel = "icon";
  1361. }
  1362. }
  1363. if (shortcutIcon) {
  1364. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1365. if (linkElement != shortcutIcon) {
  1366. linkElement.remove();
  1367. }
  1368. });
  1369. }
  1370. }
  1371. static removeSingleLineCssComments(stylesheet) {
  1372. const removedRules = [];
  1373. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1374. const ruleData = cssRule.data;
  1375. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1376. removedRules.push(cssRule);
  1377. }
  1378. }
  1379. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1380. }
  1381. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1382. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1383. const imports = Util.getImportFunctions(stylesheetContent);
  1384. await Promise.all(imports.map(async cssImport => {
  1385. const match = Util.matchImport(cssImport);
  1386. if (match) {
  1387. const regExpCssImport = Util.getRegExp(cssImport);
  1388. let resourceURL = Util.normalizeURL(match.resourceURL);
  1389. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1390. try {
  1391. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1392. } catch (error) {
  1393. // ignored
  1394. }
  1395. if (Util.testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1396. const content = await util.getContent(resourceURL, {
  1397. maxResourceSize: options.maxResourceSize,
  1398. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1399. validateTextContentType: true,
  1400. frameId: options.windowId
  1401. });
  1402. resourceURL = content.resourceURL;
  1403. content.data = Util.getUpdatedResourceContent(resourceURL, content, options);
  1404. let importedStylesheetContent = Util.removeCssComments(content.data);
  1405. if (options.compressCSS) {
  1406. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1407. }
  1408. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1409. if (stylesheetContent.includes(cssImport)) {
  1410. const ancestorStyleSheets = new Set(importedStyleSheets);
  1411. ancestorStyleSheets.add(resourceURL);
  1412. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1413. workStylesheet.textContent = importedStylesheetContent;
  1414. if (workStylesheet.sheet.cssRules.length) {
  1415. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1416. } else {
  1417. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1418. }
  1419. }
  1420. } else {
  1421. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1422. }
  1423. } else {
  1424. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1425. }
  1426. }
  1427. }));
  1428. return stylesheetContent;
  1429. }
  1430. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1431. const urlFunctions = Util.getUrlFunctions(stylesheetContent, true);
  1432. urlFunctions.map(urlFunction => {
  1433. const originalResourceURL = Util.matchURL(urlFunction);
  1434. const resourceURL = Util.normalizeURL(originalResourceURL);
  1435. if (!Util.testIgnoredPath(resourceURL)) {
  1436. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1437. let resolvedURL;
  1438. if (!originalResourceURL.startsWith("#")) {
  1439. try {
  1440. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1441. } catch (error) {
  1442. // ignored
  1443. }
  1444. }
  1445. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1446. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1447. }
  1448. } else {
  1449. let newUrlFunction;
  1450. if (originalResourceURL) {
  1451. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1452. } else {
  1453. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1454. }
  1455. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1456. }
  1457. }
  1458. });
  1459. return stylesheetContent;
  1460. }
  1461. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1462. resourceURL = Util.normalizeURL(resourceURL);
  1463. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1464. const content = await util.getContent(resourceURL, {
  1465. maxResourceSize: options.maxResourceSize,
  1466. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1467. charset: options.charset,
  1468. frameId: options.windowId
  1469. });
  1470. resourceURL = content.resourceURL;
  1471. content.data = Util.getUpdatedResourceContent(content.resourceURL, content, options);
  1472. let stylesheetContent = Util.removeCssComments(content.data);
  1473. if (options.compressCSS) {
  1474. stylesheetContent = util.compressCSS(stylesheetContent);
  1475. }
  1476. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1477. return stylesheetContent;
  1478. }
  1479. }
  1480. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1481. const promises = [];
  1482. const removedRules = [];
  1483. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1484. const ruleData = cssRule.data;
  1485. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1486. removedRules.push(cssRule);
  1487. } else if (ruleData.block && ruleData.block.children) {
  1488. if (ruleData.type == "Rule") {
  1489. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1490. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1491. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1492. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1493. promises.push(processFontFaceRule(ruleData));
  1494. }
  1495. }
  1496. }
  1497. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1498. await Promise.all(promises);
  1499. async function processFontFaceRule(ruleData) {
  1500. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1501. if (declaration.type == "Declaration" && declaration.value.children) {
  1502. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value), true);
  1503. await Promise.all(urlFunctions.map(async urlFunction => {
  1504. const originalResourceURL = Util.matchURL(urlFunction);
  1505. const resourceURL = Util.normalizeURL(originalResourceURL);
  1506. if (!Util.testIgnoredPath(resourceURL)) {
  1507. if (Util.testValidURL(resourceURL)) {
  1508. let { content } = await batchRequest.addURL(resourceURL, true);
  1509. replaceURLs(declaration, originalResourceURL, content);
  1510. }
  1511. }
  1512. }));
  1513. }
  1514. }));
  1515. function replaceURLs(declaration, oldURL, newURL) {
  1516. declaration.value.children.forEach(token => {
  1517. if (token.type == "Url" && util.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1518. token.value.value = newURL;
  1519. }
  1520. });
  1521. }
  1522. }
  1523. }
  1524. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1525. await Promise.all(declarations.map(async declaration => {
  1526. if (declaration.type == "Declaration" && declaration.value.children) {
  1527. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1528. await Promise.all(urlFunctions.map(async urlFunction => {
  1529. const originalResourceURL = Util.matchURL(urlFunction);
  1530. const resourceURL = Util.normalizeURL(originalResourceURL);
  1531. if (!Util.testIgnoredPath(resourceURL)) {
  1532. if (Util.testValidURL(resourceURL)) {
  1533. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1534. let variableDefined;
  1535. const tokens = [];
  1536. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1537. if (!originalResourceURL.startsWith("#")) {
  1538. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < VAR_MAX_SIZE) {
  1539. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1540. tokens.push({ parent, token, value });
  1541. variableDefined = true;
  1542. } else {
  1543. token.data.value.value = content;
  1544. }
  1545. }
  1546. });
  1547. if (variableDefined) {
  1548. cssVariables.set(indexResource, content);
  1549. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1550. }
  1551. }
  1552. }
  1553. }));
  1554. }
  1555. }));
  1556. function findURLToken(url, children, callback, depth = 0) {
  1557. for (let token = children.head; token; token = token.next) {
  1558. if (token.data.children) {
  1559. findURLToken(url, token.data.children, callback, depth + 1);
  1560. }
  1561. if (token.data.type == "Url" && util.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1562. callback(token, children, depth == 0);
  1563. }
  1564. }
  1565. }
  1566. }
  1567. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1568. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1569. let resourceURL = resourceElement.getAttribute(attributeName);
  1570. if (resourceURL != null) {
  1571. resourceURL = Util.normalizeURL(resourceURL);
  1572. if (!Util.testIgnoredPath(resourceURL)) {
  1573. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1574. if (Util.testValidPath(resourceURL)) {
  1575. try {
  1576. resourceURL = util.resolveURL(resourceURL, baseURI);
  1577. } catch (error) {
  1578. // ignored
  1579. }
  1580. if (Util.testValidURL(resourceURL)) {
  1581. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1582. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1583. resourceElement.remove();
  1584. } else {
  1585. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1586. if (!forbiddenPrefixFound) {
  1587. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1588. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < VAR_MAX_SIZE) {
  1589. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1590. cssVariables.set(indexResource, content);
  1591. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1592. styles.set(resourceElement, declarationList);
  1593. } else {
  1594. resourceElement.setAttribute(attributeName, content);
  1595. }
  1596. } else {
  1597. resourceElement.setAttribute(attributeName, content);
  1598. }
  1599. }
  1600. }
  1601. }
  1602. }
  1603. }
  1604. }
  1605. }));
  1606. }
  1607. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1608. const attributeName = "xlink:href";
  1609. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1610. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1611. let resourceURL = Util.normalizeURL(originalResourceURL);
  1612. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1613. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1614. try {
  1615. resourceURL = util.resolveURL(resourceURL, baseURI);
  1616. } catch (error) {
  1617. // ignored
  1618. }
  1619. if (Util.testValidURL(resourceURL)) {
  1620. const { content } = await batchRequest.addURL(resourceURL);
  1621. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1622. if (hashMatch && hashMatch[0]) {
  1623. let symbolElement;
  1624. try {
  1625. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1626. } catch (error) {
  1627. // ignored
  1628. }
  1629. if (symbolElement) {
  1630. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1631. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1632. }
  1633. } else {
  1634. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1635. }
  1636. }
  1637. } else if (resourceURL == options.url) {
  1638. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1639. }
  1640. }));
  1641. }
  1642. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1643. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1644. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1645. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1646. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1647. if (!Util.testIgnoredPath(resourceURL)) {
  1648. if (Util.testValidPath(resourceURL)) {
  1649. try {
  1650. resourceURL = util.resolveURL(resourceURL, baseURI);
  1651. } catch (error) {
  1652. // ignored
  1653. }
  1654. if (Util.testValidURL(resourceURL)) {
  1655. const { content } = await batchRequest.addURL(resourceURL, true);
  1656. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1657. if (forbiddenPrefixFound) {
  1658. return "";
  1659. }
  1660. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1661. } else {
  1662. return "";
  1663. }
  1664. } else {
  1665. return "";
  1666. }
  1667. } else {
  1668. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1669. }
  1670. }));
  1671. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1672. }));
  1673. }
  1674. static replaceImageSource(imgElement, variableName, options) {
  1675. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1676. if (attributeValue) {
  1677. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1678. if (imageData && imageData.replaceable) {
  1679. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1680. const backgroundStyle = {};
  1681. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1682. if (backgroundSize) {
  1683. backgroundStyle["background-size"] = imageData.objectFit;
  1684. }
  1685. if (imageData.objectPosition) {
  1686. backgroundStyle["background-position"] = imageData.objectPosition;
  1687. }
  1688. if (imageData.backgroundColor) {
  1689. backgroundStyle["background-color"] = imageData.backgroundColor;
  1690. }
  1691. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1692. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1693. return true;
  1694. }
  1695. }
  1696. }
  1697. }
  1698. // ----
  1699. // Util
  1700. // ----
  1701. const BLOB_URI_PREFIX = "blob:";
  1702. const HTTP_URI_PREFIX = /^https?:\/\//;
  1703. const FILE_URI_PREFIX = /^file:\/\//;
  1704. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1705. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1706. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1707. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1708. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1709. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1710. 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;
  1711. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1712. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1713. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1714. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1715. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1716. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1717. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1718. class Util {
  1719. static getUpdatedResourceContent(resourceURL, content, options) {
  1720. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1721. options.updatedResources[resourceURL].retrieved = true;
  1722. return options.updatedResources[resourceURL].content;
  1723. } else {
  1724. return content.data || "";
  1725. }
  1726. }
  1727. static normalizeURL(url) {
  1728. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1729. return url;
  1730. } else {
  1731. return url.split("#")[0];
  1732. }
  1733. }
  1734. static getCSSValue(value) {
  1735. let result = "";
  1736. try {
  1737. result = cssTree.generate(value);
  1738. } catch (error) {
  1739. // ignored
  1740. }
  1741. return result;
  1742. }
  1743. static getOnEventAttributeNames(doc) {
  1744. const element = doc.createElement("div");
  1745. const attributeNames = [];
  1746. for (const propertyName in element) {
  1747. if (propertyName.startsWith("on")) {
  1748. attributeNames.push(propertyName);
  1749. }
  1750. }
  1751. attributeNames.push("onunload");
  1752. return attributeNames;
  1753. }
  1754. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1755. const replaceRegExp = new RegExp("{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}", "g");
  1756. if (template && template.match(replaceRegExp)) {
  1757. let value = await valueGetter();
  1758. if (!dontReplaceSlash) {
  1759. value = value.replace(/\/+/g, replacementCharacter);
  1760. }
  1761. return template.replace(replaceRegExp, value);
  1762. }
  1763. return template;
  1764. }
  1765. static getLastSegment(url, replacementCharacter) {
  1766. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1767. if (!lastSegment) {
  1768. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1769. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1770. }
  1771. if (!lastSegment) {
  1772. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1773. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1774. }
  1775. if (!lastSegment) {
  1776. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1777. }
  1778. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1779. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1780. lastSegment = lastSegmentMatch[1];
  1781. }
  1782. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1783. return lastSegment;
  1784. }
  1785. static getRegExp(string) {
  1786. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1787. }
  1788. static getUrlFunctions(stylesheetContent, unique) {
  1789. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1790. if (unique) {
  1791. return [...new Set(result)];
  1792. } else {
  1793. return result;
  1794. }
  1795. }
  1796. static getImportFunctions(stylesheetContent) {
  1797. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1798. }
  1799. static findShortcutIcon(shortcutIcons) {
  1800. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1801. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1802. return shortcutIcons[0];
  1803. }
  1804. static matchURL(stylesheetContent) {
  1805. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1806. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1807. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1808. return match && match[1];
  1809. }
  1810. static testIgnoredPath(resourceURL) {
  1811. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1812. }
  1813. static testValidPath(resourceURL) {
  1814. return resourceURL && !resourceURL.match(EMPTY_URL);
  1815. }
  1816. static testValidURL(resourceURL) {
  1817. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1818. }
  1819. static matchImport(stylesheetContent) {
  1820. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1821. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1822. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1823. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1824. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1825. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1826. if (match) {
  1827. const [, resourceURL, media] = match;
  1828. return { resourceURL, media };
  1829. }
  1830. }
  1831. static removeCssComments(stylesheetContent) {
  1832. let start, end;
  1833. do {
  1834. start = stylesheetContent.indexOf("/*");
  1835. end = stylesheetContent.indexOf("*/", start + 2);
  1836. if (start != -1 && end != -1) {
  1837. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1838. }
  1839. } while (start != -1 && end != -1);
  1840. return stylesheetContent;
  1841. }
  1842. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1843. if (mediaQuery) {
  1844. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1845. } else {
  1846. return stylesheetContent;
  1847. }
  1848. }
  1849. }
  1850. function log(...args) {
  1851. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1852. }
  1853. // -----
  1854. // Stats
  1855. // -----
  1856. const STATS_DEFAULT_VALUES = {
  1857. discarded: {
  1858. "HTML bytes": 0,
  1859. "hidden elements": 0,
  1860. "HTML imports": 0,
  1861. scripts: 0,
  1862. objects: 0,
  1863. "audio sources": 0,
  1864. "video sources": 0,
  1865. frames: 0,
  1866. "CSS rules": 0,
  1867. canvas: 0,
  1868. stylesheets: 0,
  1869. resources: 0,
  1870. medias: 0
  1871. },
  1872. processed: {
  1873. "HTML bytes": 0,
  1874. "hidden elements": 0,
  1875. "HTML imports": 0,
  1876. scripts: 0,
  1877. objects: 0,
  1878. "audio sources": 0,
  1879. "video sources": 0,
  1880. frames: 0,
  1881. "CSS rules": 0,
  1882. canvas: 0,
  1883. stylesheets: 0,
  1884. resources: 0,
  1885. medias: 0
  1886. }
  1887. };
  1888. class Stats {
  1889. constructor(options) {
  1890. this.options = options;
  1891. if (options.displayStats) {
  1892. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1893. }
  1894. }
  1895. set(type, subType, value) {
  1896. if (this.options.displayStats) {
  1897. this.data[type][subType] = value;
  1898. }
  1899. }
  1900. add(type, subType, value) {
  1901. if (this.options.displayStats) {
  1902. this.data[type][subType] += value;
  1903. }
  1904. }
  1905. addAll(pageData) {
  1906. if (this.options.displayStats) {
  1907. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1908. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1909. }
  1910. }
  1911. }
  1912. return { getClass };
  1913. })();