single-file-core.js 64 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484
  1. /*
  2. * Copyright 2018 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. /* global CSSRule */
  24. this.SingleFileCore = this.SingleFileCore || (() => {
  25. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  26. const SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = "data-single-file-selected-content-root";
  27. const DEBUG = false;
  28. let Download, DOM, URL, sessionId = 0;
  29. function getClass(...args) {
  30. [Download, DOM, URL] = args;
  31. return SingleFileClass;
  32. }
  33. class SingleFileClass {
  34. constructor(options) {
  35. this.options = options;
  36. if (options.sessionId === undefined) {
  37. options.sessionId = sessionId;
  38. sessionId++;
  39. }
  40. }
  41. async initialize() {
  42. this.processor = new PageProcessor(this.options);
  43. await this.processor.loadPage();
  44. await this.processor.initialize();
  45. }
  46. async run() {
  47. await this.processor.run();
  48. }
  49. async getPageData() {
  50. return this.processor.getPageData();
  51. }
  52. }
  53. SingleFileClass.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  54. SingleFileClass.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  55. // -------------
  56. // ProgressEvent
  57. // -------------
  58. const PAGE_LOADING = "page-loading";
  59. const PAGE_LOADED = "page-loaded";
  60. const RESOURCES_INITIALIZING = "resource-initializing";
  61. const RESOURCES_INITIALIZED = "resources-initialized";
  62. const RESOURCE_LOADED = "resource-loaded";
  63. const PAGE_ENDED = "page-ended";
  64. class ProgressEvent {
  65. constructor(type, details) {
  66. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  67. }
  68. }
  69. // -------------
  70. // PageProcessor
  71. // -------------
  72. const RESOLVE_URLS_STAGE = 0;
  73. const REPLACE_DATA_STAGE = 1;
  74. const REPLACE_DOCS_STAGE = 2;
  75. const POST_PROCESS_STAGE = 3;
  76. const STAGES = [{
  77. sequential: [
  78. { action: "preProcessPage" },
  79. { action: "replaceStyleContents" },
  80. { option: "selected", action: "removeUnselectedElements" },
  81. { option: "removeVideoSrc", action: "insertVideoPosters" },
  82. { option: "removeFrames", action: "removeFrames" },
  83. { option: "removeImports", action: "removeImports" },
  84. { option: "removeScripts", action: "removeScripts" },
  85. { action: "removeDiscardedResources" },
  86. { action: "resetCharsetMeta" },
  87. { action: "setInputValues" },
  88. { option: "insertFaviconLink", action: "insertFaviconLink" },
  89. { action: "resolveHrefs" },
  90. { action: "replaceCanvasElements" },
  91. { action: "insertFonts" },
  92. { option: "removeHiddenElements", action: "removeHiddenElements" },
  93. { action: "resolveStyleAttributeURLs" }
  94. ],
  95. parallel: [
  96. { action: "resolveStylesheetURLs" },
  97. { action: "resolveLinkedStylesheetURLs" },
  98. { option: "!removeFrames", action: "resolveFrameURLs" },
  99. { option: "!removeImports", action: "resolveHtmlImportURLs" }
  100. ]
  101. }, {
  102. sequential: [
  103. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  104. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  105. { option: "removeUnusedStyles", action: "removeUnusedFonts" }
  106. ],
  107. parallel: [
  108. { action: "processStylesheets" },
  109. { action: "processStyleAttributes" },
  110. { action: "processPageResources" },
  111. { option: "!removeScripts", action: "processScripts" }
  112. ]
  113. }, {
  114. sequential: [
  115. { option: "removeAlternativeImages", action: "removeAlternativeImages" },
  116. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  117. { option: "compressCSS", action: "compressCSS" }
  118. ],
  119. parallel: [
  120. { option: "!removeFrames", action: "processFrames" },
  121. { option: "!removeImports", action: "processHtmlImports" },
  122. ]
  123. }, {
  124. sequential: [
  125. { option: "compressHTML", action: "compressHTML" }
  126. ]
  127. }];
  128. class PageProcessor {
  129. constructor(options) {
  130. this.options = options;
  131. this.options.url = this.options.url || this.options.doc.location.href;
  132. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  133. this.batchRequest = new BatchRequest();
  134. this.processor = new DOMProcessor(options, this.batchRequest);
  135. if (this.options.doc) {
  136. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  137. this.options.canvasData = docData.canvasData;
  138. this.options.fontsData = docData.fontsData;
  139. this.options.stylesheetContents = docData.stylesheetContents;
  140. this.options.imageData = docData.imageData;
  141. this.options.postersData = docData.postersData;
  142. this.options.usedFonts = docData.usedFonts;
  143. }
  144. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  145. this.onprogress = options.onprogress || (() => { });
  146. }
  147. async loadPage() {
  148. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  149. await this.processor.loadPage(this.options.content);
  150. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  151. }
  152. async initialize() {
  153. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  154. await this.executeStage(RESOLVE_URLS_STAGE);
  155. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  156. if (this.options.doc) {
  157. DOM.postProcessDoc(this.options.doc, this.options);
  158. this.options.doc = null;
  159. this.options.win = null;
  160. }
  161. }
  162. async run() {
  163. if (!this.options.windowId) {
  164. this.processor.initialize(this.batchRequest);
  165. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  166. }
  167. await this.batchRequest.run(details => {
  168. details.pageURL = this.options.url;
  169. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  170. }, this.options);
  171. await this.pendingPromises;
  172. await this.executeStage(REPLACE_DOCS_STAGE);
  173. await this.executeStage(POST_PROCESS_STAGE);
  174. await this.processor.end();
  175. }
  176. async getPageData() {
  177. if (!this.options.windowId) {
  178. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  179. }
  180. return this.processor.getPageData();
  181. }
  182. async executeStage(step) {
  183. if (DEBUG) {
  184. log("**** STARTED STAGE", step, "****");
  185. }
  186. STAGES[step].sequential.forEach(task => {
  187. let startTime;
  188. if (DEBUG) {
  189. startTime = Date.now();
  190. log(" -- STARTED task =", task.action);
  191. }
  192. this.executeTask(task);
  193. if (DEBUG) {
  194. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  195. }
  196. });
  197. if (STAGES[step].parallel) {
  198. return await Promise.all(STAGES[step].parallel.map(task => {
  199. let startTime;
  200. if (DEBUG) {
  201. startTime = Date.now();
  202. log(" // STARTED task =", task.action);
  203. }
  204. const promise = this.executeTask(task);
  205. if (DEBUG) {
  206. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  207. }
  208. return promise;
  209. }));
  210. }
  211. if (DEBUG) {
  212. log("**** ENDED STAGE", step, "****");
  213. }
  214. }
  215. executeTask(task) {
  216. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  217. return this.processor[task.action]();
  218. }
  219. }
  220. }
  221. // --------
  222. // BatchRequest
  223. // --------
  224. class BatchRequest {
  225. constructor() {
  226. this.requests = new Map();
  227. }
  228. async addURL(resourceURL, asDataURI = true) {
  229. return new Promise((resolve, reject) => {
  230. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  231. const resourceRequests = this.requests.get(requestKey);
  232. if (resourceRequests) {
  233. resourceRequests.push({ resolve, reject });
  234. } else {
  235. this.requests.set(requestKey, [{ resolve, reject }]);
  236. }
  237. });
  238. }
  239. getMaxResources() {
  240. return Array.from(this.requests.keys()).length;
  241. }
  242. async run(onloadListener, options) {
  243. const resourceURLs = Array.from(this.requests.keys());
  244. let indexResource = 0;
  245. return Promise.all(resourceURLs.map(async requestKey => {
  246. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  247. const resourceRequests = this.requests.get(requestKey);
  248. try {
  249. const resourceContent = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  250. indexResource = indexResource + 1;
  251. onloadListener({ index: indexResource, url: resourceURL });
  252. resourceRequests.forEach(resourceRequest => resourceRequest.resolve({ content: resourceContent, indexResource, duplicate: Boolean(resourceRequests.length > 1) }));
  253. } catch (error) {
  254. indexResource = indexResource + 1;
  255. onloadListener({ index: indexResource, url: resourceURL });
  256. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  257. }
  258. this.requests.delete(requestKey);
  259. }));
  260. }
  261. }
  262. // ------------
  263. // DOMProcessor
  264. // ------------
  265. const EMPTY_DATA_URI = "data:base64,";
  266. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  267. const SCRIPT_TAG_FOUND = /<script/gi;
  268. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  269. class DOMProcessor {
  270. constructor(options, batchRequest) {
  271. this.options = options;
  272. this.stats = new Stats(options);
  273. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  274. this.batchRequest = batchRequest;
  275. }
  276. initialize() {
  277. this.maxResources = this.batchRequest.getMaxResources();
  278. if (!this.options.removeFrames) {
  279. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  280. }
  281. this.stats.set("processed", "resources", this.maxResources);
  282. }
  283. async loadPage(pageContent) {
  284. if (!pageContent || this.options.saveRawPage) {
  285. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  286. }
  287. this.doc = DOM.createDoc(pageContent, this.baseURI);
  288. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  289. }
  290. async getPageData() {
  291. DOM.postProcessDoc(this.doc, this.options);
  292. const url = new URL(this.baseURI);
  293. if (this.options.insertSingleFileComment) {
  294. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  295. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  296. " \n url: " + this.options.url +
  297. " \n saved date: " + new Date() +
  298. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  299. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  300. }
  301. let size;
  302. if (this.options.displayStats) {
  303. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  304. }
  305. const content = DOM.serialize(this.doc, this.options.compressHTML);
  306. if (this.options.displayStats) {
  307. const contentSize = DOM.getContentSize(content);
  308. this.stats.set("processed", "HTML bytes", contentSize);
  309. this.stats.add("discarded", "HTML bytes", size - contentSize);
  310. }
  311. let filename = await DomProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content);
  312. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  313. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  314. if (!this.options.backgroundSave) {
  315. filename = filename.replace(/\//g, "_");
  316. }
  317. if (filename.length > 192) {
  318. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  319. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  320. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  321. }
  322. if (!filename) {
  323. filename = "Unnamed page";
  324. }
  325. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  326. return {
  327. stats: this.stats.data,
  328. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  329. filename,
  330. content
  331. };
  332. }
  333. removeUnselectedElements() {
  334. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  335. if (rootElement) {
  336. DomProcessorHelper.isolateElements(rootElement);
  337. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  338. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  339. }
  340. }
  341. setInputValues() {
  342. this.doc.querySelectorAll("input").forEach(input => {
  343. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  344. input.setAttribute("value", value || "");
  345. });
  346. this.doc.querySelectorAll("textarea").forEach(textarea => {
  347. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  348. textarea.textContent = value || "";
  349. });
  350. this.doc.querySelectorAll("select").forEach(select => {
  351. select.querySelectorAll("option").forEach(option => {
  352. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  353. if (selected) {
  354. option.setAttribute("selected", "");
  355. }
  356. });
  357. });
  358. }
  359. removeDiscardedResources() {
  360. const objectElements = this.doc.querySelectorAll("applet, meta[http-equiv=refresh], object[data]:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed[src]:not([src*=\".svg\"])");
  361. this.stats.set("discarded", "objects", objectElements.length);
  362. this.stats.set("processed", "objects", objectElements.length);
  363. objectElements.forEach(element => element.remove());
  364. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  365. replacedAttributeValue.forEach(element => {
  366. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  367. if (relValue.length) {
  368. element.setAttribute("rel", relValue);
  369. } else {
  370. element.remove();
  371. }
  372. });
  373. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  374. if (this.options.compressHTML) {
  375. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  376. }
  377. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  378. if (this.options.removeScripts) {
  379. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  380. this.doc.querySelectorAll("[href]").forEach(element => {
  381. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  382. element.setAttribute("href", "");
  383. }
  384. });
  385. this.doc.querySelectorAll("[src]").forEach(element => {
  386. if (element.src && element.src.match(/^\s*javascript:/)) {
  387. element.removeAttribute("src");
  388. }
  389. });
  390. }
  391. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  392. this.stats.set("processed", "audio sources", audioSourceElements.length);
  393. if (this.options.removeAudioSrc) {
  394. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  395. audioSourceElements.forEach(element => element.removeAttribute("src"));
  396. }
  397. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  398. this.stats.set("processed", "video sources", videoSourceElements.length);
  399. if (this.options.removeVideoSrc) {
  400. this.stats.set("discarded", "video sources", videoSourceElements.length);
  401. videoSourceElements.forEach(element => element.removeAttribute("src"));
  402. }
  403. }
  404. async end() {
  405. const metaCharset = this.doc.head.querySelector("meta[charset]");
  406. if (metaCharset) {
  407. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  408. }
  409. this.doc.querySelectorAll("style[data-single-file-sheet]").forEach(element => element.removeAttribute("data-single-file-sheet"));
  410. this.doc.querySelectorAll("base").forEach(element => element.remove());
  411. if (this.doc.head.querySelectorAll("*").length == 1 && metaCharset && this.doc.body.childNodes.length == 0) {
  412. this.doc.head.querySelector("meta[charset]").remove();
  413. }
  414. const titleElement = this.doc.querySelector("title");
  415. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  416. this.options.info = {};
  417. const descriptionElement = this.doc.querySelector("meta[name=description]");
  418. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  419. this.options.info.lang = this.doc.documentElement.lang;
  420. const authorElement = this.doc.querySelector("meta[name=author]");
  421. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  422. const creatorElement = this.doc.querySelector("meta[name=creator]");
  423. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  424. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  425. this.options.info.publisher = publisherElement ? publisherElement.content.trim() : "";
  426. this.options.infobarContent = await DomProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  427. }
  428. preProcessPage() {
  429. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  430. if (this.options.win) {
  431. this.doc.body.querySelectorAll(":not(svg) title, meta").forEach(element => element instanceof this.options.win.HTMLElement && this.doc.head.appendChild(element));
  432. }
  433. if (this.options.imageData) {
  434. const dataAttributeName = DOM.imagesAttributeName(this.options.sessionId);
  435. this.doc.querySelectorAll("img").forEach(imgElement => {
  436. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  437. if (imgData.currentSrc) {
  438. imgElement.setAttribute("src", imgData.currentSrc);
  439. }
  440. });
  441. if (this.options.lazyLoadImages) {
  442. this.doc.querySelectorAll("img[data-src]").forEach(imgElement => {
  443. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  444. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  445. imgData.src = imgElement.dataset.src;
  446. imgElement.setAttribute("src", imgElement.dataset.src);
  447. imgElement.removeAttribute("data-src");
  448. }
  449. });
  450. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  451. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  452. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  453. imgElement.removeAttribute("data-srcset");
  454. }
  455. });
  456. }
  457. }
  458. }
  459. removeScripts() {
  460. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  461. this.stats.set("discarded", "scripts", scriptElements.length);
  462. this.stats.set("processed", "scripts", scriptElements.length);
  463. scriptElements.forEach(element => element.remove());
  464. }
  465. removeFrames() {
  466. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  467. this.stats.set("discarded", "frames", frameElements.length);
  468. this.stats.set("processed", "frames", frameElements.length);
  469. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  470. }
  471. removeImports() {
  472. const importElements = this.doc.querySelectorAll("link[rel=import]");
  473. this.stats.set("discarded", "HTML imports", importElements.length);
  474. this.stats.set("processed", "HTML imports", importElements.length);
  475. importElements.forEach(element => element.remove());
  476. }
  477. resetCharsetMeta() {
  478. let charSet;
  479. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  480. const charSetDeclaration = element.content.split(";")[1];
  481. if (charSetDeclaration && !charSet) {
  482. charSet = charSetDeclaration.split("=")[1];
  483. if (charSet) {
  484. this.charSet = charSet.trim().toLowerCase();
  485. }
  486. }
  487. element.remove();
  488. });
  489. const metaElement = this.doc.createElement("meta");
  490. metaElement.setAttribute("charset", "utf-8");
  491. if (this.doc.head.firstChild) {
  492. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  493. } else {
  494. this.doc.head.appendChild(metaElement);
  495. }
  496. }
  497. insertFaviconLink() {
  498. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  499. if (!faviconElement) {
  500. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  501. }
  502. if (!faviconElement) {
  503. faviconElement = this.doc.createElement("link");
  504. faviconElement.setAttribute("type", "image/x-icon");
  505. faviconElement.setAttribute("rel", "shortcut icon");
  506. faviconElement.setAttribute("href", "/favicon.ico");
  507. }
  508. this.doc.head.appendChild(faviconElement);
  509. }
  510. resolveHrefs() {
  511. this.doc.querySelectorAll("[href]").forEach(element => {
  512. if (element.href) {
  513. const href = element.href.baseVal ? element.href.baseVal : element.href;
  514. const normalizedHref = DomUtil.normalizeURL(href);
  515. if (normalizedHref == href && href) {
  516. element.setAttribute("href", href);
  517. }
  518. }
  519. });
  520. }
  521. removeUnusedStyles() {
  522. if (!this.mediaAllInfo) {
  523. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  524. }
  525. const stats = DOM.minifyCSS(this.doc, this.mediaAllInfo);
  526. this.stats.set("processed", "CSS rules", stats.processed);
  527. this.stats.set("discarded", "CSS rules", stats.discarded);
  528. }
  529. removeUnusedFonts() {
  530. DOM.removeUnusedFonts(this.doc, this.options);
  531. }
  532. removeAlternativeFonts() {
  533. DOM.removeAlternativeFonts(this.doc);
  534. }
  535. removeAlternativeImages() {
  536. DOM.removeAlternativeImages(this.doc);
  537. }
  538. removeHiddenElements() {
  539. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  540. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  541. this.stats.set("processed", "hidden elements", hiddenElements.length);
  542. hiddenElements.forEach(element => element.remove());
  543. }
  544. compressHTML() {
  545. let size;
  546. if (this.options.displayStats) {
  547. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  548. }
  549. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  550. if (this.options.displayStats) {
  551. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  552. }
  553. }
  554. removeAlternativeMedias() {
  555. const stats = DOM.minifyMedias(this.doc);
  556. this.stats.set("processed", "medias", stats.processed);
  557. this.stats.set("discarded", "medias", stats.discarded);
  558. }
  559. compressCSS() {
  560. this.doc.querySelectorAll("style").forEach(styleElement => {
  561. if (styleElement) {
  562. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  563. }
  564. });
  565. this.doc.querySelectorAll("[style]").forEach(element => {
  566. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  567. });
  568. }
  569. replaceCanvasElements() {
  570. if (this.options.canvasData) {
  571. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  572. const canvasData = this.options.canvasData[indexCanvasElement];
  573. if (canvasData) {
  574. DomProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  575. this.stats.add("processed", "canvas", 1);
  576. }
  577. });
  578. }
  579. }
  580. insertFonts() {
  581. if (this.options.fontsData && this.options.fontsData.length) {
  582. let stylesheetContent = "";
  583. this.options.fontsData.forEach(fontStyles => {
  584. if (fontStyles["font-family"] && fontStyles.src) {
  585. stylesheetContent += "@font-face{";
  586. let stylesContent = "";
  587. Object.keys(fontStyles).forEach(fontStyle => {
  588. if (stylesContent) {
  589. stylesContent += ";";
  590. }
  591. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  592. });
  593. stylesheetContent += stylesContent + "}";
  594. }
  595. });
  596. if (stylesheetContent) {
  597. const styleElement = this.doc.createElement("style");
  598. styleElement.textContent = stylesheetContent;
  599. const existingStyleElement = this.doc.querySelector("style");
  600. if (existingStyleElement) {
  601. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  602. } else {
  603. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  604. }
  605. }
  606. }
  607. }
  608. replaceStyleContents() {
  609. if (this.options.stylesheetContents) {
  610. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  611. if (this.options.stylesheetContents[styleIndex]) {
  612. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  613. }
  614. });
  615. }
  616. }
  617. insertVideoPosters() {
  618. if (this.options.postersData) {
  619. this.doc.querySelectorAll("video[src], video > source[src]").forEach((element, videoIndex) => {
  620. let videoElement;
  621. if (element.tagName == "VIDEO") {
  622. videoElement = element;
  623. } else {
  624. videoElement = element.parentElement;
  625. }
  626. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  627. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  628. }
  629. });
  630. }
  631. }
  632. async processPageResources() {
  633. const resourcePromises = [
  634. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", "data:", this.baseURI, this.options, this.batchRequest, false, true),
  635. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.options, this.batchRequest),
  636. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("img[src], input[src][type=image]"), "src", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest, true),
  637. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.options, this.batchRequest),
  638. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  639. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  640. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  641. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  642. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest)
  643. ];
  644. if (!this.options.removeAudioSrc) {
  645. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", PREFIX_DATA_URI_AUDIO, this.baseURI, this.options, this.batchRequest));
  646. }
  647. if (!this.options.removeVideoSrc) {
  648. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("video[src], video > source[src]"), "src", PREFIX_DATA_URI_VIDEO, this.baseURI, this.options, this.batchRequest));
  649. }
  650. await Promise.all(resourcePromises);
  651. if (this.options.removeAlternativeImages) {
  652. let shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  653. if (!shortcutIcon) {
  654. shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  655. if (shortcutIcon) {
  656. shortcutIcon.rel = "icon";
  657. }
  658. }
  659. if (shortcutIcon) {
  660. this.doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  661. if (linkElement != shortcutIcon) {
  662. linkElement.remove();
  663. }
  664. });
  665. }
  666. }
  667. function findShortcutIcon(shortcutIcons) {
  668. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  669. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  670. return shortcutIcons[0];
  671. }
  672. }
  673. async resolveStylesheetURLs() {
  674. await Promise.all(Array.from(this.doc.querySelectorAll("style"))
  675. .map(async styleElement => {
  676. const options = { url: this.options.url, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled };
  677. const textContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, options);
  678. styleElement.textContent = textContent;
  679. }));
  680. }
  681. async processStylesheets() {
  682. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => {
  683. this.stats.add("processed", "stylesheets", 1);
  684. if (styleElement.sheet) {
  685. styleElement.textContent = await DomProcessorHelper.processStylesheet(this.doc, styleElement.textContent, styleElement.sheet.cssRules, this.baseURI, this.options, this.batchRequest);
  686. } else {
  687. styleElement.remove();
  688. }
  689. }));
  690. }
  691. async processScripts() {
  692. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  693. if (scriptElement.src) {
  694. this.stats.add("processed", "scripts", 1);
  695. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  696. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  697. }
  698. scriptElement.removeAttribute("src");
  699. }));
  700. }
  701. async resolveFrameURLs() {
  702. if (this.options.framesData) {
  703. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  704. await Promise.all(frameElements.map(async frameElement => {
  705. DomProcessorHelper.setFrameEmptySrc(frameElement);
  706. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  707. if (frameWindowId) {
  708. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  709. if (frameData) {
  710. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  711. }
  712. }
  713. }));
  714. }
  715. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  716. options.insertSingleFileComment = false;
  717. options.insertFaviconLink = false;
  718. options.doc = null;
  719. options.win = null;
  720. options.url = frameData.baseURI;
  721. options.windowId = frameWindowId;
  722. if (frameData.content) {
  723. options.content = frameData.content;
  724. options.canvasData = frameData.canvasData;
  725. options.stylesheetContents = frameData.stylesheetContents;
  726. options.fontsData = frameData.fontsData;
  727. options.imageData = frameData.imageData;
  728. options.usedFonts = frameData.usedFonts;
  729. frameData.processor = new PageProcessor(options);
  730. frameData.frameElement = frameElement;
  731. await frameData.processor.loadPage();
  732. await frameData.processor.initialize();
  733. frameData.maxResources = batchRequest.getMaxResources();
  734. }
  735. }
  736. }
  737. async processFrames() {
  738. if (this.options.framesData) {
  739. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  740. await Promise.all(frameElements.map(async frameElement => {
  741. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  742. if (frameWindowId) {
  743. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  744. if (frameData) {
  745. if (frameData.processor) {
  746. this.stats.add("processed", "frames", 1);
  747. await frameData.processor.run();
  748. const pageData = await frameData.processor.getPageData();
  749. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  750. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  751. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  752. } else {
  753. frameElement.setAttribute("sandbox", "");
  754. }
  755. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  756. this.stats.addAll(pageData);
  757. } else {
  758. this.stats.add("discarded", "frames", 1);
  759. }
  760. }
  761. }
  762. }));
  763. }
  764. }
  765. async resolveHtmlImportURLs() {
  766. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  767. if (!this.relImportProcessors) {
  768. this.relImportProcessors = new Map();
  769. }
  770. await Promise.all(linkElements.map(async linkElement => {
  771. const resourceURL = linkElement.href;
  772. linkElement.removeAttribute("href");
  773. const options = Object.create(this.options);
  774. options.insertSingleFileComment = false;
  775. options.insertFaviconLink = false;
  776. options.doc = null;
  777. options.win = null;
  778. options.url = resourceURL;
  779. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, this.baseURI, this.options.url)) {
  780. const processor = new PageProcessor(options);
  781. this.relImportProcessors.set(linkElement, processor);
  782. await processor.loadPage();
  783. return processor.initialize();
  784. }
  785. }));
  786. }
  787. async processHtmlImports() {
  788. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  789. await Promise.all(linkElements.map(async linkElement => {
  790. const processor = this.relImportProcessors.get(linkElement);
  791. if (processor) {
  792. this.stats.add("processed", "HTML imports", 1);
  793. this.relImportProcessors.delete(linkElement);
  794. const pageData = await processor.getPageData();
  795. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  796. this.stats.addAll(pageData);
  797. } else {
  798. this.stats.add("discarded", "HTML imports", 1);
  799. }
  800. }));
  801. }
  802. resolveStyleAttributeURLs() {
  803. Array.from(this.doc.querySelectorAll("[style]"))
  804. .map(element => element.setAttribute("style", DomProcessorHelper.resolveStylesheetURLs(element.getAttribute("style"), this.baseURI, this.options)));
  805. }
  806. async processStyleAttributes() {
  807. await Promise.all(Array.from(this.doc.querySelectorAll("[style]"))
  808. .map(async element => {
  809. const rules = [{ type: CSSRule.STYLE_RULE, cssText: element.getAttribute("style") }];
  810. const textContent = await DomProcessorHelper.processStylesheet(this.doc, element.getAttribute("style"), rules, this.baseURI, this.options, this.batchRequest);
  811. element.setAttribute("style", textContent);
  812. }));
  813. }
  814. async resolveLinkedStylesheetURLs() {
  815. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  816. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charSet: this.charSet };
  817. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, options);
  818. if (stylesheetContent && (!linkElement.rel.includes("alternate") || !linkElement.title)) {
  819. const styleElement = this.doc.createElement("style");
  820. if (linkElement.media) {
  821. styleElement.media = linkElement.media;
  822. }
  823. styleElement.textContent = stylesheetContent;
  824. linkElement.parentElement.replaceChild(styleElement, linkElement);
  825. } else {
  826. linkElement.remove();
  827. }
  828. }));
  829. }
  830. }
  831. // ---------
  832. // DomHelper
  833. // ---------
  834. const REGEXP_AMP = /&/g;
  835. const REGEXP_NBSP = /\u00a0/g;
  836. const REGEXP_START_TAG = /</g;
  837. const REGEXP_END_TAG = />/g;
  838. const REGEXP_URL_HASH = /(#.+?)$/;
  839. const PREFIX_DATA_URI_IMAGE = "data:image/";
  840. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  841. const PREFIX_DATA_URI_VIDEO = "data:video/";
  842. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  843. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  844. const PREFIX_DATA_URI_OCTET_STREAM = /^data:(application|binary)\/octet-stream/;
  845. const PREFIX_DATA_URI_VND = "data:application/vnd.";
  846. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  847. class DomProcessorHelper {
  848. static async evalTemplate(template, options, content, dontReplaceSlash) {
  849. const date = new Date();
  850. const url = new URL(options.url);
  851. template = await DomUtil.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash);
  852. template = await DomUtil.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash);
  853. template = await DomUtil.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash);
  854. template = await DomUtil.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash);
  855. template = await DomUtil.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash);
  856. template = await DomUtil.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash);
  857. template = await DomUtil.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash);
  858. template = await DomUtil.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash);
  859. template = await DomUtil.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash);
  860. template = await DomUtil.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash);
  861. template = await DomUtil.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash);
  862. template = await DomUtil.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash);
  863. template = await DomUtil.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  864. template = await DomUtil.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash);
  865. template = await DomUtil.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash);
  866. template = await DomUtil.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash);
  867. template = await DomUtil.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash);
  868. template = await DomUtil.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  869. template = await DomUtil.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash);
  870. template = await DomUtil.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash);
  871. template = await DomUtil.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash);
  872. template = await DomUtil.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash);
  873. template = await DomUtil.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash);
  874. template = await DomUtil.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash);
  875. template = await DomUtil.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash);
  876. template = await DomUtil.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1), dontReplaceSlash);
  877. template = await DomUtil.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, ""), dontReplaceSlash);
  878. template = await DomUtil.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, ""), dontReplaceSlash);
  879. template = await DomUtil.evalTemplateVariable(template, "url-href", () => url.href, dontReplaceSlash);
  880. template = await DomUtil.evalTemplateVariable(template, "url-password", () => url.password, dontReplaceSlash);
  881. template = await DomUtil.evalTemplateVariable(template, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), dontReplaceSlash === undefined ? true : dontReplaceSlash);
  882. template = await DomUtil.evalTemplateVariable(template, "url-port", () => url.port, dontReplaceSlash);
  883. template = await DomUtil.evalTemplateVariable(template, "url-protocol", () => url.protocol, dontReplaceSlash);
  884. template = await DomUtil.evalTemplateVariable(template, "url-search", () => url.search.substring(1), dontReplaceSlash);
  885. template = await DomUtil.evalTemplateVariable(template, "url-username", () => url.username, dontReplaceSlash);
  886. template = await DomUtil.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash);
  887. template = await DomUtil.evalTemplateVariable(template, "url-last-segment", () => DomUtil.getLastSegment(url), dontReplaceSlash);
  888. if (content) {
  889. template = await DomUtil.evalTemplateVariable(template, "digest-sha-256", async () => DOM.digest("SHA-256", content), dontReplaceSlash);
  890. template = await DomUtil.evalTemplateVariable(template, "digest-sha-384", async () => DOM.digest("SHA-384", content), dontReplaceSlash);
  891. template = await DomUtil.evalTemplateVariable(template, "digest-sha-512", async () => DOM.digest("SHA-512", content), dontReplaceSlash);
  892. }
  893. return template;
  894. }
  895. static setBackgroundImage(element, url, style) {
  896. element.style.setProperty("background-blend-mode", "normal", "important");
  897. element.style.setProperty("background-clip", "content-box", "important");
  898. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  899. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  900. element.style.setProperty("background-image", url, "important");
  901. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  902. element.style.setProperty("background-origin", "content-box", "important");
  903. element.style.setProperty("background-repeat", "no-repeat", "important");
  904. }
  905. static setFrameEmptySrc(frameElement) {
  906. if (frameElement.tagName == "OBJECT") {
  907. frameElement.setAttribute("data", "data:text/html,");
  908. } else {
  909. frameElement.setAttribute("srcdoc", "");
  910. frameElement.removeAttribute("src");
  911. }
  912. }
  913. static setFrameContent(frameElement, content) {
  914. if (frameElement.tagName == "OBJECT") {
  915. frameElement.setAttribute("data", "data:text/html," + content);
  916. } else {
  917. frameElement.setAttribute("srcdoc", content);
  918. frameElement.removeAttribute("src");
  919. }
  920. }
  921. static isolateElements(rootElement) {
  922. rootElement.querySelectorAll("*:not(style)").forEach(element => {
  923. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  924. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  925. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  926. element.remove();
  927. }
  928. });
  929. isolateParentElements(rootElement.parentElement, rootElement);
  930. function isolateParentElements(parentElement, element) {
  931. let elementFound = false;
  932. Array.from(parentElement.childNodes).forEach(node => {
  933. elementFound = elementFound || (node == element);
  934. const tagName = node.tagName && node.tagName.toLowerCase();
  935. if (node != element && tagName != "svg" && tagName != "style" && tagName != "link") {
  936. if (elementFound) {
  937. node.remove();
  938. } else {
  939. node.hidden = true;
  940. if (node.childNodes && node.childNodes.length) {
  941. Array.from(node.childNodes).forEach(node => node.remove());
  942. }
  943. }
  944. }
  945. });
  946. element = element.parentElement;
  947. if (element && element.parentElement && element.parentElement.tagName != "HTML") {
  948. isolateParentElements(element.parentElement, element);
  949. }
  950. }
  951. }
  952. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  953. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, options);
  954. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  955. const imports = DomUtil.getImportFunctions(stylesheetContent);
  956. await Promise.all(imports.map(async cssImport => {
  957. const match = DomUtil.matchImport(cssImport);
  958. if (match) {
  959. let resourceURL = DomUtil.normalizeURL(match.resourceURL);
  960. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  961. resourceURL = new URL(match.resourceURL, baseURI).href;
  962. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  963. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  964. let importedStylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  965. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  966. if (stylesheetContent.includes(cssImport)) {
  967. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options);
  968. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  969. }
  970. }
  971. }
  972. }
  973. }));
  974. return stylesheetContent;
  975. }
  976. static resolveStylesheetURLs(stylesheetContent, baseURI, options) {
  977. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  978. urlFunctions.map(urlFunction => {
  979. const originalResourceURL = DomUtil.matchURL(urlFunction);
  980. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  981. if (!DomUtil.testIgnoredPath(resourceURL)) {
  982. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  983. const resolvedURL = new URL(resourceURL, baseURI).href;
  984. if (DomUtil.testValidURL(resolvedURL, baseURI, options.url) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  985. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  986. }
  987. }
  988. } else {
  989. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  990. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  991. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  992. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  993. }
  994. }
  995. }
  996. });
  997. return stylesheetContent;
  998. }
  999. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options) {
  1000. resourceURL = DomUtil.normalizeURL(resourceURL);
  1001. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1002. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charSet: options.charSet };
  1003. let stylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  1004. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  1005. return stylesheetContent;
  1006. }
  1007. }
  1008. static async processStylesheet(doc, stylesheetContent, cssRules, baseURI, options, batchRequest) {
  1009. const urlFunctions = getURLFunctions(cssRules);
  1010. await Promise.all([
  1011. Promise.all(urlFunctions.style.map(urlFunction => processURLFunction(urlFunction))),
  1012. Promise.all(urlFunctions.font.map(urlFunction => processURLFunction(urlFunction, true)))]);
  1013. return stylesheetContent;
  1014. async function processURLFunction(urlFunction, isFont) {
  1015. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1016. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1017. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1018. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1019. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1020. const validResource = !isFont || content == EMPTY_DATA_URI || content.startsWith(PREFIX_DATA_URI_VND) || content.startsWith(PREFIX_DATA_URI_IMAGE_SVG) || await DOM.validFont(urlFunction);
  1021. if (!validResource) {
  1022. content = EMPTY_DATA_URI;
  1023. }
  1024. let regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1025. if (!stylesheetContent.match(regExpUrlFunction)) {
  1026. urlFunction = "url(" + originalResourceURL + ")";
  1027. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1028. }
  1029. if (!stylesheetContent.match(regExpUrlFunction)) {
  1030. urlFunction = "url('" + originalResourceURL + "')";
  1031. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1032. }
  1033. if (stylesheetContent.match(regExpUrlFunction)) {
  1034. if (duplicate && options.groupDuplicateImages && !isFont) {
  1035. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")");
  1036. DomUtil.insertVariable(doc, indexResource, content, options);
  1037. } else {
  1038. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, content));
  1039. }
  1040. }
  1041. }
  1042. }
  1043. }
  1044. function getURLFunctions(cssRules, urlFunctions = { style: [], font: [] }) {
  1045. Array.from(cssRules).forEach(cssRule => {
  1046. if (cssRule.type == CSSRule.MEDIA_RULE) {
  1047. getURLFunctions(cssRule.cssRules, urlFunctions);
  1048. } else if (cssRule.type == CSSRule.FONT_FACE_RULE) {
  1049. getURLFunctionsRule(cssRule.cssText, urlFunctions.font);
  1050. } else {
  1051. getURLFunctionsRule(cssRule.cssText, urlFunctions.style);
  1052. }
  1053. });
  1054. return { style: Array.from(urlFunctions.style), font: Array.from(urlFunctions.font) };
  1055. }
  1056. function getURLFunctionsRule(cssText, urlFunctions) {
  1057. const urlFunctionsRule = DomUtil.getUrlFunctions(cssText);
  1058. urlFunctionsRule.forEach(urlFunction => {
  1059. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1060. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1061. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1062. if (DomUtil.testValidURL(resourceURL, baseURI, options.url) && cssText.includes(urlFunction)) {
  1063. urlFunctions.push(urlFunction);
  1064. }
  1065. }
  1066. });
  1067. return urlFunctions;
  1068. }
  1069. }
  1070. static async processAttribute(doc, resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest, processDuplicates, removeElementIfMissing) {
  1071. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1072. let resourceURL = resourceElement.getAttribute(attributeName);
  1073. resourceURL = DomUtil.normalizeURL(resourceURL);
  1074. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1075. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1076. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1077. resourceURL = new URL(resourceURL, baseURI).href;
  1078. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1079. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1080. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1081. resourceElement.remove();
  1082. } else {
  1083. if (content.startsWith(prefixDataURI) || content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) || content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1084. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1085. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1086. if (DomUtil.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1087. DomUtil.insertVariable(doc, indexResource, content, options);
  1088. } else {
  1089. resourceElement.setAttribute(attributeName, content);
  1090. }
  1091. } else {
  1092. resourceElement.setAttribute(attributeName, content);
  1093. }
  1094. }
  1095. }
  1096. }
  1097. }
  1098. }
  1099. }));
  1100. }
  1101. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1102. const attributeName = "xlink:href";
  1103. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1104. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1105. let resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1106. if (DomUtil.testValidPath(resourceURL, baseURI, options.url) && !DomUtil.testIgnoredPath(resourceURL)) {
  1107. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1108. resourceURL = new URL(resourceURL, baseURI).href;
  1109. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1110. try {
  1111. const { content } = await batchRequest.addURL(resourceURL, false);
  1112. const DOMParser = DOM.getParser();
  1113. if (DOMParser) {
  1114. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  1115. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1116. if (hashMatch && hashMatch[0]) {
  1117. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1118. if (symbolElement) {
  1119. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1120. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1121. }
  1122. } else {
  1123. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1124. }
  1125. } else {
  1126. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1127. }
  1128. } catch (error) {
  1129. /* ignored */
  1130. }
  1131. }
  1132. } else if (resourceURL == options.url) {
  1133. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1134. }
  1135. }));
  1136. }
  1137. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest) {
  1138. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1139. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1140. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1141. let resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1142. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1143. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1144. resourceURL = new URL(resourceURL, baseURI).href;
  1145. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1146. const { content } = await batchRequest.addURL(resourceURL);
  1147. if (!content.startsWith(prefixDataURI) && !content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) && !content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1148. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1149. }
  1150. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1151. } else {
  1152. return "";
  1153. }
  1154. } else {
  1155. return "";
  1156. }
  1157. }
  1158. }));
  1159. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1160. }));
  1161. }
  1162. }
  1163. // -------
  1164. // DomUtil
  1165. // -------
  1166. const DATA_URI_PREFIX = "data:";
  1167. const BLOB_URI_PREFIX = "blob:";
  1168. const HTTP_URI_PREFIX = /^https?:\/\//;
  1169. const FILE_URI_PREFIX = /^file:\/\//;
  1170. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1171. const ABOUT_BLANK_URI = "about:blank";
  1172. const NOT_EMPTY_URL = /^(https?|file):\/\/.+/;
  1173. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1174. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1175. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1176. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1177. 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;
  1178. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1179. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1180. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1181. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1182. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1183. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1184. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1185. class DomUtil {
  1186. static normalizeURL(url) {
  1187. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1188. return url;
  1189. } else {
  1190. return url.split("#")[0];
  1191. }
  1192. }
  1193. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1194. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1195. if (template.match(replaceRegExp)) {
  1196. let value = await valueGetter();
  1197. if (!dontReplaceSlash) {
  1198. value = value.replace(/\/+/g, "_");
  1199. }
  1200. return template.replace(replaceRegExp, value);
  1201. }
  1202. return template;
  1203. }
  1204. static getLastSegment(url) {
  1205. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1206. if (!lastSegment) {
  1207. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1208. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1209. }
  1210. if (!lastSegment) {
  1211. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1212. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1213. }
  1214. if (!lastSegment) {
  1215. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1216. }
  1217. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1218. return lastSegment;
  1219. }
  1220. static getRegExp(string) {
  1221. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1222. }
  1223. static getUrlFunctions(stylesheetContent) {
  1224. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1225. }
  1226. static getImportFunctions(stylesheetContent) {
  1227. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1228. }
  1229. static matchURL(stylesheetContent) {
  1230. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1231. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1232. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1233. return match && match[1];
  1234. }
  1235. static testIgnoredPath(resourceURL) {
  1236. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1237. }
  1238. static testValidPath(resourceURL, baseURI, docURL) {
  1239. return resourceURL && resourceURL != baseURI && resourceURL != docURL && !resourceURL.match(EMPTY_URL);
  1240. }
  1241. static testValidURL(resourceURL, baseURI, docURL) {
  1242. return DomUtil.testValidPath(resourceURL, baseURI, docURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1243. }
  1244. static matchImport(stylesheetContent) {
  1245. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1246. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1247. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1248. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1249. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1250. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1251. if (match) {
  1252. const [, resourceURL, media] = match;
  1253. return { resourceURL, media };
  1254. }
  1255. }
  1256. static removeCssComments(stylesheetContent) {
  1257. let start, end;
  1258. do {
  1259. start = stylesheetContent.indexOf("/*");
  1260. end = stylesheetContent.indexOf("*/", start + 2);
  1261. if (start != -1 && end != -1) {
  1262. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1263. }
  1264. } while (start != -1 && end != -1);
  1265. return stylesheetContent;
  1266. }
  1267. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1268. if (mediaQuery) {
  1269. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1270. } else {
  1271. return stylesheetContent;
  1272. }
  1273. }
  1274. static insertVariable(doc, indexResource, content, options) {
  1275. const sheetAttributeName = DOM.sheetAttributeName(options.sessionId);
  1276. let styleElement = doc.querySelector("style[" + sheetAttributeName + "]"), insertedVariables;
  1277. if (!styleElement) {
  1278. styleElement = doc.createElement("style");
  1279. if (doc.head.firstChild) {
  1280. doc.head.insertBefore(styleElement, doc.head.firstChild);
  1281. } else {
  1282. doc.head.appendChild(styleElement);
  1283. }
  1284. styleElement.setAttribute(sheetAttributeName, "[]");
  1285. insertedVariables = [];
  1286. } else {
  1287. insertedVariables = JSON.parse(styleElement.getAttribute(sheetAttributeName));
  1288. }
  1289. if (!insertedVariables.includes(indexResource)) {
  1290. insertedVariables.push(indexResource);
  1291. styleElement.textContent = styleElement.textContent + `:root{${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")}`;
  1292. styleElement.setAttribute(sheetAttributeName, JSON.stringify(insertedVariables));
  1293. }
  1294. }
  1295. static replaceImageSource(imgElement, variableName, options) {
  1296. const dataAttributeName = DOM.imagesAttributeName(options.sessionId);
  1297. if (imgElement.getAttribute(dataAttributeName) != null) {
  1298. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1299. if (imgData.replaceable) {
  1300. imgElement.setAttribute("src", `data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="${imgData.size.pxWidth}" height="${imgData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1301. const backgroundStyle = {};
  1302. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1303. if (backgroundSize) {
  1304. backgroundStyle["background-size"] = imgData.objectFit;
  1305. }
  1306. if (imgData.objectPosition) {
  1307. backgroundStyle["background-position"] = imgData.objectPosition;
  1308. }
  1309. if (imgData.backgroundColor) {
  1310. backgroundStyle["background-color"] = imgData.backgroundColor;
  1311. }
  1312. DomProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1313. imgElement.removeAttribute(dataAttributeName);
  1314. return true;
  1315. }
  1316. }
  1317. }
  1318. }
  1319. function log(...args) {
  1320. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1321. }
  1322. // -----
  1323. // Stats
  1324. // -----
  1325. const STATS_DEFAULT_VALUES = {
  1326. discarded: {
  1327. "HTML bytes": 0,
  1328. "hidden elements": 0,
  1329. "HTML imports": 0,
  1330. scripts: 0,
  1331. objects: 0,
  1332. "audio sources": 0,
  1333. "video sources": 0,
  1334. frames: 0,
  1335. "CSS rules": 0,
  1336. canvas: 0,
  1337. stylesheets: 0,
  1338. resources: 0,
  1339. medias: 0
  1340. },
  1341. processed: {
  1342. "HTML bytes": 0,
  1343. "hidden elements": 0,
  1344. "HTML imports": 0,
  1345. scripts: 0,
  1346. objects: 0,
  1347. "audio sources": 0,
  1348. "video sources": 0,
  1349. frames: 0,
  1350. "CSS rules": 0,
  1351. canvas: 0,
  1352. stylesheets: 0,
  1353. resources: 0,
  1354. medias: 0
  1355. }
  1356. };
  1357. class Stats {
  1358. constructor(options) {
  1359. this.options = options;
  1360. if (options.displayStats) {
  1361. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1362. }
  1363. }
  1364. set(type, subType, value) {
  1365. if (this.options.displayStats) {
  1366. this.data[type][subType] = value;
  1367. }
  1368. }
  1369. add(type, subType, value) {
  1370. if (this.options.displayStats) {
  1371. this.data[type][subType] += value;
  1372. }
  1373. }
  1374. addAll(pageData) {
  1375. if (this.options.displayStats) {
  1376. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1377. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1378. }
  1379. }
  1380. }
  1381. return { getClass };
  1382. })();