single-file-core.js 64 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481
  1. /*
  2. * Copyright 2018 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * SingleFile is free software: you can redistribute it and/or modify
  8. * it under the terms of the GNU Lesser General Public License as published by
  9. * the Free Software Foundation, either version 3 of the License, or
  10. * (at your option) any later version.
  11. *
  12. * SingleFile 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
  15. * GNU Lesser General Public License for more details.
  16. *
  17. * You should have received a copy of the GNU Lesser General Public License
  18. * along with SingleFile. If not, see <http://www.gnu.org/licenses/>.
  19. */
  20. /* global CSSRule */
  21. this.SingleFileCore = this.SingleFileCore || (() => {
  22. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  23. const SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = "data-single-file-selected-content-root";
  24. const DEBUG = false;
  25. let Download, DOM, URL, sessionId = 0;
  26. function getClass(...args) {
  27. [Download, DOM, URL] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. if (options.sessionId === undefined) {
  34. options.sessionId = sessionId;
  35. sessionId++;
  36. }
  37. }
  38. async initialize() {
  39. this.processor = new PageProcessor(this.options);
  40. await this.processor.loadPage();
  41. await this.processor.initialize();
  42. }
  43. async run() {
  44. await this.processor.run();
  45. }
  46. async getPageData() {
  47. return this.processor.getPageData();
  48. }
  49. }
  50. SingleFileClass.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  51. SingleFileClass.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  52. // -------------
  53. // ProgressEvent
  54. // -------------
  55. const PAGE_LOADING = "page-loading";
  56. const PAGE_LOADED = "page-loaded";
  57. const RESOURCES_INITIALIZING = "resource-initializing";
  58. const RESOURCES_INITIALIZED = "resources-initialized";
  59. const RESOURCE_LOADED = "resource-loaded";
  60. const PAGE_ENDED = "page-ended";
  61. class ProgressEvent {
  62. constructor(type, details) {
  63. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  64. }
  65. }
  66. // -------------
  67. // PageProcessor
  68. // -------------
  69. const RESOLVE_URLS_STAGE = 0;
  70. const REPLACE_DATA_STAGE = 1;
  71. const REPLACE_DOCS_STAGE = 2;
  72. const POST_PROCESS_STAGE = 3;
  73. const STAGES = [{
  74. sequential: [
  75. { action: "preProcessPage" },
  76. { action: "replaceStyleContents" },
  77. { option: "selected", action: "removeUnselectedElements" },
  78. { option: "removeVideoSrc", action: "insertVideoPosters" },
  79. { option: "removeFrames", action: "removeFrames" },
  80. { option: "removeImports", action: "removeImports" },
  81. { option: "removeScripts", action: "removeScripts" },
  82. { action: "removeDiscardedResources" },
  83. { action: "resetCharsetMeta" },
  84. { action: "setInputValues" },
  85. { option: "insertFaviconLink", action: "insertFaviconLink" },
  86. { action: "resolveHrefs" },
  87. { action: "replaceCanvasElements" },
  88. { action: "insertFonts" },
  89. { option: "removeHiddenElements", action: "removeHiddenElements" },
  90. { action: "resolveStyleAttributeURLs" }
  91. ],
  92. parallel: [
  93. { action: "resolveStylesheetURLs" },
  94. { action: "resolveLinkedStylesheetURLs" },
  95. { option: "!removeFrames", action: "resolveFrameURLs" },
  96. { option: "!removeImports", action: "resolveHtmlImportURLs" }
  97. ]
  98. }, {
  99. sequential: [
  100. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  101. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  102. { option: "removeUnusedStyles", action: "removeUnusedFonts" }
  103. ],
  104. parallel: [
  105. { action: "processStylesheets" },
  106. { action: "processStyleAttributes" },
  107. { action: "processPageResources" },
  108. { option: "!removeScripts", action: "processScripts" }
  109. ]
  110. }, {
  111. sequential: [
  112. { option: "removeAlternativeImages", action: "removeAlternativeImages" },
  113. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  114. { option: "compressCSS", action: "compressCSS" }
  115. ],
  116. parallel: [
  117. { option: "!removeFrames", action: "processFrames" },
  118. { option: "!removeImports", action: "processHtmlImports" },
  119. ]
  120. }, {
  121. sequential: [
  122. { option: "compressHTML", action: "compressHTML" }
  123. ]
  124. }];
  125. class PageProcessor {
  126. constructor(options) {
  127. this.options = options;
  128. this.options.url = this.options.url || this.options.doc.location.href;
  129. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  130. this.batchRequest = new BatchRequest();
  131. this.processor = new DOMProcessor(options, this.batchRequest);
  132. if (this.options.doc) {
  133. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  134. this.options.canvasData = docData.canvasData;
  135. this.options.fontsData = docData.fontsData;
  136. this.options.stylesheetContents = docData.stylesheetContents;
  137. this.options.imageData = docData.imageData;
  138. this.options.postersData = docData.postersData;
  139. this.options.usedFonts = docData.usedFonts;
  140. }
  141. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  142. this.onprogress = options.onprogress || (() => { });
  143. }
  144. async loadPage() {
  145. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  146. await this.processor.loadPage(this.options.content);
  147. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  148. }
  149. async initialize() {
  150. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  151. await this.executeStage(RESOLVE_URLS_STAGE);
  152. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  153. if (this.options.doc) {
  154. DOM.postProcessDoc(this.options.doc, this.options);
  155. this.options.doc = null;
  156. this.options.win = null;
  157. }
  158. }
  159. async run() {
  160. if (!this.options.windowId) {
  161. this.processor.initialize(this.batchRequest);
  162. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  163. }
  164. await this.batchRequest.run(details => {
  165. details.pageURL = this.options.url;
  166. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  167. }, this.options);
  168. await this.pendingPromises;
  169. await this.executeStage(REPLACE_DOCS_STAGE);
  170. await this.executeStage(POST_PROCESS_STAGE);
  171. await this.processor.end();
  172. }
  173. async getPageData() {
  174. if (!this.options.windowId) {
  175. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  176. }
  177. return this.processor.getPageData();
  178. }
  179. async executeStage(step) {
  180. if (DEBUG) {
  181. log("**** STARTED STAGE", step, "****");
  182. }
  183. STAGES[step].sequential.forEach(task => {
  184. let startTime;
  185. if (DEBUG) {
  186. startTime = Date.now();
  187. log(" -- STARTED task =", task.action);
  188. }
  189. this.executeTask(task);
  190. if (DEBUG) {
  191. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  192. }
  193. });
  194. if (STAGES[step].parallel) {
  195. return await Promise.all(STAGES[step].parallel.map(task => {
  196. let startTime;
  197. if (DEBUG) {
  198. startTime = Date.now();
  199. log(" // STARTED task =", task.action);
  200. }
  201. const promise = this.executeTask(task);
  202. if (DEBUG) {
  203. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  204. }
  205. return promise;
  206. }));
  207. }
  208. if (DEBUG) {
  209. log("**** ENDED STAGE", step, "****");
  210. }
  211. }
  212. executeTask(task) {
  213. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  214. return this.processor[task.action]();
  215. }
  216. }
  217. }
  218. // --------
  219. // BatchRequest
  220. // --------
  221. class BatchRequest {
  222. constructor() {
  223. this.requests = new Map();
  224. }
  225. async addURL(resourceURL, asDataURI = true) {
  226. return new Promise((resolve, reject) => {
  227. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  228. const resourceRequests = this.requests.get(requestKey);
  229. if (resourceRequests) {
  230. resourceRequests.push({ resolve, reject });
  231. } else {
  232. this.requests.set(requestKey, [{ resolve, reject }]);
  233. }
  234. });
  235. }
  236. getMaxResources() {
  237. return Array.from(this.requests.keys()).length;
  238. }
  239. async run(onloadListener, options) {
  240. const resourceURLs = Array.from(this.requests.keys());
  241. let indexResource = 0;
  242. return Promise.all(resourceURLs.map(async requestKey => {
  243. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  244. const resourceRequests = this.requests.get(requestKey);
  245. try {
  246. const resourceContent = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  247. indexResource = indexResource + 1;
  248. onloadListener({ index: indexResource, url: resourceURL });
  249. resourceRequests.forEach(resourceRequest => resourceRequest.resolve({ content: resourceContent, indexResource, duplicate: Boolean(resourceRequests.length > 1) }));
  250. } catch (error) {
  251. indexResource = indexResource + 1;
  252. onloadListener({ index: indexResource, url: resourceURL });
  253. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  254. }
  255. this.requests.delete(requestKey);
  256. }));
  257. }
  258. }
  259. // ------------
  260. // DOMProcessor
  261. // ------------
  262. const EMPTY_DATA_URI = "data:base64,";
  263. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  264. const SCRIPT_TAG_FOUND = /<script/gi;
  265. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  266. class DOMProcessor {
  267. constructor(options, batchRequest) {
  268. this.options = options;
  269. this.stats = new Stats(options);
  270. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  271. this.batchRequest = batchRequest;
  272. }
  273. initialize() {
  274. this.maxResources = this.batchRequest.getMaxResources();
  275. if (!this.options.removeFrames) {
  276. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  277. }
  278. this.stats.set("processed", "resources", this.maxResources);
  279. }
  280. async loadPage(pageContent) {
  281. if (!pageContent || this.options.saveRawPage) {
  282. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  283. }
  284. this.doc = DOM.createDoc(pageContent, this.baseURI);
  285. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  286. }
  287. async getPageData() {
  288. DOM.postProcessDoc(this.doc, this.options);
  289. const url = new URL(this.baseURI);
  290. if (this.options.insertSingleFileComment) {
  291. let infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  292. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  293. " \n url: " + this.options.url +
  294. " \n saved date: " + new Date() +
  295. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  296. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  297. }
  298. let size;
  299. if (this.options.displayStats) {
  300. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  301. }
  302. const content = DOM.serialize(this.doc, this.options.compressHTML);
  303. if (this.options.displayStats) {
  304. const contentSize = DOM.getContentSize(content);
  305. this.stats.set("processed", "HTML bytes", contentSize);
  306. this.stats.add("discarded", "HTML bytes", size - contentSize);
  307. }
  308. let filename = await DomProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content);
  309. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  310. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  311. if (!this.options.backgroundSave) {
  312. filename = filename.replace(/\//g, "_");
  313. }
  314. if (filename.length > 192) {
  315. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  316. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  317. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  318. }
  319. if (!filename) {
  320. filename = "Unnamed page";
  321. }
  322. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  323. return {
  324. stats: this.stats.data,
  325. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  326. filename,
  327. content
  328. };
  329. }
  330. removeUnselectedElements() {
  331. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  332. if (rootElement) {
  333. DomProcessorHelper.isolateElements(rootElement);
  334. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  335. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  336. }
  337. }
  338. setInputValues() {
  339. this.doc.querySelectorAll("input").forEach(input => {
  340. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  341. input.setAttribute("value", value || "");
  342. });
  343. this.doc.querySelectorAll("textarea").forEach(textarea => {
  344. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  345. textarea.textContent = value || "";
  346. });
  347. this.doc.querySelectorAll("select").forEach(select => {
  348. select.querySelectorAll("option").forEach(option => {
  349. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  350. if (selected) {
  351. option.setAttribute("selected", "");
  352. }
  353. });
  354. });
  355. }
  356. removeDiscardedResources() {
  357. 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\"])");
  358. this.stats.set("discarded", "objects", objectElements.length);
  359. this.stats.set("processed", "objects", objectElements.length);
  360. objectElements.forEach(element => element.remove());
  361. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  362. replacedAttributeValue.forEach(element => {
  363. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  364. if (relValue.length) {
  365. element.setAttribute("rel", relValue);
  366. } else {
  367. element.remove();
  368. }
  369. });
  370. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  371. if (this.options.compressHTML) {
  372. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  373. }
  374. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  375. if (this.options.removeScripts) {
  376. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  377. this.doc.querySelectorAll("[href]").forEach(element => {
  378. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  379. element.setAttribute("href", "");
  380. }
  381. });
  382. this.doc.querySelectorAll("[src]").forEach(element => {
  383. if (element.src && element.src.match(/^\s*javascript:/)) {
  384. element.removeAttribute("src");
  385. }
  386. });
  387. }
  388. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  389. this.stats.set("processed", "audio sources", audioSourceElements.length);
  390. if (this.options.removeAudioSrc) {
  391. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  392. audioSourceElements.forEach(element => element.removeAttribute("src"));
  393. }
  394. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  395. this.stats.set("processed", "video sources", videoSourceElements.length);
  396. if (this.options.removeVideoSrc) {
  397. this.stats.set("discarded", "video sources", videoSourceElements.length);
  398. videoSourceElements.forEach(element => element.removeAttribute("src"));
  399. }
  400. }
  401. async end() {
  402. const metaCharset = this.doc.head.querySelector("meta[charset]");
  403. if (metaCharset) {
  404. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  405. }
  406. this.doc.querySelectorAll("style[data-single-file-sheet]").forEach(element => element.removeAttribute("data-single-file-sheet"));
  407. this.doc.querySelectorAll("base").forEach(element => element.remove());
  408. if (this.doc.head.querySelectorAll("*").length == 1 && metaCharset && this.doc.body.childNodes.length == 0) {
  409. this.doc.head.querySelector("meta[charset]").remove();
  410. }
  411. const titleElement = this.doc.querySelector("title");
  412. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  413. this.options.info = {};
  414. const descriptionElement = this.doc.querySelector("meta[name=description]");
  415. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  416. this.options.info.lang = this.doc.documentElement.lang;
  417. const authorElement = this.doc.querySelector("meta[name=author]");
  418. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  419. const creatorElement = this.doc.querySelector("meta[name=creator]");
  420. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  421. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  422. this.options.info.publisher = publisherElement ? publisherElement.content.trim() : "";
  423. this.options.infobarContent = await DomProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  424. }
  425. preProcessPage() {
  426. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  427. if (this.options.win) {
  428. this.doc.body.querySelectorAll(":not(svg) title, meta").forEach(element => element instanceof this.options.win.HTMLElement && this.doc.head.appendChild(element));
  429. }
  430. if (this.options.imageData) {
  431. const dataAttributeName = DOM.imagesAttributeName(this.options.sessionId);
  432. this.doc.querySelectorAll("img").forEach(imgElement => {
  433. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  434. if (imgData.currentSrc) {
  435. imgElement.setAttribute("src", imgData.currentSrc);
  436. }
  437. });
  438. if (this.options.lazyLoadImages) {
  439. this.doc.querySelectorAll("img[data-src]").forEach(imgElement => {
  440. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  441. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  442. imgData.src = imgElement.dataset.src;
  443. imgElement.setAttribute("src", imgElement.dataset.src);
  444. imgElement.removeAttribute("data-src");
  445. }
  446. });
  447. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  448. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  449. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  450. imgElement.removeAttribute("data-srcset");
  451. }
  452. });
  453. }
  454. }
  455. }
  456. removeScripts() {
  457. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  458. this.stats.set("discarded", "scripts", scriptElements.length);
  459. this.stats.set("processed", "scripts", scriptElements.length);
  460. scriptElements.forEach(element => element.remove());
  461. }
  462. removeFrames() {
  463. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  464. this.stats.set("discarded", "frames", frameElements.length);
  465. this.stats.set("processed", "frames", frameElements.length);
  466. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  467. }
  468. removeImports() {
  469. const importElements = this.doc.querySelectorAll("link[rel=import]");
  470. this.stats.set("discarded", "HTML imports", importElements.length);
  471. this.stats.set("processed", "HTML imports", importElements.length);
  472. importElements.forEach(element => element.remove());
  473. }
  474. resetCharsetMeta() {
  475. let charSet;
  476. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  477. const charSetDeclaration = element.content.split(";")[1];
  478. if (charSetDeclaration && !charSet) {
  479. charSet = charSetDeclaration.split("=")[1];
  480. if (charSet) {
  481. this.charSet = charSet.trim().toLowerCase();
  482. }
  483. }
  484. element.remove();
  485. });
  486. const metaElement = this.doc.createElement("meta");
  487. metaElement.setAttribute("charset", "utf-8");
  488. if (this.doc.head.firstChild) {
  489. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  490. } else {
  491. this.doc.head.appendChild(metaElement);
  492. }
  493. }
  494. insertFaviconLink() {
  495. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  496. if (!faviconElement) {
  497. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  498. }
  499. if (!faviconElement) {
  500. faviconElement = this.doc.createElement("link");
  501. faviconElement.setAttribute("type", "image/x-icon");
  502. faviconElement.setAttribute("rel", "shortcut icon");
  503. faviconElement.setAttribute("href", "/favicon.ico");
  504. }
  505. this.doc.head.appendChild(faviconElement);
  506. }
  507. resolveHrefs() {
  508. this.doc.querySelectorAll("[href]").forEach(element => {
  509. if (element.href) {
  510. const href = element.href.baseVal ? element.href.baseVal : element.href;
  511. const normalizedHref = DomUtil.normalizeURL(href);
  512. if (normalizedHref == href && href) {
  513. element.setAttribute("href", href);
  514. }
  515. }
  516. });
  517. }
  518. removeUnusedStyles() {
  519. if (!this.mediaAllInfo) {
  520. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  521. }
  522. const stats = DOM.minifyCSS(this.doc, this.mediaAllInfo);
  523. this.stats.set("processed", "CSS rules", stats.processed);
  524. this.stats.set("discarded", "CSS rules", stats.discarded);
  525. }
  526. removeUnusedFonts() {
  527. DOM.removeUnusedFonts(this.doc, this.options);
  528. }
  529. removeAlternativeFonts() {
  530. DOM.removeAlternativeFonts(this.doc);
  531. }
  532. removeAlternativeImages() {
  533. DOM.removeAlternativeImages(this.doc);
  534. }
  535. removeHiddenElements() {
  536. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  537. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  538. this.stats.set("processed", "hidden elements", hiddenElements.length);
  539. hiddenElements.forEach(element => element.remove());
  540. }
  541. compressHTML() {
  542. let size;
  543. if (this.options.displayStats) {
  544. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  545. }
  546. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  547. if (this.options.displayStats) {
  548. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  549. }
  550. }
  551. removeAlternativeMedias() {
  552. const stats = DOM.minifyMedias(this.doc);
  553. this.stats.set("processed", "medias", stats.processed);
  554. this.stats.set("discarded", "medias", stats.discarded);
  555. }
  556. compressCSS() {
  557. this.doc.querySelectorAll("style").forEach(styleElement => {
  558. if (styleElement) {
  559. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  560. }
  561. });
  562. this.doc.querySelectorAll("[style]").forEach(element => {
  563. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  564. });
  565. }
  566. replaceCanvasElements() {
  567. if (this.options.canvasData) {
  568. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  569. const canvasData = this.options.canvasData[indexCanvasElement];
  570. if (canvasData) {
  571. DomProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  572. this.stats.add("processed", "canvas", 1);
  573. }
  574. });
  575. }
  576. }
  577. insertFonts() {
  578. if (this.options.fontsData && this.options.fontsData.length) {
  579. let stylesheetContent = "";
  580. this.options.fontsData.forEach(fontStyles => {
  581. if (fontStyles["font-family"] && fontStyles.src) {
  582. stylesheetContent += "@font-face{";
  583. let stylesContent = "";
  584. Object.keys(fontStyles).forEach(fontStyle => {
  585. if (stylesContent) {
  586. stylesContent += ";";
  587. }
  588. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  589. });
  590. stylesheetContent += stylesContent + "}";
  591. }
  592. });
  593. if (stylesheetContent) {
  594. const styleElement = this.doc.createElement("style");
  595. styleElement.textContent = stylesheetContent;
  596. const existingStyleElement = this.doc.querySelector("style");
  597. if (existingStyleElement) {
  598. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  599. } else {
  600. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  601. }
  602. }
  603. }
  604. }
  605. replaceStyleContents() {
  606. if (this.options.stylesheetContents) {
  607. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  608. if (this.options.stylesheetContents[styleIndex]) {
  609. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  610. }
  611. });
  612. }
  613. }
  614. insertVideoPosters() {
  615. if (this.options.postersData) {
  616. this.doc.querySelectorAll("video[src], video > source[src]").forEach((element, videoIndex) => {
  617. let videoElement;
  618. if (element.tagName == "VIDEO") {
  619. videoElement = element;
  620. } else {
  621. videoElement = element.parentElement;
  622. }
  623. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  624. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  625. }
  626. });
  627. }
  628. }
  629. async processPageResources() {
  630. const resourcePromises = [
  631. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", "data:", this.baseURI, this.options, this.batchRequest, false, true),
  632. 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),
  633. 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),
  634. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.options, this.batchRequest),
  635. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  636. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  637. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  638. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  639. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest)
  640. ];
  641. if (!this.options.removeAudioSrc) {
  642. 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));
  643. }
  644. if (!this.options.removeVideoSrc) {
  645. 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));
  646. }
  647. await Promise.all(resourcePromises);
  648. if (this.options.removeAlternativeImages) {
  649. let shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  650. if (!shortcutIcon) {
  651. shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  652. if (shortcutIcon) {
  653. shortcutIcon.rel = "icon";
  654. }
  655. }
  656. if (shortcutIcon) {
  657. this.doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  658. if (linkElement != shortcutIcon) {
  659. linkElement.remove();
  660. }
  661. });
  662. }
  663. }
  664. function findShortcutIcon(shortcutIcons) {
  665. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  666. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  667. return shortcutIcons[0];
  668. }
  669. }
  670. async resolveStylesheetURLs() {
  671. await Promise.all(Array.from(this.doc.querySelectorAll("style"))
  672. .map(async styleElement => {
  673. const options = { url: this.options.url, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled };
  674. const textContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, options);
  675. styleElement.textContent = textContent;
  676. }));
  677. }
  678. async processStylesheets() {
  679. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => {
  680. this.stats.add("processed", "stylesheets", 1);
  681. if (styleElement.sheet) {
  682. styleElement.textContent = await DomProcessorHelper.processStylesheet(this.doc, styleElement.textContent, styleElement.sheet.cssRules, this.baseURI, this.options, this.batchRequest);
  683. } else {
  684. styleElement.remove();
  685. }
  686. }));
  687. }
  688. async processScripts() {
  689. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  690. if (scriptElement.src) {
  691. this.stats.add("processed", "scripts", 1);
  692. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  693. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  694. }
  695. scriptElement.removeAttribute("src");
  696. }));
  697. }
  698. async resolveFrameURLs() {
  699. if (this.options.framesData) {
  700. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  701. await Promise.all(frameElements.map(async frameElement => {
  702. DomProcessorHelper.setFrameEmptySrc(frameElement);
  703. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  704. if (frameWindowId) {
  705. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  706. if (frameData) {
  707. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  708. }
  709. }
  710. }));
  711. }
  712. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  713. options.insertSingleFileComment = false;
  714. options.insertFaviconLink = false;
  715. options.doc = null;
  716. options.win = null;
  717. options.url = frameData.baseURI;
  718. options.windowId = frameWindowId;
  719. if (frameData.content) {
  720. options.content = frameData.content;
  721. options.canvasData = frameData.canvasData;
  722. options.stylesheetContents = frameData.stylesheetContents;
  723. options.fontsData = frameData.fontsData;
  724. options.imageData = frameData.imageData;
  725. options.usedFonts = frameData.usedFonts;
  726. frameData.processor = new PageProcessor(options);
  727. frameData.frameElement = frameElement;
  728. await frameData.processor.loadPage();
  729. await frameData.processor.initialize();
  730. frameData.maxResources = batchRequest.getMaxResources();
  731. }
  732. }
  733. }
  734. async processFrames() {
  735. if (this.options.framesData) {
  736. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  737. await Promise.all(frameElements.map(async frameElement => {
  738. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  739. if (frameWindowId) {
  740. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  741. if (frameData) {
  742. if (frameData.processor) {
  743. this.stats.add("processed", "frames", 1);
  744. await frameData.processor.run();
  745. const pageData = await frameData.processor.getPageData();
  746. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  747. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  748. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  749. } else {
  750. frameElement.setAttribute("sandbox", "");
  751. }
  752. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  753. this.stats.addAll(pageData);
  754. } else {
  755. this.stats.add("discarded", "frames", 1);
  756. }
  757. }
  758. }
  759. }));
  760. }
  761. }
  762. async resolveHtmlImportURLs() {
  763. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  764. if (!this.relImportProcessors) {
  765. this.relImportProcessors = new Map();
  766. }
  767. await Promise.all(linkElements.map(async linkElement => {
  768. const resourceURL = linkElement.href;
  769. linkElement.removeAttribute("href");
  770. const options = Object.create(this.options);
  771. options.insertSingleFileComment = false;
  772. options.insertFaviconLink = false;
  773. options.doc = null;
  774. options.win = null;
  775. options.url = resourceURL;
  776. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, this.baseURI, this.options.url)) {
  777. const processor = new PageProcessor(options);
  778. this.relImportProcessors.set(linkElement, processor);
  779. await processor.loadPage();
  780. return processor.initialize();
  781. }
  782. }));
  783. }
  784. async processHtmlImports() {
  785. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  786. await Promise.all(linkElements.map(async linkElement => {
  787. const processor = this.relImportProcessors.get(linkElement);
  788. if (processor) {
  789. this.stats.add("processed", "HTML imports", 1);
  790. this.relImportProcessors.delete(linkElement);
  791. const pageData = await processor.getPageData();
  792. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  793. this.stats.addAll(pageData);
  794. } else {
  795. this.stats.add("discarded", "HTML imports", 1);
  796. }
  797. }));
  798. }
  799. resolveStyleAttributeURLs() {
  800. Array.from(this.doc.querySelectorAll("[style]"))
  801. .map(element => element.setAttribute("style", DomProcessorHelper.resolveStylesheetURLs(element.getAttribute("style"), this.baseURI, this.options)));
  802. }
  803. async processStyleAttributes() {
  804. await Promise.all(Array.from(this.doc.querySelectorAll("[style]"))
  805. .map(async element => {
  806. const rules = [{ type: CSSRule.STYLE_RULE, cssText: element.getAttribute("style") }];
  807. const textContent = await DomProcessorHelper.processStylesheet(this.doc, element.getAttribute("style"), rules, this.baseURI, this.options, this.batchRequest);
  808. element.setAttribute("style", textContent);
  809. }));
  810. }
  811. async resolveLinkedStylesheetURLs() {
  812. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  813. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charSet: this.charSet };
  814. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, options);
  815. if (stylesheetContent && (!linkElement.rel.includes("alternate") || !linkElement.title)) {
  816. const styleElement = this.doc.createElement("style");
  817. if (linkElement.media) {
  818. styleElement.media = linkElement.media;
  819. }
  820. styleElement.textContent = stylesheetContent;
  821. linkElement.parentElement.replaceChild(styleElement, linkElement);
  822. } else {
  823. linkElement.remove();
  824. }
  825. }));
  826. }
  827. }
  828. // ---------
  829. // DomHelper
  830. // ---------
  831. const REGEXP_AMP = /&/g;
  832. const REGEXP_NBSP = /\u00a0/g;
  833. const REGEXP_START_TAG = /</g;
  834. const REGEXP_END_TAG = />/g;
  835. const REGEXP_URL_HASH = /(#.+?)$/;
  836. const PREFIX_DATA_URI_IMAGE = "data:image/";
  837. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  838. const PREFIX_DATA_URI_VIDEO = "data:video/";
  839. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  840. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  841. const PREFIX_DATA_URI_OCTET_STREAM = /^data:(application|binary)\/octet-stream/;
  842. const PREFIX_DATA_URI_VND = "data:application/vnd.";
  843. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  844. class DomProcessorHelper {
  845. static async evalTemplate(template, options, content, dontReplaceSlash) {
  846. const date = new Date();
  847. const url = new URL(options.url);
  848. template = await DomUtil.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash);
  849. template = await DomUtil.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash);
  850. template = await DomUtil.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash);
  851. template = await DomUtil.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash);
  852. template = await DomUtil.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash);
  853. template = await DomUtil.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash);
  854. template = await DomUtil.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash);
  855. template = await DomUtil.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash);
  856. template = await DomUtil.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash);
  857. template = await DomUtil.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash);
  858. template = await DomUtil.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash);
  859. template = await DomUtil.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash);
  860. template = await DomUtil.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  861. template = await DomUtil.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash);
  862. template = await DomUtil.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash);
  863. template = await DomUtil.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash);
  864. template = await DomUtil.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash);
  865. template = await DomUtil.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  866. template = await DomUtil.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash);
  867. template = await DomUtil.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash);
  868. template = await DomUtil.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash);
  869. template = await DomUtil.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash);
  870. template = await DomUtil.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash);
  871. template = await DomUtil.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash);
  872. template = await DomUtil.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash);
  873. template = await DomUtil.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1), dontReplaceSlash);
  874. template = await DomUtil.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, ""), dontReplaceSlash);
  875. template = await DomUtil.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, ""), dontReplaceSlash);
  876. template = await DomUtil.evalTemplateVariable(template, "url-href", () => url.href, dontReplaceSlash);
  877. template = await DomUtil.evalTemplateVariable(template, "url-password", () => url.password, dontReplaceSlash);
  878. template = await DomUtil.evalTemplateVariable(template, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), dontReplaceSlash === undefined ? true : dontReplaceSlash);
  879. template = await DomUtil.evalTemplateVariable(template, "url-port", () => url.port, dontReplaceSlash);
  880. template = await DomUtil.evalTemplateVariable(template, "url-protocol", () => url.protocol, dontReplaceSlash);
  881. template = await DomUtil.evalTemplateVariable(template, "url-search", () => url.search.substring(1), dontReplaceSlash);
  882. template = await DomUtil.evalTemplateVariable(template, "url-username", () => url.username, dontReplaceSlash);
  883. template = await DomUtil.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash);
  884. template = await DomUtil.evalTemplateVariable(template, "url-last-segment", () => DomUtil.getLastSegment(url), dontReplaceSlash);
  885. if (content) {
  886. template = await DomUtil.evalTemplateVariable(template, "digest-sha-256", async () => DOM.digest("SHA-256", content), dontReplaceSlash);
  887. template = await DomUtil.evalTemplateVariable(template, "digest-sha-384", async () => DOM.digest("SHA-384", content), dontReplaceSlash);
  888. template = await DomUtil.evalTemplateVariable(template, "digest-sha-512", async () => DOM.digest("SHA-512", content), dontReplaceSlash);
  889. }
  890. return template;
  891. }
  892. static setBackgroundImage(element, url, style) {
  893. element.style.setProperty("background-blend-mode", "normal", "important");
  894. element.style.setProperty("background-clip", "content-box", "important");
  895. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  896. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  897. element.style.setProperty("background-image", url, "important");
  898. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  899. element.style.setProperty("background-origin", "content-box", "important");
  900. element.style.setProperty("background-repeat", "no-repeat", "important");
  901. }
  902. static setFrameEmptySrc(frameElement) {
  903. if (frameElement.tagName == "OBJECT") {
  904. frameElement.setAttribute("data", "data:text/html,");
  905. } else {
  906. frameElement.setAttribute("srcdoc", "");
  907. frameElement.removeAttribute("src");
  908. }
  909. }
  910. static setFrameContent(frameElement, content) {
  911. if (frameElement.tagName == "OBJECT") {
  912. frameElement.setAttribute("data", "data:text/html," + content);
  913. } else {
  914. frameElement.setAttribute("srcdoc", content);
  915. frameElement.removeAttribute("src");
  916. }
  917. }
  918. static isolateElements(rootElement) {
  919. rootElement.querySelectorAll("*:not(style)").forEach(element => {
  920. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  921. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  922. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  923. element.remove();
  924. }
  925. });
  926. isolateParentElements(rootElement.parentElement, rootElement);
  927. function isolateParentElements(parentElement, element) {
  928. let elementFound = false;
  929. Array.from(parentElement.childNodes).forEach(node => {
  930. elementFound = elementFound || (node == element);
  931. const tagName = node.tagName && node.tagName.toLowerCase();
  932. if (node != element && tagName != "svg" && tagName != "style" && tagName != "link") {
  933. if (elementFound) {
  934. node.remove();
  935. } else {
  936. node.hidden = true;
  937. if (node.childNodes && node.childNodes.length) {
  938. Array.from(node.childNodes).forEach(node => node.remove());
  939. }
  940. }
  941. }
  942. });
  943. element = element.parentElement;
  944. if (element && element.parentElement && element.parentElement.tagName != "HTML") {
  945. isolateParentElements(element.parentElement, element);
  946. }
  947. }
  948. }
  949. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  950. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, options);
  951. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  952. const imports = DomUtil.getImportFunctions(stylesheetContent);
  953. await Promise.all(imports.map(async cssImport => {
  954. const match = DomUtil.matchImport(cssImport);
  955. if (match) {
  956. let resourceURL = DomUtil.normalizeURL(match.resourceURL);
  957. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  958. resourceURL = new URL(match.resourceURL, baseURI).href;
  959. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  960. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  961. let importedStylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  962. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  963. if (stylesheetContent.includes(cssImport)) {
  964. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options);
  965. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  966. }
  967. }
  968. }
  969. }
  970. }));
  971. return stylesheetContent;
  972. }
  973. static resolveStylesheetURLs(stylesheetContent, baseURI, options) {
  974. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  975. urlFunctions.map(urlFunction => {
  976. const originalResourceURL = DomUtil.matchURL(urlFunction);
  977. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  978. if (!DomUtil.testIgnoredPath(resourceURL)) {
  979. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  980. const resolvedURL = new URL(resourceURL, baseURI).href;
  981. if (DomUtil.testValidURL(resolvedURL, baseURI, options.url) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  982. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  983. }
  984. }
  985. } else {
  986. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  987. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  988. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  989. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  990. }
  991. }
  992. }
  993. });
  994. return stylesheetContent;
  995. }
  996. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options) {
  997. resourceURL = DomUtil.normalizeURL(resourceURL);
  998. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  999. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charSet: options.charSet };
  1000. let stylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  1001. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  1002. return stylesheetContent;
  1003. }
  1004. }
  1005. static async processStylesheet(doc, stylesheetContent, cssRules, baseURI, options, batchRequest) {
  1006. const urlFunctions = getURLFunctions(cssRules);
  1007. await Promise.all([
  1008. Promise.all(urlFunctions.style.map(urlFunction => processURLFunction(urlFunction))),
  1009. Promise.all(urlFunctions.font.map(urlFunction => processURLFunction(urlFunction, true)))]);
  1010. return stylesheetContent;
  1011. async function processURLFunction(urlFunction, isFont) {
  1012. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1013. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1014. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1015. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1016. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1017. const validResource = !isFont || content == EMPTY_DATA_URI || content.startsWith(PREFIX_DATA_URI_VND) || content.startsWith(PREFIX_DATA_URI_IMAGE_SVG) || await DOM.validFont(urlFunction);
  1018. if (!validResource) {
  1019. content = EMPTY_DATA_URI;
  1020. }
  1021. let regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1022. if (!stylesheetContent.match(regExpUrlFunction)) {
  1023. urlFunction = "url(" + originalResourceURL + ")";
  1024. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1025. }
  1026. if (!stylesheetContent.match(regExpUrlFunction)) {
  1027. urlFunction = "url('" + originalResourceURL + "')";
  1028. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1029. }
  1030. if (stylesheetContent.match(regExpUrlFunction)) {
  1031. if (duplicate && options.groupDuplicateImages && !isFont) {
  1032. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")");
  1033. DomUtil.insertVariable(doc, indexResource, content, options);
  1034. } else {
  1035. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, content));
  1036. }
  1037. }
  1038. }
  1039. }
  1040. }
  1041. function getURLFunctions(cssRules, urlFunctions = { style: [], font: [] }) {
  1042. Array.from(cssRules).forEach(cssRule => {
  1043. if (cssRule.type == CSSRule.MEDIA_RULE) {
  1044. getURLFunctions(cssRule.cssRules, urlFunctions);
  1045. } else if (cssRule.type == CSSRule.FONT_FACE_RULE) {
  1046. getURLFunctionsRule(cssRule.cssText, urlFunctions.font);
  1047. } else {
  1048. getURLFunctionsRule(cssRule.cssText, urlFunctions.style);
  1049. }
  1050. });
  1051. return { style: Array.from(urlFunctions.style), font: Array.from(urlFunctions.font) };
  1052. }
  1053. function getURLFunctionsRule(cssText, urlFunctions) {
  1054. const urlFunctionsRule = DomUtil.getUrlFunctions(cssText);
  1055. urlFunctionsRule.forEach(urlFunction => {
  1056. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1057. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1058. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1059. if (DomUtil.testValidURL(resourceURL, baseURI, options.url) && cssText.includes(urlFunction)) {
  1060. urlFunctions.push(urlFunction);
  1061. }
  1062. }
  1063. });
  1064. return urlFunctions;
  1065. }
  1066. }
  1067. static async processAttribute(doc, resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest, processDuplicates, removeElementIfMissing) {
  1068. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1069. let resourceURL = resourceElement.getAttribute(attributeName);
  1070. resourceURL = DomUtil.normalizeURL(resourceURL);
  1071. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1072. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1073. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1074. resourceURL = new URL(resourceURL, baseURI).href;
  1075. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1076. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1077. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1078. resourceElement.remove();
  1079. } else {
  1080. if (content.startsWith(prefixDataURI) || content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) || content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1081. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1082. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1083. if (DomUtil.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1084. DomUtil.insertVariable(doc, indexResource, content, options);
  1085. } else {
  1086. resourceElement.setAttribute(attributeName, content);
  1087. }
  1088. } else {
  1089. resourceElement.setAttribute(attributeName, content);
  1090. }
  1091. }
  1092. }
  1093. }
  1094. }
  1095. }
  1096. }));
  1097. }
  1098. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1099. const attributeName = "xlink:href";
  1100. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1101. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1102. let resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1103. if (DomUtil.testValidPath(resourceURL, baseURI, options.url) && !DomUtil.testIgnoredPath(resourceURL)) {
  1104. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1105. resourceURL = new URL(resourceURL, baseURI).href;
  1106. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1107. try {
  1108. const { content } = await batchRequest.addURL(resourceURL, false);
  1109. const DOMParser = DOM.getParser();
  1110. if (DOMParser) {
  1111. let svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  1112. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1113. if (hashMatch && hashMatch[0]) {
  1114. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1115. if (symbolElement) {
  1116. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1117. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1118. }
  1119. } else {
  1120. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1121. }
  1122. } else {
  1123. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1124. }
  1125. } catch (error) {
  1126. /* ignored */
  1127. }
  1128. }
  1129. } else if (resourceURL == options.url) {
  1130. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1131. }
  1132. }));
  1133. }
  1134. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest) {
  1135. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1136. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1137. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1138. let resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1139. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1140. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1141. resourceURL = new URL(resourceURL, baseURI).href;
  1142. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1143. const { content } = await batchRequest.addURL(resourceURL);
  1144. if (!content.startsWith(prefixDataURI) && !content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) && !content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1145. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1146. }
  1147. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1148. } else {
  1149. return "";
  1150. }
  1151. } else {
  1152. return "";
  1153. }
  1154. }
  1155. }));
  1156. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1157. }));
  1158. }
  1159. }
  1160. // -------
  1161. // DomUtil
  1162. // -------
  1163. const DATA_URI_PREFIX = "data:";
  1164. const BLOB_URI_PREFIX = "blob:";
  1165. const HTTP_URI_PREFIX = /^https?:\/\//;
  1166. const FILE_URI_PREFIX = /^file:\/\//;
  1167. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1168. const ABOUT_BLANK_URI = "about:blank";
  1169. const NOT_EMPTY_URL = /^(https?|file):\/\/.+/;
  1170. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1171. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1172. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1173. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1174. 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;
  1175. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1176. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1177. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1178. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1179. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1180. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1181. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1182. class DomUtil {
  1183. static normalizeURL(url) {
  1184. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1185. return url;
  1186. } else {
  1187. return url.split("#")[0];
  1188. }
  1189. }
  1190. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1191. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1192. if (template.match(replaceRegExp)) {
  1193. let value = await valueGetter();
  1194. if (!dontReplaceSlash) {
  1195. value = value.replace(/\/+/g, "_");
  1196. }
  1197. return template.replace(replaceRegExp, value);
  1198. }
  1199. return template;
  1200. }
  1201. static getLastSegment(url) {
  1202. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1203. if (!lastSegment) {
  1204. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1205. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1206. }
  1207. if (!lastSegment) {
  1208. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1209. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1210. }
  1211. if (!lastSegment) {
  1212. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1213. }
  1214. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1215. return lastSegment;
  1216. }
  1217. static getRegExp(string) {
  1218. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1219. }
  1220. static getUrlFunctions(stylesheetContent) {
  1221. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1222. }
  1223. static getImportFunctions(stylesheetContent) {
  1224. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1225. }
  1226. static matchURL(stylesheetContent) {
  1227. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1228. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1229. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1230. return match && match[1];
  1231. }
  1232. static testIgnoredPath(resourceURL) {
  1233. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1234. }
  1235. static testValidPath(resourceURL, baseURI, docURL) {
  1236. return resourceURL && resourceURL != baseURI && resourceURL != docURL && !resourceURL.match(EMPTY_URL);
  1237. }
  1238. static testValidURL(resourceURL, baseURI, docURL) {
  1239. return DomUtil.testValidPath(resourceURL, baseURI, docURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1240. }
  1241. static matchImport(stylesheetContent) {
  1242. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1243. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1244. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1245. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1246. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1247. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1248. if (match) {
  1249. const [, resourceURL, media] = match;
  1250. return { resourceURL, media };
  1251. }
  1252. }
  1253. static removeCssComments(stylesheetContent) {
  1254. let start, end;
  1255. do {
  1256. start = stylesheetContent.indexOf("/*");
  1257. end = stylesheetContent.indexOf("*/", start + 2);
  1258. if (start != -1 && end != -1) {
  1259. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1260. }
  1261. } while (start != -1 && end != -1);
  1262. return stylesheetContent;
  1263. }
  1264. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1265. if (mediaQuery) {
  1266. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1267. } else {
  1268. return stylesheetContent;
  1269. }
  1270. }
  1271. static insertVariable(doc, indexResource, content, options) {
  1272. const sheetAttributeName = DOM.sheetAttributeName(options.sessionId);
  1273. let styleElement = doc.querySelector("style[" + sheetAttributeName + "]"), insertedVariables;
  1274. if (!styleElement) {
  1275. styleElement = doc.createElement("style");
  1276. if (doc.head.firstChild) {
  1277. doc.head.insertBefore(styleElement, doc.head.firstChild);
  1278. } else {
  1279. doc.head.appendChild(styleElement);
  1280. }
  1281. styleElement.setAttribute(sheetAttributeName, "[]");
  1282. insertedVariables = [];
  1283. } else {
  1284. insertedVariables = JSON.parse(styleElement.getAttribute(sheetAttributeName));
  1285. }
  1286. if (!insertedVariables.includes(indexResource)) {
  1287. insertedVariables.push(indexResource);
  1288. styleElement.textContent = styleElement.textContent + `:root{${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")}`;
  1289. styleElement.setAttribute(sheetAttributeName, JSON.stringify(insertedVariables));
  1290. }
  1291. }
  1292. static replaceImageSource(imgElement, variableName, options) {
  1293. const dataAttributeName = DOM.imagesAttributeName(options.sessionId);
  1294. if (imgElement.getAttribute(dataAttributeName) != null) {
  1295. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1296. if (imgData.replaceable) {
  1297. 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>`);
  1298. const backgroundStyle = {};
  1299. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1300. if (backgroundSize) {
  1301. backgroundStyle["background-size"] = imgData.objectFit;
  1302. }
  1303. if (imgData.objectPosition) {
  1304. backgroundStyle["background-position"] = imgData.objectPosition;
  1305. }
  1306. if (imgData.backgroundColor) {
  1307. backgroundStyle["background-color"] = imgData.backgroundColor;
  1308. }
  1309. DomProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1310. imgElement.removeAttribute(dataAttributeName);
  1311. return true;
  1312. }
  1313. }
  1314. }
  1315. }
  1316. function log(...args) {
  1317. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1318. }
  1319. // -----
  1320. // Stats
  1321. // -----
  1322. const STATS_DEFAULT_VALUES = {
  1323. discarded: {
  1324. "HTML bytes": 0,
  1325. "hidden elements": 0,
  1326. "HTML imports": 0,
  1327. scripts: 0,
  1328. objects: 0,
  1329. "audio sources": 0,
  1330. "video sources": 0,
  1331. frames: 0,
  1332. "CSS rules": 0,
  1333. canvas: 0,
  1334. stylesheets: 0,
  1335. resources: 0,
  1336. medias: 0
  1337. },
  1338. processed: {
  1339. "HTML bytes": 0,
  1340. "hidden elements": 0,
  1341. "HTML imports": 0,
  1342. scripts: 0,
  1343. objects: 0,
  1344. "audio sources": 0,
  1345. "video sources": 0,
  1346. frames: 0,
  1347. "CSS rules": 0,
  1348. canvas: 0,
  1349. stylesheets: 0,
  1350. resources: 0,
  1351. medias: 0
  1352. }
  1353. };
  1354. class Stats {
  1355. constructor(options) {
  1356. this.options = options;
  1357. if (options.displayStats) {
  1358. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1359. }
  1360. }
  1361. set(type, subType, value) {
  1362. if (this.options.displayStats) {
  1363. this.data[type][subType] = value;
  1364. }
  1365. }
  1366. add(type, subType, value) {
  1367. if (this.options.displayStats) {
  1368. this.data[type][subType] += value;
  1369. }
  1370. }
  1371. addAll(pageData) {
  1372. if (this.options.displayStats) {
  1373. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1374. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1375. }
  1376. }
  1377. }
  1378. return { getClass };
  1379. })();