single-file-core.js 64 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485
  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. const descriptionElement = this.doc.querySelector("meta[name=description]");
  416. const authorElement = this.doc.querySelector("meta[name=author]");
  417. const creatorElement = this.doc.querySelector("meta[name=creator]");
  418. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  419. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  420. this.options.infobarContent = await DomProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  421. this.options.info = {
  422. description: descriptionElement ? descriptionElement.content.trim() : "",
  423. lang: this.doc.documentElement.lang,
  424. author: authorElement ? authorElement.content.trim() : "",
  425. creator: creatorElement ? creatorElement.content.trim() : "",
  426. publisher: publisherElement ? publisherElement.content.trim() : ""
  427. };
  428. }
  429. preProcessPage() {
  430. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  431. if (this.options.win) {
  432. this.doc.body.querySelectorAll(":not(svg) title, meta").forEach(element => element instanceof this.options.win.HTMLElement && this.doc.head.appendChild(element));
  433. }
  434. if (this.options.imageData) {
  435. const dataAttributeName = DOM.imagesAttributeName(this.options.sessionId);
  436. this.doc.querySelectorAll("img").forEach(imgElement => {
  437. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  438. if (imgData.currentSrc) {
  439. imgElement.setAttribute("src", imgData.currentSrc);
  440. }
  441. });
  442. if (this.options.lazyLoadImages) {
  443. this.doc.querySelectorAll("img[data-src]").forEach(imgElement => {
  444. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  445. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  446. imgData.src = imgElement.dataset.src;
  447. imgElement.setAttribute("src", imgElement.dataset.src);
  448. imgElement.removeAttribute("data-src");
  449. }
  450. });
  451. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  452. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  453. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  454. imgElement.removeAttribute("data-srcset");
  455. }
  456. });
  457. }
  458. }
  459. }
  460. removeScripts() {
  461. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  462. this.stats.set("discarded", "scripts", scriptElements.length);
  463. this.stats.set("processed", "scripts", scriptElements.length);
  464. scriptElements.forEach(element => element.remove());
  465. }
  466. removeFrames() {
  467. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  468. this.stats.set("discarded", "frames", frameElements.length);
  469. this.stats.set("processed", "frames", frameElements.length);
  470. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  471. }
  472. removeImports() {
  473. const importElements = this.doc.querySelectorAll("link[rel=import]");
  474. this.stats.set("discarded", "HTML imports", importElements.length);
  475. this.stats.set("processed", "HTML imports", importElements.length);
  476. importElements.forEach(element => element.remove());
  477. }
  478. resetCharsetMeta() {
  479. let charSet;
  480. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  481. const charSetDeclaration = element.content.split(";")[1];
  482. if (charSetDeclaration && !charSet) {
  483. charSet = charSetDeclaration.split("=")[1];
  484. if (charSet) {
  485. this.charSet = charSet.trim().toLowerCase();
  486. }
  487. }
  488. element.remove();
  489. });
  490. const metaElement = this.doc.createElement("meta");
  491. metaElement.setAttribute("charset", "utf-8");
  492. if (this.doc.head.firstChild) {
  493. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  494. } else {
  495. this.doc.head.appendChild(metaElement);
  496. }
  497. }
  498. insertFaviconLink() {
  499. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  500. if (!faviconElement) {
  501. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  502. }
  503. if (!faviconElement) {
  504. faviconElement = this.doc.createElement("link");
  505. faviconElement.setAttribute("type", "image/x-icon");
  506. faviconElement.setAttribute("rel", "shortcut icon");
  507. faviconElement.setAttribute("href", "/favicon.ico");
  508. }
  509. this.doc.head.appendChild(faviconElement);
  510. }
  511. resolveHrefs() {
  512. this.doc.querySelectorAll("[href]").forEach(element => {
  513. if (element.href) {
  514. const href = element.href.baseVal ? element.href.baseVal : element.href;
  515. const normalizedHref = DomUtil.normalizeURL(href);
  516. if (normalizedHref == href && href) {
  517. element.setAttribute("href", href);
  518. }
  519. }
  520. });
  521. }
  522. removeUnusedStyles() {
  523. if (!this.mediaAllInfo) {
  524. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  525. }
  526. const stats = DOM.minifyCSS(this.doc, this.mediaAllInfo);
  527. this.stats.set("processed", "CSS rules", stats.processed);
  528. this.stats.set("discarded", "CSS rules", stats.discarded);
  529. }
  530. removeUnusedFonts() {
  531. DOM.removeUnusedFonts(this.doc, this.options);
  532. }
  533. removeAlternativeFonts() {
  534. DOM.removeAlternativeFonts(this.doc);
  535. }
  536. removeAlternativeImages() {
  537. DOM.removeAlternativeImages(this.doc);
  538. }
  539. removeHiddenElements() {
  540. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  541. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  542. this.stats.set("processed", "hidden elements", hiddenElements.length);
  543. hiddenElements.forEach(element => element.remove());
  544. }
  545. compressHTML() {
  546. let size;
  547. if (this.options.displayStats) {
  548. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  549. }
  550. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  551. if (this.options.displayStats) {
  552. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  553. }
  554. }
  555. removeAlternativeMedias() {
  556. const stats = DOM.minifyMedias(this.doc);
  557. this.stats.set("processed", "medias", stats.processed);
  558. this.stats.set("discarded", "medias", stats.discarded);
  559. }
  560. compressCSS() {
  561. this.doc.querySelectorAll("style").forEach(styleElement => {
  562. if (styleElement) {
  563. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  564. }
  565. });
  566. this.doc.querySelectorAll("[style]").forEach(element => {
  567. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  568. });
  569. }
  570. replaceCanvasElements() {
  571. if (this.options.canvasData) {
  572. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  573. const canvasData = this.options.canvasData[indexCanvasElement];
  574. if (canvasData) {
  575. DomProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  576. this.stats.add("processed", "canvas", 1);
  577. }
  578. });
  579. }
  580. }
  581. insertFonts() {
  582. if (this.options.fontsData && this.options.fontsData.length) {
  583. let stylesheetContent = "";
  584. this.options.fontsData.forEach(fontStyles => {
  585. if (fontStyles["font-family"] && fontStyles.src) {
  586. stylesheetContent += "@font-face{";
  587. let stylesContent = "";
  588. Object.keys(fontStyles).forEach(fontStyle => {
  589. if (stylesContent) {
  590. stylesContent += ";";
  591. }
  592. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  593. });
  594. stylesheetContent += stylesContent + "}";
  595. }
  596. });
  597. if (stylesheetContent) {
  598. const styleElement = this.doc.createElement("style");
  599. styleElement.textContent = stylesheetContent;
  600. const existingStyleElement = this.doc.querySelector("style");
  601. if (existingStyleElement) {
  602. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  603. } else {
  604. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  605. }
  606. }
  607. }
  608. }
  609. replaceStyleContents() {
  610. if (this.options.stylesheetContents) {
  611. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  612. if (this.options.stylesheetContents[styleIndex]) {
  613. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  614. }
  615. });
  616. }
  617. }
  618. insertVideoPosters() {
  619. if (this.options.postersData) {
  620. this.doc.querySelectorAll("video[src], video > source[src]").forEach((element, videoIndex) => {
  621. let videoElement;
  622. if (element.tagName == "VIDEO") {
  623. videoElement = element;
  624. } else {
  625. videoElement = element.parentElement;
  626. }
  627. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  628. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  629. }
  630. });
  631. }
  632. }
  633. async processPageResources() {
  634. const resourcePromises = [
  635. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", "data:", this.baseURI, this.options, this.batchRequest, false, true),
  636. 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),
  637. 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),
  638. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.options, this.batchRequest),
  639. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  640. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  641. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  642. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  643. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest)
  644. ];
  645. if (!this.options.removeAudioSrc) {
  646. 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));
  647. }
  648. if (!this.options.removeVideoSrc) {
  649. 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));
  650. }
  651. await Promise.all(resourcePromises);
  652. if (this.options.removeAlternativeImages) {
  653. let shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  654. if (!shortcutIcon) {
  655. shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  656. if (shortcutIcon) {
  657. shortcutIcon.rel = "icon";
  658. }
  659. }
  660. if (shortcutIcon) {
  661. this.doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  662. if (linkElement != shortcutIcon) {
  663. linkElement.remove();
  664. }
  665. });
  666. }
  667. }
  668. function findShortcutIcon(shortcutIcons) {
  669. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  670. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  671. return shortcutIcons[0];
  672. }
  673. }
  674. async resolveStylesheetURLs() {
  675. await Promise.all(Array.from(this.doc.querySelectorAll("style"))
  676. .map(async styleElement => {
  677. const options = { url: this.options.url, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled };
  678. const textContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, options);
  679. styleElement.textContent = textContent;
  680. }));
  681. }
  682. async processStylesheets() {
  683. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => {
  684. this.stats.add("processed", "stylesheets", 1);
  685. if (styleElement.sheet) {
  686. styleElement.textContent = await DomProcessorHelper.processStylesheet(this.doc, styleElement.textContent, styleElement.sheet.cssRules, this.baseURI, this.options, this.batchRequest);
  687. } else {
  688. styleElement.remove();
  689. }
  690. }));
  691. }
  692. async processScripts() {
  693. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  694. if (scriptElement.src) {
  695. this.stats.add("processed", "scripts", 1);
  696. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  697. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  698. }
  699. scriptElement.removeAttribute("src");
  700. }));
  701. }
  702. async resolveFrameURLs() {
  703. if (this.options.framesData) {
  704. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  705. await Promise.all(frameElements.map(async frameElement => {
  706. DomProcessorHelper.setFrameEmptySrc(frameElement);
  707. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  708. if (frameWindowId) {
  709. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  710. if (frameData) {
  711. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  712. }
  713. }
  714. }));
  715. }
  716. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  717. options.insertSingleFileComment = false;
  718. options.insertFaviconLink = false;
  719. options.doc = null;
  720. options.win = null;
  721. options.url = frameData.baseURI;
  722. options.windowId = frameWindowId;
  723. if (frameData.content) {
  724. options.content = frameData.content;
  725. options.canvasData = frameData.canvasData;
  726. options.stylesheetContents = frameData.stylesheetContents;
  727. options.fontsData = frameData.fontsData;
  728. options.imageData = frameData.imageData;
  729. options.usedFonts = frameData.usedFonts;
  730. frameData.processor = new PageProcessor(options);
  731. frameData.frameElement = frameElement;
  732. await frameData.processor.loadPage();
  733. await frameData.processor.initialize();
  734. frameData.maxResources = batchRequest.getMaxResources();
  735. }
  736. }
  737. }
  738. async processFrames() {
  739. if (this.options.framesData) {
  740. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  741. await Promise.all(frameElements.map(async frameElement => {
  742. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  743. if (frameWindowId) {
  744. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  745. if (frameData) {
  746. if (frameData.processor) {
  747. this.stats.add("processed", "frames", 1);
  748. await frameData.processor.run();
  749. const pageData = await frameData.processor.getPageData();
  750. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  751. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  752. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  753. } else {
  754. frameElement.setAttribute("sandbox", "");
  755. }
  756. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  757. this.stats.addAll(pageData);
  758. } else {
  759. this.stats.add("discarded", "frames", 1);
  760. }
  761. }
  762. }
  763. }));
  764. }
  765. }
  766. async resolveHtmlImportURLs() {
  767. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  768. if (!this.relImportProcessors) {
  769. this.relImportProcessors = new Map();
  770. }
  771. await Promise.all(linkElements.map(async linkElement => {
  772. const resourceURL = linkElement.href;
  773. linkElement.removeAttribute("href");
  774. const options = Object.create(this.options);
  775. options.insertSingleFileComment = false;
  776. options.insertFaviconLink = false;
  777. options.doc = null;
  778. options.win = null;
  779. options.url = resourceURL;
  780. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, this.baseURI, this.options.url)) {
  781. const processor = new PageProcessor(options);
  782. this.relImportProcessors.set(linkElement, processor);
  783. await processor.loadPage();
  784. return processor.initialize();
  785. }
  786. }));
  787. }
  788. async processHtmlImports() {
  789. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  790. await Promise.all(linkElements.map(async linkElement => {
  791. const processor = this.relImportProcessors.get(linkElement);
  792. if (processor) {
  793. this.stats.add("processed", "HTML imports", 1);
  794. this.relImportProcessors.delete(linkElement);
  795. const pageData = await processor.getPageData();
  796. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  797. this.stats.addAll(pageData);
  798. } else {
  799. this.stats.add("discarded", "HTML imports", 1);
  800. }
  801. }));
  802. }
  803. resolveStyleAttributeURLs() {
  804. Array.from(this.doc.querySelectorAll("[style]"))
  805. .map(element => element.setAttribute("style", DomProcessorHelper.resolveStylesheetURLs(element.getAttribute("style"), this.baseURI, this.options)));
  806. }
  807. async processStyleAttributes() {
  808. await Promise.all(Array.from(this.doc.querySelectorAll("[style]"))
  809. .map(async element => {
  810. const rules = [{ type: CSSRule.STYLE_RULE, cssText: element.getAttribute("style") }];
  811. const textContent = await DomProcessorHelper.processStylesheet(this.doc, element.getAttribute("style"), rules, this.baseURI, this.options, this.batchRequest);
  812. element.setAttribute("style", textContent);
  813. }));
  814. }
  815. async resolveLinkedStylesheetURLs() {
  816. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  817. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charSet: this.charSet };
  818. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, options);
  819. if (stylesheetContent && (!linkElement.rel.includes("alternate") || !linkElement.title)) {
  820. const styleElement = this.doc.createElement("style");
  821. if (linkElement.media) {
  822. styleElement.media = linkElement.media;
  823. }
  824. styleElement.textContent = stylesheetContent;
  825. linkElement.parentElement.replaceChild(styleElement, linkElement);
  826. } else {
  827. linkElement.remove();
  828. }
  829. }));
  830. }
  831. }
  832. // ---------
  833. // DomHelper
  834. // ---------
  835. const REGEXP_AMP = /&/g;
  836. const REGEXP_NBSP = /\u00a0/g;
  837. const REGEXP_START_TAG = /</g;
  838. const REGEXP_END_TAG = />/g;
  839. const REGEXP_URL_HASH = /(#.+?)$/;
  840. const PREFIX_DATA_URI_IMAGE = "data:image/";
  841. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  842. const PREFIX_DATA_URI_VIDEO = "data:video/";
  843. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  844. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  845. const PREFIX_DATA_URI_OCTET_STREAM = /^data:(application|binary)\/octet-stream/;
  846. const PREFIX_DATA_URI_VND = "data:application/vnd.";
  847. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  848. class DomProcessorHelper {
  849. static async evalTemplate(template, options, content, dontReplaceSlash) {
  850. const date = new Date();
  851. const url = new URL(options.url);
  852. template = await DomUtil.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash);
  853. template = await DomUtil.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash);
  854. template = await DomUtil.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash);
  855. template = await DomUtil.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash);
  856. template = await DomUtil.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash);
  857. template = await DomUtil.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash);
  858. template = await DomUtil.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash);
  859. template = await DomUtil.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash);
  860. template = await DomUtil.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash);
  861. template = await DomUtil.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash);
  862. template = await DomUtil.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash);
  863. template = await DomUtil.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash);
  864. template = await DomUtil.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  865. template = await DomUtil.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash);
  866. template = await DomUtil.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash);
  867. template = await DomUtil.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash);
  868. template = await DomUtil.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash);
  869. template = await DomUtil.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  870. template = await DomUtil.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash);
  871. template = await DomUtil.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash);
  872. template = await DomUtil.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash);
  873. template = await DomUtil.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash);
  874. template = await DomUtil.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash);
  875. template = await DomUtil.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash);
  876. template = await DomUtil.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash);
  877. template = await DomUtil.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1), dontReplaceSlash);
  878. template = await DomUtil.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, ""), dontReplaceSlash);
  879. template = await DomUtil.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, ""), dontReplaceSlash);
  880. template = await DomUtil.evalTemplateVariable(template, "url-href", () => url.href, dontReplaceSlash);
  881. template = await DomUtil.evalTemplateVariable(template, "url-password", () => url.password, dontReplaceSlash);
  882. template = await DomUtil.evalTemplateVariable(template, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), dontReplaceSlash === undefined ? true : dontReplaceSlash);
  883. template = await DomUtil.evalTemplateVariable(template, "url-port", () => url.port, dontReplaceSlash);
  884. template = await DomUtil.evalTemplateVariable(template, "url-protocol", () => url.protocol, dontReplaceSlash);
  885. template = await DomUtil.evalTemplateVariable(template, "url-search", () => url.search.substring(1), dontReplaceSlash);
  886. template = await DomUtil.evalTemplateVariable(template, "url-username", () => url.username, dontReplaceSlash);
  887. template = await DomUtil.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash);
  888. template = await DomUtil.evalTemplateVariable(template, "url-last-segment", () => DomUtil.getLastSegment(url), dontReplaceSlash);
  889. if (content) {
  890. template = await DomUtil.evalTemplateVariable(template, "digest-sha-256", async () => DOM.digest("SHA-256", content), dontReplaceSlash);
  891. template = await DomUtil.evalTemplateVariable(template, "digest-sha-384", async () => DOM.digest("SHA-384", content), dontReplaceSlash);
  892. template = await DomUtil.evalTemplateVariable(template, "digest-sha-512", async () => DOM.digest("SHA-512", content), dontReplaceSlash);
  893. }
  894. return template;
  895. }
  896. static setBackgroundImage(element, url, style) {
  897. element.style.setProperty("background-blend-mode", "normal", "important");
  898. element.style.setProperty("background-clip", "content-box", "important");
  899. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  900. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  901. element.style.setProperty("background-image", url, "important");
  902. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  903. element.style.setProperty("background-origin", "content-box", "important");
  904. element.style.setProperty("background-repeat", "no-repeat", "important");
  905. }
  906. static setFrameEmptySrc(frameElement) {
  907. if (frameElement.tagName == "OBJECT") {
  908. frameElement.setAttribute("data", "data:text/html,");
  909. } else {
  910. frameElement.setAttribute("srcdoc", "");
  911. frameElement.removeAttribute("src");
  912. }
  913. }
  914. static setFrameContent(frameElement, content) {
  915. if (frameElement.tagName == "OBJECT") {
  916. frameElement.setAttribute("data", "data:text/html," + content);
  917. } else {
  918. frameElement.setAttribute("srcdoc", content);
  919. frameElement.removeAttribute("src");
  920. }
  921. }
  922. static isolateElements(rootElement) {
  923. rootElement.querySelectorAll("*:not(style)").forEach(element => {
  924. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  925. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  926. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  927. element.remove();
  928. }
  929. });
  930. isolateParentElements(rootElement.parentElement, rootElement);
  931. function isolateParentElements(parentElement, element) {
  932. let elementFound = false;
  933. Array.from(parentElement.childNodes).forEach(node => {
  934. elementFound = elementFound || (node == element);
  935. const tagName = node.tagName && node.tagName.toLowerCase();
  936. if (node != element && tagName != "svg" && tagName != "style" && tagName != "link") {
  937. if (elementFound) {
  938. node.remove();
  939. } else {
  940. node.hidden = true;
  941. if (node.childNodes && node.childNodes.length) {
  942. Array.from(node.childNodes).forEach(node => node.remove());
  943. }
  944. }
  945. }
  946. });
  947. element = element.parentElement;
  948. if (element && element.parentElement && element.parentElement.tagName != "HTML") {
  949. isolateParentElements(element.parentElement, element);
  950. }
  951. }
  952. }
  953. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  954. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, options);
  955. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  956. const imports = DomUtil.getImportFunctions(stylesheetContent);
  957. await Promise.all(imports.map(async cssImport => {
  958. const match = DomUtil.matchImport(cssImport);
  959. if (match) {
  960. let resourceURL = DomUtil.normalizeURL(match.resourceURL);
  961. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  962. resourceURL = new URL(match.resourceURL, baseURI).href;
  963. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  964. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  965. let importedStylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  966. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  967. if (stylesheetContent.includes(cssImport)) {
  968. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options);
  969. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  970. }
  971. }
  972. }
  973. }
  974. }));
  975. return stylesheetContent;
  976. }
  977. static resolveStylesheetURLs(stylesheetContent, baseURI, options) {
  978. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  979. urlFunctions.map(urlFunction => {
  980. const originalResourceURL = DomUtil.matchURL(urlFunction);
  981. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  982. if (!DomUtil.testIgnoredPath(resourceURL)) {
  983. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  984. const resolvedURL = new URL(resourceURL, baseURI).href;
  985. if (DomUtil.testValidURL(resolvedURL, baseURI, options.url) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  986. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  987. }
  988. }
  989. } else {
  990. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  991. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  992. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  993. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  994. }
  995. }
  996. }
  997. });
  998. return stylesheetContent;
  999. }
  1000. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options) {
  1001. resourceURL = DomUtil.normalizeURL(resourceURL);
  1002. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1003. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charSet: options.charSet };
  1004. let stylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  1005. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  1006. return stylesheetContent;
  1007. }
  1008. }
  1009. static async processStylesheet(doc, stylesheetContent, cssRules, baseURI, options, batchRequest) {
  1010. const urlFunctions = getURLFunctions(cssRules);
  1011. await Promise.all([
  1012. Promise.all(urlFunctions.style.map(urlFunction => processURLFunction(urlFunction))),
  1013. Promise.all(urlFunctions.font.map(urlFunction => processURLFunction(urlFunction, true)))]);
  1014. return stylesheetContent;
  1015. async function processURLFunction(urlFunction, isFont) {
  1016. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1017. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1018. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1019. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1020. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1021. const validResource = !isFont || content == EMPTY_DATA_URI || content.startsWith(PREFIX_DATA_URI_VND) || content.startsWith(PREFIX_DATA_URI_IMAGE_SVG) || await DOM.validFont(urlFunction);
  1022. if (!validResource) {
  1023. content = EMPTY_DATA_URI;
  1024. }
  1025. let regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1026. if (!stylesheetContent.match(regExpUrlFunction)) {
  1027. urlFunction = "url(" + originalResourceURL + ")";
  1028. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1029. }
  1030. if (!stylesheetContent.match(regExpUrlFunction)) {
  1031. urlFunction = "url('" + originalResourceURL + "')";
  1032. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1033. }
  1034. if (stylesheetContent.match(regExpUrlFunction)) {
  1035. if (duplicate && options.groupDuplicateImages && !isFont) {
  1036. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")");
  1037. DomUtil.insertVariable(doc, indexResource, content, options);
  1038. } else {
  1039. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, content));
  1040. }
  1041. }
  1042. }
  1043. }
  1044. }
  1045. function getURLFunctions(cssRules, urlFunctions = { style: [], font: [] }) {
  1046. Array.from(cssRules).forEach(cssRule => {
  1047. if (cssRule.type == CSSRule.MEDIA_RULE) {
  1048. getURLFunctions(cssRule.cssRules, urlFunctions);
  1049. } else if (cssRule.type == CSSRule.FONT_FACE_RULE) {
  1050. getURLFunctionsRule(cssRule.cssText, urlFunctions.font);
  1051. } else {
  1052. getURLFunctionsRule(cssRule.cssText, urlFunctions.style);
  1053. }
  1054. });
  1055. return { style: Array.from(urlFunctions.style), font: Array.from(urlFunctions.font) };
  1056. }
  1057. function getURLFunctionsRule(cssText, urlFunctions) {
  1058. const urlFunctionsRule = DomUtil.getUrlFunctions(cssText);
  1059. urlFunctionsRule.forEach(urlFunction => {
  1060. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1061. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1062. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1063. if (DomUtil.testValidURL(resourceURL, baseURI, options.url) && cssText.includes(urlFunction)) {
  1064. urlFunctions.push(urlFunction);
  1065. }
  1066. }
  1067. });
  1068. return urlFunctions;
  1069. }
  1070. }
  1071. static async processAttribute(doc, resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest, processDuplicates, removeElementIfMissing) {
  1072. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1073. let resourceURL = resourceElement.getAttribute(attributeName);
  1074. resourceURL = DomUtil.normalizeURL(resourceURL);
  1075. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1076. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1077. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1078. resourceURL = new URL(resourceURL, baseURI).href;
  1079. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1080. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1081. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1082. resourceElement.remove();
  1083. } else {
  1084. if (content.startsWith(prefixDataURI) || content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) || content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1085. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1086. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1087. if (DomUtil.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1088. DomUtil.insertVariable(doc, indexResource, content, options);
  1089. } else {
  1090. resourceElement.setAttribute(attributeName, content);
  1091. }
  1092. } else {
  1093. resourceElement.setAttribute(attributeName, content);
  1094. }
  1095. }
  1096. }
  1097. }
  1098. }
  1099. }
  1100. }));
  1101. }
  1102. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1103. const attributeName = "xlink:href";
  1104. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1105. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1106. let resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1107. if (DomUtil.testValidPath(resourceURL, baseURI, options.url) && !DomUtil.testIgnoredPath(resourceURL)) {
  1108. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1109. resourceURL = new URL(resourceURL, baseURI).href;
  1110. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1111. try {
  1112. const { content } = await batchRequest.addURL(resourceURL, false);
  1113. const DOMParser = DOM.getParser();
  1114. if (DOMParser) {
  1115. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  1116. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1117. if (hashMatch && hashMatch[0]) {
  1118. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1119. if (symbolElement) {
  1120. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1121. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1122. }
  1123. } else {
  1124. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1125. }
  1126. } else {
  1127. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1128. }
  1129. } catch (error) {
  1130. /* ignored */
  1131. }
  1132. }
  1133. } else if (resourceURL == options.url) {
  1134. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1135. }
  1136. }));
  1137. }
  1138. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest) {
  1139. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1140. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1141. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1142. let resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1143. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1144. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1145. resourceURL = new URL(resourceURL, baseURI).href;
  1146. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1147. const { content } = await batchRequest.addURL(resourceURL);
  1148. if (!content.startsWith(prefixDataURI) && !content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) && !content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1149. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1150. }
  1151. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1152. } else {
  1153. return "";
  1154. }
  1155. } else {
  1156. return "";
  1157. }
  1158. }
  1159. }));
  1160. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1161. }));
  1162. }
  1163. }
  1164. // -------
  1165. // DomUtil
  1166. // -------
  1167. const DATA_URI_PREFIX = "data:";
  1168. const BLOB_URI_PREFIX = "blob:";
  1169. const HTTP_URI_PREFIX = /^https?:\/\//;
  1170. const FILE_URI_PREFIX = /^file:\/\//;
  1171. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1172. const ABOUT_BLANK_URI = "about:blank";
  1173. const NOT_EMPTY_URL = /^(https?|file):\/\/.+/;
  1174. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1175. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1176. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1177. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1178. 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;
  1179. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1180. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1181. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1182. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1183. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1184. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1185. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1186. class DomUtil {
  1187. static normalizeURL(url) {
  1188. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1189. return url;
  1190. } else {
  1191. return url.split("#")[0];
  1192. }
  1193. }
  1194. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1195. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1196. if (template.match(replaceRegExp)) {
  1197. let value = await valueGetter();
  1198. if (!dontReplaceSlash) {
  1199. value = value.replace(/\/+/g, "_");
  1200. }
  1201. return template.replace(replaceRegExp, value);
  1202. }
  1203. return template;
  1204. }
  1205. static getLastSegment(url) {
  1206. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1207. if (!lastSegment) {
  1208. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1209. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1210. }
  1211. if (!lastSegment) {
  1212. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1213. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1214. }
  1215. if (!lastSegment) {
  1216. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1217. }
  1218. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1219. return lastSegment;
  1220. }
  1221. static getRegExp(string) {
  1222. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1223. }
  1224. static getUrlFunctions(stylesheetContent) {
  1225. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1226. }
  1227. static getImportFunctions(stylesheetContent) {
  1228. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1229. }
  1230. static matchURL(stylesheetContent) {
  1231. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1232. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1233. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1234. return match && match[1];
  1235. }
  1236. static testIgnoredPath(resourceURL) {
  1237. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1238. }
  1239. static testValidPath(resourceURL, baseURI, docURL) {
  1240. return resourceURL && resourceURL != baseURI && resourceURL != docURL && !resourceURL.match(EMPTY_URL);
  1241. }
  1242. static testValidURL(resourceURL, baseURI, docURL) {
  1243. return DomUtil.testValidPath(resourceURL, baseURI, docURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1244. }
  1245. static matchImport(stylesheetContent) {
  1246. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1247. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1248. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1249. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1250. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1251. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1252. if (match) {
  1253. const [, resourceURL, media] = match;
  1254. return { resourceURL, media };
  1255. }
  1256. }
  1257. static removeCssComments(stylesheetContent) {
  1258. let start, end;
  1259. do {
  1260. start = stylesheetContent.indexOf("/*");
  1261. end = stylesheetContent.indexOf("*/", start + 2);
  1262. if (start != -1 && end != -1) {
  1263. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1264. }
  1265. } while (start != -1 && end != -1);
  1266. return stylesheetContent;
  1267. }
  1268. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1269. if (mediaQuery) {
  1270. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1271. } else {
  1272. return stylesheetContent;
  1273. }
  1274. }
  1275. static insertVariable(doc, indexResource, content, options) {
  1276. const sheetAttributeName = DOM.sheetAttributeName(options.sessionId);
  1277. let styleElement = doc.querySelector("style[" + sheetAttributeName + "]"), insertedVariables;
  1278. if (!styleElement) {
  1279. styleElement = doc.createElement("style");
  1280. if (doc.head.firstChild) {
  1281. doc.head.insertBefore(styleElement, doc.head.firstChild);
  1282. } else {
  1283. doc.head.appendChild(styleElement);
  1284. }
  1285. styleElement.setAttribute(sheetAttributeName, "[]");
  1286. insertedVariables = [];
  1287. } else {
  1288. insertedVariables = JSON.parse(styleElement.getAttribute(sheetAttributeName));
  1289. }
  1290. if (!insertedVariables.includes(indexResource)) {
  1291. insertedVariables.push(indexResource);
  1292. styleElement.textContent = styleElement.textContent + `:root{${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")}`;
  1293. styleElement.setAttribute(sheetAttributeName, JSON.stringify(insertedVariables));
  1294. }
  1295. }
  1296. static replaceImageSource(imgElement, variableName, options) {
  1297. const dataAttributeName = DOM.imagesAttributeName(options.sessionId);
  1298. if (imgElement.getAttribute(dataAttributeName) != null) {
  1299. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1300. if (imgData.replaceable) {
  1301. 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>`);
  1302. const backgroundStyle = {};
  1303. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1304. if (backgroundSize) {
  1305. backgroundStyle["background-size"] = imgData.objectFit;
  1306. }
  1307. if (imgData.objectPosition) {
  1308. backgroundStyle["background-position"] = imgData.objectPosition;
  1309. }
  1310. if (imgData.backgroundColor) {
  1311. backgroundStyle["background-color"] = imgData.backgroundColor;
  1312. }
  1313. DomProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1314. imgElement.removeAttribute(dataAttributeName);
  1315. return true;
  1316. }
  1317. }
  1318. }
  1319. }
  1320. function log(...args) {
  1321. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1322. }
  1323. // -----
  1324. // Stats
  1325. // -----
  1326. const STATS_DEFAULT_VALUES = {
  1327. discarded: {
  1328. "HTML bytes": 0,
  1329. "hidden elements": 0,
  1330. "HTML imports": 0,
  1331. scripts: 0,
  1332. objects: 0,
  1333. "audio sources": 0,
  1334. "video sources": 0,
  1335. frames: 0,
  1336. "CSS rules": 0,
  1337. canvas: 0,
  1338. stylesheets: 0,
  1339. resources: 0,
  1340. medias: 0
  1341. },
  1342. processed: {
  1343. "HTML bytes": 0,
  1344. "hidden elements": 0,
  1345. "HTML imports": 0,
  1346. scripts: 0,
  1347. objects: 0,
  1348. "audio sources": 0,
  1349. "video sources": 0,
  1350. frames: 0,
  1351. "CSS rules": 0,
  1352. canvas: 0,
  1353. stylesheets: 0,
  1354. resources: 0,
  1355. medias: 0
  1356. }
  1357. };
  1358. class Stats {
  1359. constructor(options) {
  1360. this.options = options;
  1361. if (options.displayStats) {
  1362. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1363. }
  1364. }
  1365. set(type, subType, value) {
  1366. if (this.options.displayStats) {
  1367. this.data[type][subType] = value;
  1368. }
  1369. }
  1370. add(type, subType, value) {
  1371. if (this.options.displayStats) {
  1372. this.data[type][subType] += value;
  1373. }
  1374. }
  1375. addAll(pageData) {
  1376. if (this.options.displayStats) {
  1377. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1378. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1379. }
  1380. }
  1381. }
  1382. return { getClass };
  1383. })();