single-file-core.js 31 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801
  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. this.SingleFileCore = (() => {
  21. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  22. const REMOVED_CONTENT_ATTRIBUTE_NAME = "data-single-file-removed-content";
  23. let Download, DOM, URL;
  24. function SingleFileCore(...args) {
  25. [Download, DOM, URL] = args;
  26. return class {
  27. constructor(options) {
  28. this.options = options;
  29. this.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  30. this.REMOVED_CONTENT_ATTRIBUTE_NAME = REMOVED_CONTENT_ATTRIBUTE_NAME;
  31. }
  32. async initialize() {
  33. this.processor = new PageProcessor(this.options);
  34. await this.processor.loadPage();
  35. await this.processor.initialize();
  36. }
  37. async preparePageData() {
  38. await this.processor.preparePageData();
  39. }
  40. getPageData() {
  41. return this.processor.getPageData();
  42. }
  43. };
  44. }
  45. // -------------
  46. // ProgressEvent
  47. // -------------
  48. const PAGE_LOADING = "page-loading";
  49. const PAGE_LOADED = "page-loaded";
  50. const RESOURCES_INITIALIZING = "resource-initializing";
  51. const RESOURCES_INITIALIZED = "resources-initialized";
  52. const RESOURCE_LOADING = "resource-loading";
  53. const RESOURCE_LOADED = "resource-loaded";
  54. const PAGE_ENDED = "page-ended";
  55. class ProgressEvent {
  56. constructor(type, details) {
  57. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADING, RESOURCE_LOADED, PAGE_ENDED };
  58. }
  59. }
  60. // -------------
  61. // PageProcessor
  62. // -------------
  63. class PageProcessor {
  64. constructor(options) {
  65. this.options = options;
  66. this.processor = new DOMProcessor(options);
  67. this.onprogress = options.onprogress;
  68. }
  69. async loadPage() {
  70. if (this.onprogress) {
  71. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  72. }
  73. await this.processor.loadPage(this.options.content);
  74. if (this.onprogress) {
  75. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  76. }
  77. }
  78. async initialize() {
  79. if (this.onprogress) {
  80. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  81. }
  82. if (!this.options.jsEnabled || (this.options.saveRawPage && this.options.removeScripts)) {
  83. this.processor.insertNoscriptContents();
  84. }
  85. if (this.options.removeFrames) {
  86. this.processor.removeFrames();
  87. }
  88. if (this.options.removeImports) {
  89. this.processor.removeImports();
  90. }
  91. if (this.options.removeScripts) {
  92. this.processor.removeScripts();
  93. }
  94. this.processor.removeDiscardedResources();
  95. this.processor.resetCharsetMeta();
  96. if (this.options.compress) {
  97. this.processor.compressHTML();
  98. }
  99. if (this.options.insertFaviconLink) {
  100. this.processor.insertFaviconLink();
  101. }
  102. this.processor.resolveHrefs();
  103. if (this.options.insertSingleFileComment) {
  104. this.processor.insertSingleFileCommentNode();
  105. }
  106. this.processor.replaceCanvasElements();
  107. if (this.options.removeHiddenElements) {
  108. this.processor.removeHiddenElements();
  109. }
  110. const initializationPromises = [this.processor.inlineStylesheets(true), this.processor.linkStylesheets(), this.processor.attributeStyles(true)];
  111. if (!this.options.removeFrames) {
  112. initializationPromises.push(this.processor.frames(true));
  113. }
  114. if (!this.options.removeImports) {
  115. initializationPromises.push(this.processor.htmlImports(true));
  116. }
  117. await Promise.all(initializationPromises);
  118. if (this.options.removeUnusedCSSRules) {
  119. this.processor.removeUnusedCSSRules();
  120. }
  121. this.pendingPromises = [this.processor.inlineStylesheets(), this.processor.attributeStyles(), this.processor.pageResources()];
  122. if (!this.options.removeScripts) {
  123. this.pendingPromises.push(this.processor.scripts());
  124. }
  125. if (this.onprogress) {
  126. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: batchRequest.getMaxResources() }));
  127. }
  128. }
  129. async preparePageData() {
  130. await this.processor.retrieveResources(
  131. details => {
  132. if (this.onprogress) {
  133. details.pageURL = this.options.url;
  134. this.onprogress(new ProgressEvent(RESOURCE_LOADING, details));
  135. }
  136. },
  137. details => {
  138. if (this.onprogress) {
  139. details.pageURL = this.options.url;
  140. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  141. }
  142. });
  143. await this.pendingPromises;
  144. if (this.options.lazyLoadImages) {
  145. this.processor.lazyLoadImages();
  146. }
  147. if (!this.options.removeFrames) {
  148. await this.processor.frames();
  149. }
  150. if (!this.options.removeImports) {
  151. await this.processor.htmlImports();
  152. }
  153. }
  154. getPageData() {
  155. if (this.onprogress) {
  156. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  157. }
  158. return this.processor.getPageData();
  159. }
  160. }
  161. // --------
  162. // BatchRequest
  163. // --------
  164. class BatchRequest {
  165. constructor() {
  166. this.requests = new Map();
  167. }
  168. async addURL(resourceURL) {
  169. return new Promise((resolve, reject) => {
  170. const resourceRequests = this.requests.get(resourceURL);
  171. if (resourceRequests) {
  172. resourceRequests.push({ resolve, reject });
  173. } else {
  174. this.requests.set(resourceURL, [{ resolve, reject }]);
  175. }
  176. });
  177. }
  178. getMaxResources() {
  179. return Array.from(this.requests.keys()).length;
  180. }
  181. async run(beforeListener, afterListener, options) {
  182. const resourceURLs = Array.from(this.requests.keys());
  183. let indexResource = 1, indexAfterResource = 1;
  184. return Promise.all(resourceURLs.map(async resourceURL => {
  185. let error;
  186. const resourceRequests = this.requests.get(resourceURL);
  187. beforeListener({ index: indexResource, max: resourceURLs.length, url: resourceURL, error });
  188. indexResource = indexResource + 1;
  189. try {
  190. const dataURI = await Download.getContent(resourceURL, { asDataURI: true, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  191. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  192. } catch (responseError) {
  193. error = responseError;
  194. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  195. }
  196. afterListener({ index: indexAfterResource, max: resourceURLs.length, url: resourceURL, error });
  197. indexAfterResource = indexAfterResource + 1;
  198. this.requests.delete(resourceURL);
  199. }));
  200. }
  201. }
  202. // ------------
  203. // DOMProcessor
  204. // ------------
  205. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  206. const EMPTY_DATA_URI = "data:base64,";
  207. const batchRequest = new BatchRequest();
  208. class DOMProcessor {
  209. constructor(options) {
  210. this.options = options;
  211. this.baseURI = options.url;
  212. }
  213. async loadPage(pageContent) {
  214. if (!pageContent || this.options.saveRawPage) {
  215. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  216. }
  217. this.dom = DOM.create(pageContent, this.baseURI);
  218. this.DOMParser = this.dom.DOMParser;
  219. this.doc = this.dom.document;
  220. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  221. await DOMProcessor.loadEscapedFragmentPage();
  222. }
  223. }
  224. async loadEscapedFragmentPage() {
  225. if (this.baseURI.includes("?")) {
  226. this.baseURI += "&";
  227. } else {
  228. this.baseURI += "?";
  229. }
  230. this.baseURI += ESCAPED_FRAGMENT;
  231. await this.loadPage();
  232. }
  233. async retrieveResources(beforeListener, afterListener) {
  234. await batchRequest.run(beforeListener, afterListener, this.options);
  235. }
  236. getPageData() {
  237. if (this.options.selected) {
  238. const selectedElement = this.doc.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]");
  239. DomProcessorHelper.isolateElement(selectedElement.parentElement, selectedElement);
  240. selectedElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  241. }
  242. const titleElement = this.doc.querySelector("title");
  243. let title;
  244. if (titleElement) {
  245. title = titleElement.textContent.trim();
  246. }
  247. return {
  248. title: title || (this.baseURI ? this.baseURI.match(/([^/]*)\/?$/) : ""),
  249. content: this.dom.serialize()
  250. };
  251. }
  252. insertNoscriptContents() {
  253. if (this.DOMParser) {
  254. this.doc.querySelectorAll("noscript").forEach(element => {
  255. const fragment = this.doc.createDocumentFragment();
  256. Array.from(element.childNodes).forEach(node => {
  257. const parsedNode = new this.DOMParser().parseFromString(node.nodeValue, "text/html");
  258. Array.from(parsedNode.head.childNodes).concat(Array.from(parsedNode.body.childNodes)).forEach(node => {
  259. this.doc.importNode(node);
  260. fragment.appendChild(node);
  261. });
  262. });
  263. element.parentElement.replaceChild(fragment, element);
  264. });
  265. }
  266. }
  267. lazyLoadImages() {
  268. DomProcessorHelper.replaceSrc(this.doc.querySelectorAll("img[data-src]"), "src");
  269. DomProcessorHelper.replaceSrc(this.doc.querySelectorAll("img[data-original]"), "original");
  270. this.doc.querySelectorAll("[data-bg]").forEach(element => {
  271. const dataBg = element.dataset.bg;
  272. if (dataBg && dataBg.startsWith(DATA_URI_PREFIX) && dataBg != EMPTY_DATA_URI && !element.style.backgroundImage.includes(dataBg)) {
  273. element.style.backgroundImage = "url(" + element.dataset.bg + ")";
  274. element.removeAttribute("data-bg");
  275. }
  276. });
  277. this.doc.querySelectorAll("[data-srcset]").forEach(imgElement => {
  278. if (imgElement.dataset.srcset && imgElement.srcset != imgElement.dataset.srcset) {
  279. imgElement.srcset = imgElement.dataset.srcset;
  280. imgElement.removeAttribute("data-srcset");
  281. }
  282. });
  283. }
  284. removeDiscardedResources() {
  285. this.doc.querySelectorAll("applet, meta[http-equiv=refresh], object:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed:not([src*=\".svg\"]), link[rel*=preload], link[rel*=prefetch]").forEach(element => element.remove());
  286. this.doc.querySelectorAll("[onload]").forEach(element => element.removeAttribute("onload"));
  287. this.doc.querySelectorAll("[onerror]").forEach(element => element.removeAttribute("onerror"));
  288. this.doc.querySelectorAll("audio[src], video[src]").forEach(element => element.removeAttribute("src"));
  289. }
  290. removeScripts() {
  291. this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])").forEach(element => element.remove());
  292. }
  293. removeFrames() {
  294. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  295. }
  296. removeImports() {
  297. this.doc.querySelectorAll("link[rel=import]").forEach(element => element.remove());
  298. }
  299. resetCharsetMeta() {
  300. this.doc.querySelectorAll("meta[charset]").forEach(element => element.remove());
  301. const metaElement = this.doc.createElement("meta");
  302. metaElement.setAttribute("charset", "utf-8");
  303. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  304. }
  305. insertFaviconLink() {
  306. let faviconElement = this.doc.querySelectorAll("link[href][rel*=\"icon\"]")[0];
  307. if (!faviconElement) {
  308. faviconElement = this.doc.createElement("link");
  309. faviconElement.setAttribute("type", "image/x-icon");
  310. faviconElement.setAttribute("rel", "shortcut icon");
  311. faviconElement.setAttribute("href", "/favicon.ico");
  312. this.doc.head.appendChild(faviconElement);
  313. }
  314. }
  315. resolveHrefs() {
  316. this.doc.querySelectorAll("[href]").forEach(element => element.setAttribute("href", element.href));
  317. }
  318. removeUnusedCSSRules() {
  319. const doc = this.doc;
  320. doc.querySelectorAll("style").forEach(style => {
  321. const cssRules = [];
  322. if (style.sheet) {
  323. DomProcessorHelper.processRules(this.doc, style.sheet.cssRules, cssRules);
  324. const stylesheetContent = cssRules.join("");
  325. style.textContent = this.options.compress ? this.dom.uglifycss(stylesheetContent) : stylesheetContent;
  326. }
  327. });
  328. }
  329. removeHiddenElements() {
  330. this.doc.querySelectorAll("[" + REMOVED_CONTENT_ATTRIBUTE_NAME + "]").forEach(element => element.remove());
  331. }
  332. compressHTML() {
  333. const textNodesWalker = this.doc.createTreeWalker(this.doc.documentElement, 4, null, false);
  334. let node = textNodesWalker.nextNode();
  335. while (node) {
  336. let element = node.parentElement;
  337. while (element && element.tagName != "PRE") {
  338. element = element.parentElement;
  339. }
  340. if (!element) {
  341. node.textContent = node.textContent.replace(/ +/g, " ");
  342. node.textContent = node.textContent.replace(/\n+/g, " ");
  343. }
  344. node = textNodesWalker.nextNode();
  345. }
  346. const commentNodesWalker = this.doc.createTreeWalker(this.doc.documentElement, 128, null, false);
  347. node = commentNodesWalker.nextNode();
  348. let removedNodes = [];
  349. while (node) {
  350. removedNodes.push(node);
  351. node = commentNodesWalker.nextNode();
  352. }
  353. removedNodes.forEach(node => node.remove());
  354. }
  355. insertSingleFileCommentNode() {
  356. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  357. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  358. }
  359. replaceCanvasElements() {
  360. if (this.options.canvasData) {
  361. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  362. const canvasData = this.options.canvasData[indexCanvasElement];
  363. if (canvasData) {
  364. const imgElement = this.doc.createElement("img");
  365. imgElement.setAttribute("src", canvasData.dataURI);
  366. Array.from(canvasElement.attributes).forEach(attribute => {
  367. if (attribute.value) {
  368. imgElement.setAttribute(attribute.name, attribute.value);
  369. }
  370. });
  371. if (!imgElement.width && canvasData.width) {
  372. imgElement.style.pixelWidth = canvasData.width;
  373. }
  374. if (!imgElement.height && canvasData.height) {
  375. imgElement.style.pixelHeight = canvasData.height;
  376. }
  377. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  378. }
  379. });
  380. }
  381. }
  382. async pageResources() {
  383. const resourcePromises = [
  384. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI),
  385. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"], embed[src*=\".svg\"]"), "src", this.baseURI),
  386. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI),
  387. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI),
  388. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image, use"), "xlink:href", this.baseURI),
  389. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.dom.parseSrcset)
  390. ];
  391. if (this.options.lazyLoadImages) {
  392. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[data-src]"), "data-src", this.baseURI));
  393. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[data-original]"), "data-original", this.baseURI));
  394. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[data-bg]"), "data-bg", this.baseURI));
  395. resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[data-srcset]"), "data-srcset", this.baseURI, this.dom.parseSrcset));
  396. }
  397. await resourcePromises;
  398. }
  399. async inlineStylesheets(initialization) {
  400. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => {
  401. const stylesheetContent = initialization ? await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled }) : await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI);
  402. styleElement.textContent = this.options.compress && !this.options.removeUnusedCSSRules ? this.dom.uglifycss(stylesheetContent) : stylesheetContent;
  403. }));
  404. }
  405. async scripts() {
  406. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  407. if (scriptElement.src) {
  408. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  409. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  410. }
  411. scriptElement.removeAttribute("src");
  412. }));
  413. }
  414. async frames(initialization) {
  415. let frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  416. frameElements = DomUtil.removeNoScriptFrames(frameElements);
  417. await Promise.all(frameElements.map(async (frameElement, frameIndex) => {
  418. const frameWindowId = (this.options.windowId || "0") + "." + frameIndex;
  419. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  420. const setFrameEmptySrc = frameElement => frameElement.setAttribute(frameElement.tagName == "OBJECT" ? "data" : "srcdoc", (frameElement.tagName == "OBJECT" ? "data:text/html," : ""));
  421. if (frameData) {
  422. if (initialization) {
  423. const options = Object.create(this.options);
  424. options.insertSingleFileComment = false;
  425. options.insertFaviconLink = false;
  426. options.url = frameData.baseURI;
  427. options.windowId = frameWindowId;
  428. if (frameData.content) {
  429. options.content = frameData.content;
  430. frameData.processor = new PageProcessor(options);
  431. frameData.frameElement = frameElement;
  432. await frameData.processor.loadPage();
  433. return frameData.processor.initialize();
  434. }
  435. } else {
  436. setFrameEmptySrc(frameElement);
  437. if (frameData.processor) {
  438. const pageData = await frameData.processor.getPageData();
  439. frameElement.setAttribute(frameElement.tagName == "OBJECT" ? "data" : "srcdoc", (frameElement.tagName == "OBJECT" ? "data:text/html," : "") + pageData.content);
  440. frameElement.setAttribute("sandbox", "");
  441. }
  442. }
  443. } else {
  444. setFrameEmptySrc(frameElement);
  445. }
  446. }));
  447. }
  448. async htmlImports(initialization) {
  449. let linkElements = this.doc.querySelectorAll("link[rel=import][href]");
  450. linkElements = DomUtil.removeNoScriptFrames(linkElements);
  451. if (!this.relImportProcessors) {
  452. this.relImportProcessors = new Map();
  453. }
  454. await Promise.all(linkElements.map(async linkElement => {
  455. if (initialization) {
  456. const resourceURL = linkElement.href;
  457. const options = Object.create(this.options);
  458. options.insertSingleFileComment = false;
  459. options.insertFaviconLink = false;
  460. options.url = resourceURL;
  461. if (resourceURL) {
  462. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  463. const processor = new PageProcessor(options);
  464. this.relImportProcessors.set(linkElement, processor);
  465. await processor.loadPage();
  466. return processor.initialize();
  467. }
  468. }
  469. } else {
  470. const processor = this.relImportProcessors.get(linkElement);
  471. if (processor) {
  472. this.relImportProcessors.delete(linkElement);
  473. const pageData = await processor.getPageData();
  474. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  475. } else {
  476. linkElement.setAttribute("href", EMPTY_DATA_URI);
  477. }
  478. }
  479. }));
  480. }
  481. async attributeStyles(initialization) {
  482. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  483. const stylesheetContent = initialization ? await DomProcessorHelper.resolveImportURLs(element.getAttribute("style"), this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled }) : await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI);
  484. element.setAttribute("style", stylesheetContent);
  485. }));
  486. }
  487. async linkStylesheets() {
  488. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  489. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  490. const styleElement = this.doc.createElement("style");
  491. styleElement.textContent = this.options.compress && !this.options.removeUnusedCSSRules ? this.dom.uglifycss(stylesheetContent) : stylesheetContent;
  492. linkElement.parentElement.replaceChild(styleElement, linkElement);
  493. }));
  494. }
  495. }
  496. // ---------
  497. // DomHelper
  498. // ---------
  499. class DomProcessorHelper {
  500. static isolateElement(parentElement, element) {
  501. Array.from(parentElement.childNodes).forEach(node => {
  502. if (node == element) {
  503. node.removeAttribute("style");
  504. node.style.all = "unset";
  505. } else {
  506. if (node.tagName != "HEAD" && node.tagName != "STYLE") {
  507. node.remove();
  508. }
  509. }
  510. });
  511. element = element.parentElement;
  512. if (element.parentElement) {
  513. DomProcessorHelper.isolateElement(element.parentElement, element);
  514. }
  515. }
  516. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  517. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  518. const imports = DomUtil.getImportFunctions(stylesheetContent);
  519. await Promise.all(imports.map(async cssImport => {
  520. const match = DomUtil.matchImport(cssImport);
  521. if (match) {
  522. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  523. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  524. let importedStylesheetContent = await Download.getContent(new URL(match.resourceURL, baseURI).href, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  525. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  526. if (stylesheetContent.indexOf(cssImport) != -1) {
  527. stylesheetContent = stylesheetContent.replace(cssImport, importedStylesheetContent);
  528. }
  529. }
  530. }
  531. }));
  532. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  533. if (imports.length) {
  534. return await DomProcessorHelper.resolveImportURLs(stylesheetContent, baseURI, options);
  535. } else {
  536. return stylesheetContent;
  537. }
  538. }
  539. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  540. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  541. urlFunctions.map(urlFunction => {
  542. let resourceURL = DomUtil.matchURL(urlFunction);
  543. resourceURL = DomUtil.normalizeURL(resourceURL);
  544. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  545. stylesheetContent = stylesheetContent.replace(urlFunction, urlFunction.replace(resourceURL, new URL(resourceURL, baseURI).href));
  546. }
  547. });
  548. return stylesheetContent;
  549. }
  550. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  551. resourceURL = DomUtil.normalizeURL(resourceURL);
  552. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  553. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  554. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  555. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  556. return stylesheetContent;
  557. }
  558. }
  559. static async processStylesheet(stylesheetContent, baseURI) {
  560. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  561. await Promise.all(urlFunctions.map(async urlFunction => {
  562. let resourceURL = DomUtil.matchURL(urlFunction);
  563. resourceURL = DomUtil.normalizeURL(resourceURL);
  564. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  565. const dataURI = await batchRequest.addURL(resourceURL);
  566. stylesheetContent = stylesheetContent.replace(urlFunction, urlFunction.replace(resourceURL, dataURI));
  567. }
  568. }));
  569. return stylesheetContent;
  570. }
  571. static async processAttribute(resourceElements, attributeName, baseURI) {
  572. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  573. let resourceURL = resourceElement.getAttribute(attributeName);
  574. if (resourceURL) {
  575. resourceURL = DomUtil.normalizeURL(resourceURL);
  576. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  577. try {
  578. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  579. resourceElement.setAttribute(attributeName, dataURI);
  580. } catch (e) {
  581. // ignored
  582. }
  583. }
  584. }
  585. }));
  586. }
  587. static replaceSrc(elements, attributeName) {
  588. elements.forEach(element => {
  589. const dataSrc = element.dataset[attributeName];
  590. if (dataSrc && dataSrc.startsWith(DATA_URI_PREFIX) && dataSrc != EMPTY_DATA_URI && element.src != dataSrc) {
  591. element.src = element.dataset[attributeName];
  592. element.removeAttribute("data-" + attributeName);
  593. }
  594. });
  595. }
  596. static async processSrcset(resourceElements, attributeName, baseURI, parseSrcset) {
  597. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  598. const srcset = parseSrcset(resourceElement.getAttribute(attributeName));
  599. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  600. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  601. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  602. try {
  603. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  604. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  605. } catch (e) {
  606. // ignored
  607. }
  608. }
  609. }));
  610. resourceElement.setAttribute(attributeName, srcsetValues.join(","));
  611. }));
  612. }
  613. static processRules(doc, rules, cssRules) {
  614. if (rules) {
  615. Array.from(rules).forEach(rule => {
  616. if (rule.media) {
  617. cssRules.push("@media " + Array.prototype.join.call(rule.media, ",") + " {");
  618. DomProcessorHelper.processRules(doc, rule.cssRules, cssRules);
  619. cssRules.push("}");
  620. } else if (rule.selectorText) {
  621. const selector = DomUtil.getFilteredSelector(rule.selectorText);
  622. if (selector) {
  623. try {
  624. if (doc.querySelector(selector)) {
  625. cssRules.push(rule.cssText);
  626. }
  627. } catch (e) {
  628. cssRules.push(rule.cssText);
  629. }
  630. }
  631. } else {
  632. cssRules.push(rule.cssText);
  633. }
  634. });
  635. }
  636. }
  637. }
  638. // -------
  639. // DomUtil
  640. // -------
  641. const DATA_URI_PREFIX = "data:";
  642. const BLOB_URI_PREFIX = "blob:";
  643. const ABOUT_BLANK_URI = "about:blank";
  644. const REGEXP_URL_FN = /(url\s*\(\s*'([^']*)'\s*\))|(url\s*\(\s*"([^"]*)"\s*\))|(url\s*\(\s*([^)]*)\s*\))/gi;
  645. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'([^']*)'\s*\)$/i;
  646. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"([^"]*)"\s*\)$/i;
  647. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*([^)]*)\s*\)$/i;
  648. 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;
  649. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'([^']*)'\s*\)\s*([^;]*)/i;
  650. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"([^"]*)"\s*\)\s*([^;]*)/i;
  651. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*([^)]*)\s*\)\s*([^;]*)/i;
  652. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'([^']*)'\s*([^;]*)/i;
  653. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"([^"]*)"\s*([^;]*)/i;
  654. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*([^;]*)\s*([^;]*)/i;
  655. const REGEXP_PSEUDO_CLASSES = /::after|::before|::first-line|::first-letter|:focus|:focus-within|:hover|:link|:visited|:active/gi;
  656. class DomUtil {
  657. static normalizeURL(url) {
  658. return url.split("#")[0];
  659. }
  660. static getUrlFunctions(stylesheetContent) {
  661. return stylesheetContent.match(REGEXP_URL_FN) || [];
  662. }
  663. static getImportFunctions(stylesheetContent) {
  664. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  665. }
  666. static matchURL(stylesheetContent) {
  667. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  668. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  669. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  670. return match && match[1];
  671. }
  672. static testValidPath(resourceURL) {
  673. return !resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(BLOB_URI_PREFIX) && resourceURL != ABOUT_BLANK_URI;
  674. }
  675. static matchImport(stylesheetContent) {
  676. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  677. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  678. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  679. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  680. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  681. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  682. if (match) {
  683. const [, resourceURL, media] = match;
  684. return { resourceURL, media };
  685. }
  686. }
  687. static removeCssComments(stylesheetContent) {
  688. let start, end;
  689. do {
  690. start = stylesheetContent.indexOf("/*");
  691. end = stylesheetContent.indexOf("*/", start);
  692. if (start != -1 && end != -1) {
  693. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  694. }
  695. } while (start != -1 && end != -1);
  696. return stylesheetContent;
  697. }
  698. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  699. if (mediaQuery) {
  700. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  701. } else {
  702. return stylesheetContent;
  703. }
  704. }
  705. static removeNoScriptFrames(frameElements) {
  706. return Array.from(frameElements).filter(element => {
  707. element = element.parentElement;
  708. while (element && element.tagName != "NOSCRIPT") {
  709. element = element.parentElement;
  710. }
  711. return !element;
  712. });
  713. }
  714. static getFilteredSelector(selector) {
  715. if (selector.match(REGEXP_PSEUDO_CLASSES)) {
  716. let selectors = selector.split(/\s*,\s*/g);
  717. selector = selectors.map(selector => {
  718. const simpleSelectors = selector.split(/\s*[ >~+]\s*/g);
  719. const separators = selector.match(/\s*[ >~+]\s*/g);
  720. return simpleSelectors.map((selector, selectorIndex) => {
  721. while (selector.match(REGEXP_PSEUDO_CLASSES)) {
  722. selector = selector.replace(REGEXP_PSEUDO_CLASSES, "").trim();
  723. }
  724. selector = selector.replace(/:?:[^(]+\(\)/g, "");
  725. if (selector == "") {
  726. selector = "*";
  727. }
  728. return selector + (separators && separators[selectorIndex] ? separators[selectorIndex] : "");
  729. }).join("");
  730. }).join(",");
  731. }
  732. return selector;
  733. }
  734. }
  735. return SingleFileCore;
  736. })();