single-file-core.js 31 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793
  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. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  71. await this.processor.loadPage(this.options.content);
  72. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  73. }
  74. async initialize() {
  75. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  76. if (!this.options.jsEnabled || (this.options.saveRawPage && this.options.removeScripts)) {
  77. this.processor.insertNoscriptContents();
  78. }
  79. if (this.options.removeFrames) {
  80. this.processor.removeFrames();
  81. }
  82. if (this.options.removeImports) {
  83. this.processor.removeImports();
  84. }
  85. if (this.options.removeScripts) {
  86. this.processor.removeScripts();
  87. }
  88. this.processor.removeDiscardedResources();
  89. this.processor.resetCharsetMeta();
  90. if (this.options.compress) {
  91. this.processor.compressHTML();
  92. }
  93. if (this.options.insertFaviconLink) {
  94. this.processor.insertFaviconLink();
  95. }
  96. this.processor.resolveHrefs();
  97. if (this.options.insertSingleFileComment) {
  98. this.processor.insertSingleFileCommentNode();
  99. }
  100. this.processor.replaceCanvasElements();
  101. if (this.options.removeHiddenElements) {
  102. this.processor.removeHiddenElements();
  103. }
  104. const initializationPromises = [this.processor.inlineStylesheets(true), this.processor.linkStylesheets(), this.processor.attributeStyles(true)];
  105. if (!this.options.removeFrames) {
  106. initializationPromises.push(this.processor.frames(true));
  107. }
  108. if (!this.options.removeImports) {
  109. initializationPromises.push(this.processor.htmlImports(true));
  110. }
  111. await Promise.all(initializationPromises);
  112. if (this.options.removeUnusedCSSRules) {
  113. this.processor.removeUnusedCSSRules();
  114. }
  115. this.pendingPromises = [this.processor.inlineStylesheets(), this.processor.attributeStyles(), this.processor.pageResources()];
  116. if (!this.options.removeScripts) {
  117. this.pendingPromises.push(this.processor.scripts());
  118. }
  119. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: batchRequest.getMaxResources() }));
  120. }
  121. async preparePageData() {
  122. await this.processor.retrieveResources(
  123. details => {
  124. details.pageURL = this.options.url;
  125. this.onprogress(new ProgressEvent(RESOURCE_LOADING, details));
  126. },
  127. details => {
  128. details.pageURL = this.options.url;
  129. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  130. });
  131. await this.pendingPromises;
  132. if (this.options.lazyLoadImages) {
  133. this.processor.lazyLoadImages();
  134. }
  135. if (!this.options.removeFrames) {
  136. await this.processor.frames();
  137. }
  138. if (!this.options.removeImports) {
  139. await this.processor.htmlImports();
  140. }
  141. }
  142. getPageData() {
  143. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  144. return this.processor.getPageData();
  145. }
  146. }
  147. // --------
  148. // BatchRequest
  149. // --------
  150. class BatchRequest {
  151. constructor() {
  152. this.requests = new Map();
  153. }
  154. async addURL(resourceURL) {
  155. return new Promise((resolve, reject) => {
  156. const resourceRequests = this.requests.get(resourceURL);
  157. if (resourceRequests) {
  158. resourceRequests.push({ resolve, reject });
  159. } else {
  160. this.requests.set(resourceURL, [{ resolve, reject }]);
  161. }
  162. });
  163. }
  164. getMaxResources() {
  165. return Array.from(this.requests.keys()).length;
  166. }
  167. async run(beforeListener, afterListener, options) {
  168. const resourceURLs = Array.from(this.requests.keys());
  169. let indexResource = 1, indexAfterResource = 1;
  170. return Promise.all(resourceURLs.map(async resourceURL => {
  171. let error;
  172. const resourceRequests = this.requests.get(resourceURL);
  173. beforeListener({ index: indexResource, max: resourceURLs.length, url: resourceURL, error });
  174. indexResource = indexResource + 1;
  175. try {
  176. const dataURI = await Download.getContent(resourceURL, { asDataURI: true, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  177. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  178. } catch (responseError) {
  179. error = responseError;
  180. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  181. }
  182. afterListener({ index: indexAfterResource, max: resourceURLs.length, url: resourceURL, error });
  183. indexAfterResource = indexAfterResource + 1;
  184. this.requests.delete(resourceURL);
  185. }));
  186. }
  187. }
  188. // ------------
  189. // DOMProcessor
  190. // ------------
  191. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  192. const EMPTY_DATA_URI = "data:base64,";
  193. const batchRequest = new BatchRequest();
  194. class DOMProcessor {
  195. constructor(options) {
  196. this.options = options;
  197. this.baseURI = options.url;
  198. }
  199. async loadPage(pageContent) {
  200. if (!pageContent || this.options.saveRawPage) {
  201. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  202. }
  203. this.dom = DOM.create(pageContent, this.baseURI);
  204. this.DOMParser = this.dom.DOMParser;
  205. this.doc = this.dom.document;
  206. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  207. await DOMProcessor.loadEscapedFragmentPage();
  208. }
  209. }
  210. async loadEscapedFragmentPage() {
  211. if (this.baseURI.includes("?")) {
  212. this.baseURI += "&";
  213. } else {
  214. this.baseURI += "?";
  215. }
  216. this.baseURI += ESCAPED_FRAGMENT;
  217. await this.loadPage();
  218. }
  219. async retrieveResources(beforeListener, afterListener) {
  220. await batchRequest.run(beforeListener, afterListener, this.options);
  221. }
  222. getPageData() {
  223. if (this.options.selected) {
  224. const selectedElement = this.doc.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]");
  225. if (selectedElement) {
  226. DomProcessorHelper.isolateElement(selectedElement.parentElement, selectedElement);
  227. selectedElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  228. }
  229. }
  230. const titleElement = this.doc.querySelector("title");
  231. let title;
  232. if (titleElement) {
  233. title = titleElement.textContent.trim();
  234. }
  235. return {
  236. title: title || (this.baseURI ? this.baseURI.match(/([^/]*)\/?$/) : ""),
  237. content: this.dom.serialize()
  238. };
  239. }
  240. insertNoscriptContents() {
  241. if (this.DOMParser) {
  242. this.doc.querySelectorAll("noscript").forEach(element => {
  243. const fragment = this.doc.createDocumentFragment();
  244. Array.from(element.childNodes).forEach(node => {
  245. const parsedNode = new this.DOMParser().parseFromString(node.nodeValue, "text/html");
  246. Array.from(parsedNode.head.childNodes).concat(Array.from(parsedNode.body.childNodes)).forEach(node => {
  247. this.doc.importNode(node);
  248. fragment.appendChild(node);
  249. });
  250. });
  251. element.parentElement.replaceChild(fragment, element);
  252. });
  253. }
  254. }
  255. lazyLoadImages() {
  256. DomProcessorHelper.replaceSrc(this.doc.querySelectorAll("img[data-src]"), "src");
  257. DomProcessorHelper.replaceSrc(this.doc.querySelectorAll("img[data-original]"), "original");
  258. this.doc.querySelectorAll("[data-bg]").forEach(element => {
  259. const dataBg = element.dataset.bg;
  260. if (dataBg && dataBg.startsWith(DATA_URI_PREFIX) && dataBg != EMPTY_DATA_URI && !element.style.backgroundImage.includes(dataBg)) {
  261. element.style.backgroundImage = "url(" + element.dataset.bg + ")";
  262. element.removeAttribute("data-bg");
  263. element.removeAttribute("data-lazyload");
  264. }
  265. });
  266. this.doc.querySelectorAll("[data-srcset]").forEach(imgElement => {
  267. const srcset = imgElement.dataset.srcset;
  268. if (srcset && imgElement.srcset != srcset) {
  269. imgElement.srcset = srcset;
  270. imgElement.removeAttribute("data-srcset");
  271. imgElement.removeAttribute("data-lazyload");
  272. }
  273. });
  274. }
  275. removeDiscardedResources() {
  276. 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());
  277. this.doc.querySelectorAll("[onload]").forEach(element => element.removeAttribute("onload"));
  278. this.doc.querySelectorAll("[onerror]").forEach(element => element.removeAttribute("onerror"));
  279. this.doc.querySelectorAll("audio[src], audio > source[src], video[src], video > source[src]").forEach(element => element.removeAttribute("src"));
  280. }
  281. removeScripts() {
  282. this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])").forEach(element => element.remove());
  283. }
  284. removeFrames() {
  285. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  286. }
  287. removeImports() {
  288. this.doc.querySelectorAll("link[rel=import]").forEach(element => element.remove());
  289. }
  290. resetCharsetMeta() {
  291. this.doc.querySelectorAll("meta[charset]").forEach(element => element.remove());
  292. const metaElement = this.doc.createElement("meta");
  293. metaElement.setAttribute("charset", "utf-8");
  294. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  295. }
  296. insertFaviconLink() {
  297. let faviconElement = this.doc.querySelectorAll("link[href][rel*=\"icon\"]")[0];
  298. if (!faviconElement) {
  299. faviconElement = this.doc.createElement("link");
  300. faviconElement.setAttribute("type", "image/x-icon");
  301. faviconElement.setAttribute("rel", "shortcut icon");
  302. faviconElement.setAttribute("href", "/favicon.ico");
  303. this.doc.head.appendChild(faviconElement);
  304. }
  305. }
  306. resolveHrefs() {
  307. this.doc.querySelectorAll("[href]").forEach(element => element.setAttribute("href", element.href));
  308. }
  309. removeUnusedCSSRules() {
  310. const doc = this.doc;
  311. doc.querySelectorAll("style").forEach(style => {
  312. const cssRules = [];
  313. if (style.sheet) {
  314. DomProcessorHelper.processRules(this.doc, style.sheet.cssRules, cssRules);
  315. const stylesheetContent = cssRules.join("");
  316. style.textContent = this.options.compress ? this.dom.uglifycss(stylesheetContent) : stylesheetContent;
  317. }
  318. });
  319. }
  320. removeHiddenElements() {
  321. this.doc.querySelectorAll("[" + REMOVED_CONTENT_ATTRIBUTE_NAME + "]").forEach(element => element.remove());
  322. }
  323. compressHTML() {
  324. const textNodesWalker = this.doc.createTreeWalker(this.doc.documentElement, 4, null, false);
  325. let node = textNodesWalker.nextNode();
  326. while (node) {
  327. let element = node.parentElement;
  328. while (element && element.tagName != "PRE") {
  329. element = element.parentElement;
  330. }
  331. if (!element) {
  332. node.textContent = node.textContent.replace(/ +/g, " ");
  333. node.textContent = node.textContent.replace(/\n+/g, " ");
  334. }
  335. node = textNodesWalker.nextNode();
  336. }
  337. const commentNodesWalker = this.doc.createTreeWalker(this.doc.documentElement, 128, null, false);
  338. node = commentNodesWalker.nextNode();
  339. let removedNodes = [];
  340. while (node) {
  341. removedNodes.push(node);
  342. node = commentNodesWalker.nextNode();
  343. }
  344. removedNodes.forEach(node => node.remove());
  345. }
  346. insertSingleFileCommentNode() {
  347. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  348. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  349. }
  350. replaceCanvasElements() {
  351. if (this.options.canvasData) {
  352. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  353. const canvasData = this.options.canvasData[indexCanvasElement];
  354. if (canvasData) {
  355. const imgElement = this.doc.createElement("img");
  356. imgElement.setAttribute("src", canvasData.dataURI);
  357. Array.from(canvasElement.attributes).forEach(attribute => {
  358. if (attribute.value) {
  359. imgElement.setAttribute(attribute.name, attribute.value);
  360. }
  361. });
  362. if (!imgElement.width && canvasData.width) {
  363. imgElement.style.pixelWidth = canvasData.width;
  364. }
  365. if (!imgElement.height && canvasData.height) {
  366. imgElement.style.pixelHeight = canvasData.height;
  367. }
  368. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  369. }
  370. });
  371. }
  372. }
  373. async pageResources() {
  374. const resourcePromises = [
  375. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI),
  376. 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),
  377. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI),
  378. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI),
  379. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image, use"), "xlink:href", this.baseURI),
  380. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.dom.parseSrcset)
  381. ];
  382. if (this.options.lazyLoadImages) {
  383. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[data-src]"), "data-src", this.baseURI));
  384. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[data-original]"), "data-original", this.baseURI));
  385. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[data-bg]"), "data-bg", this.baseURI));
  386. resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[data-srcset]"), "data-srcset", this.baseURI, this.dom.parseSrcset));
  387. }
  388. await resourcePromises;
  389. }
  390. async inlineStylesheets(initialization) {
  391. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => {
  392. 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);
  393. styleElement.textContent = this.options.compress && !this.options.removeUnusedCSSRules ? this.dom.uglifycss(stylesheetContent) : stylesheetContent;
  394. }));
  395. }
  396. async scripts() {
  397. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  398. if (scriptElement.src) {
  399. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  400. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  401. }
  402. scriptElement.removeAttribute("src");
  403. }));
  404. }
  405. async frames(initialization) {
  406. let frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  407. frameElements = DomUtil.removeNoScriptFrames(frameElements);
  408. await Promise.all(frameElements.map(async (frameElement, frameIndex) => {
  409. const frameWindowId = (this.options.windowId || "0") + "." + frameIndex;
  410. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  411. const setFrameEmptySrc = frameElement => frameElement.setAttribute(frameElement.tagName == "OBJECT" ? "data" : "srcdoc", (frameElement.tagName == "OBJECT" ? "data:text/html," : ""));
  412. if (frameData) {
  413. if (initialization) {
  414. const options = Object.create(this.options);
  415. options.insertSingleFileComment = false;
  416. options.insertFaviconLink = false;
  417. options.url = frameData.baseURI;
  418. options.windowId = frameWindowId;
  419. if (frameData.content) {
  420. options.content = frameData.content;
  421. frameData.processor = new PageProcessor(options);
  422. frameData.frameElement = frameElement;
  423. await frameData.processor.loadPage();
  424. return frameData.processor.initialize();
  425. }
  426. } else {
  427. setFrameEmptySrc(frameElement);
  428. if (frameData.processor) {
  429. const pageData = await frameData.processor.getPageData();
  430. frameElement.setAttribute(frameElement.tagName == "OBJECT" ? "data" : "srcdoc", (frameElement.tagName == "OBJECT" ? "data:text/html," : "") + pageData.content);
  431. frameElement.setAttribute("sandbox", "");
  432. }
  433. }
  434. } else {
  435. setFrameEmptySrc(frameElement);
  436. }
  437. }));
  438. }
  439. async htmlImports(initialization) {
  440. let linkElements = this.doc.querySelectorAll("link[rel=import][href]");
  441. linkElements = DomUtil.removeNoScriptFrames(linkElements);
  442. if (!this.relImportProcessors) {
  443. this.relImportProcessors = new Map();
  444. }
  445. await Promise.all(linkElements.map(async linkElement => {
  446. if (initialization) {
  447. const resourceURL = linkElement.href;
  448. const options = Object.create(this.options);
  449. options.insertSingleFileComment = false;
  450. options.insertFaviconLink = false;
  451. options.url = resourceURL;
  452. if (resourceURL) {
  453. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  454. const processor = new PageProcessor(options);
  455. this.relImportProcessors.set(linkElement, processor);
  456. await processor.loadPage();
  457. return processor.initialize();
  458. }
  459. }
  460. } else {
  461. const processor = this.relImportProcessors.get(linkElement);
  462. if (processor) {
  463. this.relImportProcessors.delete(linkElement);
  464. const pageData = await processor.getPageData();
  465. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  466. } else {
  467. linkElement.setAttribute("href", EMPTY_DATA_URI);
  468. }
  469. }
  470. }));
  471. }
  472. async attributeStyles(initialization) {
  473. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  474. 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);
  475. element.setAttribute("style", stylesheetContent);
  476. }));
  477. }
  478. async linkStylesheets() {
  479. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  480. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  481. const styleElement = this.doc.createElement("style");
  482. styleElement.textContent = this.options.compress && !this.options.removeUnusedCSSRules ? this.dom.uglifycss(stylesheetContent) : stylesheetContent;
  483. linkElement.parentElement.replaceChild(styleElement, linkElement);
  484. }));
  485. }
  486. }
  487. // ---------
  488. // DomHelper
  489. // ---------
  490. class DomProcessorHelper {
  491. static isolateElement(parentElement, element) {
  492. Array.from(parentElement.childNodes).forEach(node => {
  493. if (node == element) {
  494. node.removeAttribute("style");
  495. node.style.all = "unset";
  496. } else {
  497. if (node.tagName != "HEAD" && node.tagName != "STYLE") {
  498. node.remove();
  499. }
  500. }
  501. });
  502. element = element.parentElement;
  503. if (element.parentElement) {
  504. DomProcessorHelper.isolateElement(element.parentElement, element);
  505. }
  506. }
  507. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  508. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  509. const imports = DomUtil.getImportFunctions(stylesheetContent);
  510. await Promise.all(imports.map(async cssImport => {
  511. const match = DomUtil.matchImport(cssImport);
  512. if (match) {
  513. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  514. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  515. let importedStylesheetContent = await Download.getContent(new URL(match.resourceURL, baseURI).href, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  516. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  517. if (stylesheetContent.indexOf(cssImport) != -1) {
  518. stylesheetContent = stylesheetContent.replace(cssImport, importedStylesheetContent);
  519. }
  520. }
  521. }
  522. }));
  523. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  524. if (imports.length) {
  525. return await DomProcessorHelper.resolveImportURLs(stylesheetContent, baseURI, options);
  526. } else {
  527. return stylesheetContent;
  528. }
  529. }
  530. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  531. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  532. urlFunctions.map(urlFunction => {
  533. let resourceURL = DomUtil.matchURL(urlFunction);
  534. resourceURL = DomUtil.normalizeURL(resourceURL);
  535. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  536. stylesheetContent = stylesheetContent.replace(urlFunction, urlFunction.replace(resourceURL, new URL(resourceURL, baseURI).href));
  537. }
  538. });
  539. return stylesheetContent;
  540. }
  541. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  542. resourceURL = DomUtil.normalizeURL(resourceURL);
  543. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  544. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  545. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  546. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  547. return stylesheetContent;
  548. }
  549. }
  550. static async processStylesheet(stylesheetContent, baseURI) {
  551. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  552. await Promise.all(urlFunctions.map(async urlFunction => {
  553. let resourceURL = DomUtil.matchURL(urlFunction);
  554. resourceURL = DomUtil.normalizeURL(resourceURL);
  555. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  556. const dataURI = await batchRequest.addURL(resourceURL);
  557. stylesheetContent = stylesheetContent.replace(urlFunction, urlFunction.replace(resourceURL, dataURI));
  558. }
  559. }));
  560. return stylesheetContent;
  561. }
  562. static async processAttribute(resourceElements, attributeName, baseURI) {
  563. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  564. let resourceURL = resourceElement.getAttribute(attributeName);
  565. if (resourceURL) {
  566. resourceURL = DomUtil.normalizeURL(resourceURL);
  567. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  568. try {
  569. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  570. resourceElement.setAttribute(attributeName, dataURI);
  571. } catch (e) {
  572. // ignored
  573. }
  574. }
  575. }
  576. }));
  577. }
  578. static replaceSrc(elements, attributeName) {
  579. elements.forEach(element => {
  580. const dataSrc = element.dataset[attributeName];
  581. if (dataSrc && dataSrc.startsWith(DATA_URI_PREFIX) && dataSrc != EMPTY_DATA_URI && element.src != dataSrc) {
  582. element.src = element.dataset[attributeName];
  583. element.removeAttribute("data-" + attributeName);
  584. element.removeAttribute("data-lazyload");
  585. }
  586. });
  587. }
  588. static async processSrcset(resourceElements, attributeName, baseURI, parseSrcset) {
  589. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  590. const srcset = parseSrcset(resourceElement.getAttribute(attributeName));
  591. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  592. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  593. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  594. try {
  595. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  596. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  597. } catch (e) {
  598. // ignored
  599. }
  600. }
  601. }));
  602. resourceElement.setAttribute(attributeName, srcsetValues.join(","));
  603. }));
  604. }
  605. static processRules(doc, rules, cssRules) {
  606. if (rules) {
  607. Array.from(rules).forEach(rule => {
  608. if (rule.media) {
  609. cssRules.push("@media " + Array.prototype.join.call(rule.media, ",") + " {");
  610. DomProcessorHelper.processRules(doc, rule.cssRules, cssRules);
  611. cssRules.push("}");
  612. } else if (rule.selectorText) {
  613. const selector = DomUtil.getFilteredSelector(rule.selectorText);
  614. if (selector) {
  615. try {
  616. if (doc.querySelector(selector)) {
  617. cssRules.push(rule.cssText);
  618. }
  619. } catch (e) {
  620. cssRules.push(rule.cssText);
  621. }
  622. }
  623. } else {
  624. cssRules.push(rule.cssText);
  625. }
  626. });
  627. }
  628. }
  629. }
  630. // -------
  631. // DomUtil
  632. // -------
  633. const DATA_URI_PREFIX = "data:";
  634. const BLOB_URI_PREFIX = "blob:";
  635. const ABOUT_BLANK_URI = "about:blank";
  636. const REGEXP_URL_FN = /(url\s*\(\s*'([^']*)'\s*\))|(url\s*\(\s*"([^"]*)"\s*\))|(url\s*\(\s*([^)]*)\s*\))/gi;
  637. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'([^']*)'\s*\)$/i;
  638. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"([^"]*)"\s*\)$/i;
  639. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*([^)]*)\s*\)$/i;
  640. 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;
  641. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'([^']*)'\s*\)\s*([^;]*)/i;
  642. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"([^"]*)"\s*\)\s*([^;]*)/i;
  643. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*([^)]*)\s*\)\s*([^;]*)/i;
  644. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'([^']*)'\s*([^;]*)/i;
  645. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"([^"]*)"\s*([^;]*)/i;
  646. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*([^;]*)\s*([^;]*)/i;
  647. const REGEXP_PSEUDO_CLASSES = /::after|::before|::first-line|::first-letter|:focus|:focus-within|:hover|:link|:visited|:active/gi;
  648. class DomUtil {
  649. static normalizeURL(url) {
  650. return url.split("#")[0];
  651. }
  652. static getUrlFunctions(stylesheetContent) {
  653. return stylesheetContent.match(REGEXP_URL_FN) || [];
  654. }
  655. static getImportFunctions(stylesheetContent) {
  656. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  657. }
  658. static matchURL(stylesheetContent) {
  659. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  660. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  661. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  662. return match && match[1];
  663. }
  664. static testValidPath(resourceURL) {
  665. return !resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(BLOB_URI_PREFIX) && resourceURL != ABOUT_BLANK_URI;
  666. }
  667. static matchImport(stylesheetContent) {
  668. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  669. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  670. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  671. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  672. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  673. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  674. if (match) {
  675. const [, resourceURL, media] = match;
  676. return { resourceURL, media };
  677. }
  678. }
  679. static removeCssComments(stylesheetContent) {
  680. let start, end;
  681. do {
  682. start = stylesheetContent.indexOf("/*");
  683. end = stylesheetContent.indexOf("*/", start);
  684. if (start != -1 && end != -1) {
  685. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  686. }
  687. } while (start != -1 && end != -1);
  688. return stylesheetContent;
  689. }
  690. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  691. if (mediaQuery) {
  692. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  693. } else {
  694. return stylesheetContent;
  695. }
  696. }
  697. static removeNoScriptFrames(frameElements) {
  698. return Array.from(frameElements).filter(element => {
  699. element = element.parentElement;
  700. while (element && element.tagName != "NOSCRIPT") {
  701. element = element.parentElement;
  702. }
  703. return !element;
  704. });
  705. }
  706. static getFilteredSelector(selector) {
  707. if (selector.match(REGEXP_PSEUDO_CLASSES)) {
  708. let selectors = selector.split(/\s*,\s*/g);
  709. selector = selectors.map(selector => {
  710. const simpleSelectors = selector.split(/\s*[ >~+]\s*/g);
  711. const separators = selector.match(/\s*[ >~+]\s*/g);
  712. return simpleSelectors.map((selector, selectorIndex) => {
  713. while (selector.match(REGEXP_PSEUDO_CLASSES)) {
  714. selector = selector.replace(REGEXP_PSEUDO_CLASSES, "").trim();
  715. }
  716. selector = selector.replace(/:?:[^(]+\(\)/g, "");
  717. if (selector == "") {
  718. selector = "*";
  719. }
  720. return selector + (separators && separators[selectorIndex] ? separators[selectorIndex] : "");
  721. }).join("");
  722. }).join(",");
  723. }
  724. return selector;
  725. }
  726. }
  727. return SingleFileCore;
  728. })();