1
0

single-file-core.js 52 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259
  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 = this.SingleFileCore || (() => {
  21. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  22. const SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = "data-single-file-selected-content-root";
  23. let Download, DOM, URL, sessionId = 0;
  24. function getClass(...args) {
  25. [Download, DOM, URL] = args;
  26. return class {
  27. constructor(options) {
  28. this.options = options;
  29. options.sessionId = sessionId;
  30. sessionId++;
  31. this.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  32. this.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  33. }
  34. async initialize() {
  35. this.processor = new PageProcessor(this.options);
  36. await this.processor.loadPage();
  37. await this.processor.initialize();
  38. }
  39. async preparePageData() {
  40. await this.processor.preparePageData();
  41. }
  42. async getPageData() {
  43. return this.processor.getPageData();
  44. }
  45. };
  46. }
  47. // -------------
  48. // ProgressEvent
  49. // -------------
  50. const PAGE_LOADING = "page-loading";
  51. const PAGE_LOADED = "page-loaded";
  52. const RESOURCES_INITIALIZING = "resource-initializing";
  53. const RESOURCES_INITIALIZED = "resources-initialized";
  54. const RESOURCE_LOADED = "resource-loaded";
  55. const PAGE_ENDED = "page-ended";
  56. class ProgressEvent {
  57. constructor(type, details) {
  58. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  59. }
  60. }
  61. // -------------
  62. // PageProcessor
  63. // -------------
  64. const STAGES = [{
  65. sync: [
  66. { action: "removeUIElements" },
  67. { action: "replaceStyleContents" },
  68. { option: "removeSrcSet", action: "removeSrcSet" },
  69. { option: "removeFrames", action: "removeFrames" },
  70. { option: "removeImports", action: "removeImports" },
  71. { option: "removeScripts", action: "removeScripts" },
  72. { action: "removeDiscardedResources" },
  73. { action: "resetCharsetMeta" },
  74. { action: "setInputValues" },
  75. { option: "compressHTML", action: "compressHTML" },
  76. { option: "insertFaviconLink", action: "insertFaviconLink" },
  77. { action: "resolveHrefs" },
  78. { action: "replaceCanvasElements" },
  79. { action: "insertFonts" },
  80. { option: "removeHiddenElements", action: "removeHiddenElements" }
  81. ],
  82. async: [
  83. { action: "inlineStylesheets" },
  84. { action: "linkStylesheets" },
  85. { action: "attributeStyles" },
  86. { option: "!removeFrames", action: "frames" },
  87. { option: "!removeImports", action: "htmlImports" }
  88. ]
  89. }, {
  90. sync: [
  91. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  92. { option: "compressHTML", action: "compressHTML" },
  93. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  94. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  95. ],
  96. async: [
  97. { action: "inlineStylesheets" },
  98. { action: "attributeStyles" },
  99. { action: "pageResources" },
  100. { option: "!removeScripts", action: "scripts" }
  101. ]
  102. }, {
  103. sync: [
  104. { option: "lazyLoadImages", action: "lazyLoadImages" },
  105. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  106. { option: "compressCSS", action: "compressCSS" }
  107. ],
  108. async: [
  109. { option: "!removeFrames", action: "frames" },
  110. { option: "!removeImports", action: "htmlImports" },
  111. ]
  112. }, {
  113. sync: [
  114. { option: "compressHTML", action: "postCompressHTML" },
  115. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  116. { action: "removeDefaultHeadTags" }
  117. ]
  118. }];
  119. class PageProcessor {
  120. constructor(options) {
  121. this.options = options;
  122. this.options.url = this.options.url || this.options.doc.location.href;
  123. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  124. this.batchRequest = new BatchRequest();
  125. this.processor = new DOMProcessor(options, this.batchRequest);
  126. if (this.options.doc) {
  127. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  128. this.options.canvasData = docData.canvasData;
  129. this.options.fontsData = docData.fontsData;
  130. this.options.stylesheetContents = docData.stylesheetContents;
  131. this.options.imageData = docData.imageData;
  132. }
  133. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  134. this.onprogress = options.onprogress || (() => { });
  135. }
  136. async loadPage() {
  137. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  138. await this.processor.loadPage(this.options.content);
  139. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  140. }
  141. async initialize() {
  142. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  143. await this.executeStage(0);
  144. this.pendingPromises = this.executeStage(1);
  145. if (this.options.doc) {
  146. DOM.postProcessDoc(this.options.doc, this.options);
  147. this.options.doc = null;
  148. this.options.win = null;
  149. }
  150. }
  151. async preparePageData() {
  152. if (!this.options.windowId) {
  153. this.processor.initialize(this.batchRequest);
  154. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  155. }
  156. await this.batchRequest.run(details => {
  157. details.pageURL = this.options.url;
  158. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  159. }, this.options);
  160. await this.pendingPromises;
  161. await this.executeStage(2);
  162. await this.executeStage(3);
  163. }
  164. async getPageData() {
  165. if (!this.options.windowId) {
  166. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  167. }
  168. return this.processor.getPageData();
  169. }
  170. async executeStage(step) {
  171. STAGES[step].sync.forEach(task => {
  172. // const startTime = Date.now();
  173. // console.log(task.action, "...");
  174. this.executeTask(task, !step);
  175. // console.log(task.action, Date.now() - startTime);
  176. });
  177. if (STAGES[step].async) {
  178. return Promise.all(STAGES[step].async.map(task => {
  179. // const startTime = Date.now();
  180. // console.log(task.action, "...");
  181. const promise = this.executeTask(task, !step);
  182. // promise.then(() => console.log(task.action, Date.now() - startTime));
  183. return promise;
  184. }));
  185. }
  186. }
  187. executeTask(task, initialization) {
  188. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  189. return this.processor[task.action](initialization);
  190. }
  191. }
  192. }
  193. // --------
  194. // BatchRequest
  195. // --------
  196. class BatchRequest {
  197. constructor() {
  198. this.requests = new Map();
  199. }
  200. async addURL(resourceURL, asDataURI = true) {
  201. return new Promise((resolve, reject) => {
  202. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  203. const resourceRequests = this.requests.get(requestKey);
  204. if (resourceRequests) {
  205. resourceRequests.push({ resolve, reject });
  206. } else {
  207. this.requests.set(requestKey, [{ resolve, reject }]);
  208. }
  209. });
  210. }
  211. getMaxResources() {
  212. return Array.from(this.requests.keys()).length;
  213. }
  214. async run(onloadListener, options) {
  215. const resourceURLs = Array.from(this.requests.keys());
  216. let indexResource = 0;
  217. return Promise.all(resourceURLs.map(async requestKey => {
  218. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  219. const resourceRequests = this.requests.get(requestKey);
  220. try {
  221. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  222. indexResource = indexResource + 1;
  223. onloadListener({ index: indexResource, url: resourceURL });
  224. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  225. } catch (error) {
  226. indexResource = indexResource + 1;
  227. onloadListener({ index: indexResource, url: resourceURL });
  228. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  229. }
  230. this.requests.delete(requestKey);
  231. }));
  232. }
  233. }
  234. // ------------
  235. // DOMProcessor
  236. // ------------
  237. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  238. const EMPTY_DATA_URI = "data:base64,";
  239. class DOMProcessor {
  240. constructor(options, batchRequest) {
  241. this.options = options;
  242. this.stats = new Stats(options);
  243. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  244. this.batchRequest = batchRequest;
  245. }
  246. initialize() {
  247. this.maxResources = this.batchRequest.getMaxResources();
  248. if (!this.options.removeFrames) {
  249. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  250. }
  251. this.stats.set("processed", "resources", this.maxResources);
  252. }
  253. async loadPage(pageContent) {
  254. if (!pageContent || this.options.saveRawPage) {
  255. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  256. }
  257. this.doc = DOM.createDoc(pageContent, this.baseURI);
  258. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  259. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  260. await DOMProcessor.loadEscapedFragmentPage();
  261. }
  262. }
  263. async loadEscapedFragmentPage() {
  264. if (this.baseURI.includes("?")) {
  265. this.baseURI += "&";
  266. } else {
  267. this.baseURI += "?";
  268. }
  269. this.baseURI += ESCAPED_FRAGMENT;
  270. await this.loadPage();
  271. }
  272. async getPageData() {
  273. DOM.postProcessDoc(this.doc, this.options);
  274. if (this.options.selected) {
  275. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  276. if (rootElement) {
  277. DomProcessorHelper.isolateElements(rootElement);
  278. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  279. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  280. }
  281. }
  282. const titleElement = this.doc.querySelector("title");
  283. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  284. this.options.info = {};
  285. const descriptionElement = this.doc.querySelector("meta[name=description]");
  286. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  287. this.options.info.lang = this.doc.documentElement.lang;
  288. const authorElement = this.doc.querySelector("meta[name=author]");
  289. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  290. const creatorElement = this.doc.querySelector("meta[name=creator]");
  291. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  292. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  293. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  294. const url = new URL(this.baseURI);
  295. let size;
  296. if (this.options.displayStats) {
  297. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  298. }
  299. const content = DOM.serialize(this.doc, this.options.compressHTML);
  300. if (this.options.displayStats) {
  301. const contentSize = DOM.getContentSize(content);
  302. this.stats.set("processed", "HTML bytes", contentSize);
  303. this.stats.add("discarded", "HTML bytes", size - contentSize);
  304. }
  305. const filename = await DomProcessorHelper.getFilename(this.options, content);
  306. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  307. return {
  308. stats: this.stats.data,
  309. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  310. filename,
  311. content
  312. };
  313. }
  314. setInputValues() {
  315. this.doc.querySelectorAll("input").forEach(input => {
  316. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  317. if (value) {
  318. input.setAttribute("value", value);
  319. }
  320. });
  321. this.doc.querySelectorAll("textarea").forEach(textarea => {
  322. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  323. if (value) {
  324. textarea.textContent = value;
  325. }
  326. });
  327. this.doc.querySelectorAll("select").forEach(select => {
  328. select.querySelectorAll("option").forEach(option => {
  329. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  330. if (selected) {
  331. option.setAttribute("selected", "");
  332. }
  333. });
  334. });
  335. }
  336. lazyLoadImages() {
  337. DOM.lazyLoad(this.doc);
  338. }
  339. removeDiscardedResources() {
  340. const objectElements = 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\"])");
  341. this.stats.set("discarded", "objects", objectElements.length);
  342. this.stats.set("processed", "objects", objectElements.length);
  343. objectElements.forEach(element => element.remove());
  344. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  345. replacedAttributeValue.forEach(element => {
  346. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  347. if (relValue.length) {
  348. element.setAttribute("rel", relValue);
  349. } else {
  350. element.remove();
  351. }
  352. });
  353. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  354. if (this.options.compressHTML) {
  355. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  356. }
  357. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  358. if (this.options.removeScripts) {
  359. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  360. this.doc.querySelectorAll("[href]").forEach(element => {
  361. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  362. element.removeAttribute("href");
  363. }
  364. });
  365. this.doc.querySelectorAll("[src]").forEach(element => {
  366. if (element.src && element.src.match(/^\s*javascript:/)) {
  367. element.removeAttribute("src");
  368. }
  369. });
  370. }
  371. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  372. this.stats.set("processed", "audio sources", audioSourceElements.length);
  373. if (this.options.removeAudioSrc) {
  374. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  375. audioSourceElements.forEach(element => element.removeAttribute("src"));
  376. }
  377. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  378. this.stats.set("processed", "video sources", videoSourceElements.length);
  379. if (this.options.removeVideoSrc) {
  380. this.stats.set("discarded", "video sources", videoSourceElements.length);
  381. videoSourceElements.forEach(element => element.removeAttribute("src"));
  382. }
  383. }
  384. removeDefaultHeadTags() {
  385. this.doc.querySelectorAll("base").forEach(element => element.remove());
  386. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  387. this.doc.head.querySelector("meta[charset]").remove();
  388. }
  389. }
  390. removeUIElements() {
  391. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  392. }
  393. removeScripts() {
  394. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  395. this.stats.set("discarded", "scripts", scriptElements.length);
  396. this.stats.set("processed", "scripts", scriptElements.length);
  397. scriptElements.forEach(element => element.remove());
  398. }
  399. removeFrames() {
  400. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  401. this.stats.set("discarded", "frames", frameElements.length);
  402. this.stats.set("processed", "frames", frameElements.length);
  403. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  404. }
  405. removeImports() {
  406. const importElements = this.doc.querySelectorAll("link[rel=import]");
  407. this.stats.set("discarded", "HTML imports", importElements.length);
  408. this.stats.set("processed", "HTML imports", importElements.length);
  409. importElements.forEach(element => element.remove());
  410. }
  411. resetCharsetMeta() {
  412. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  413. const metaElement = this.doc.createElement("meta");
  414. metaElement.setAttribute("charset", "utf-8");
  415. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  416. }
  417. insertFaviconLink() {
  418. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  419. if (!faviconElement) {
  420. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  421. }
  422. if (!faviconElement) {
  423. faviconElement = this.doc.createElement("link");
  424. faviconElement.setAttribute("type", "image/x-icon");
  425. faviconElement.setAttribute("rel", "shortcut icon");
  426. faviconElement.setAttribute("href", "/favicon.ico");
  427. }
  428. this.doc.head.appendChild(faviconElement);
  429. }
  430. resolveHrefs() {
  431. this.doc.querySelectorAll("[href]").forEach(element => {
  432. if (element.href) {
  433. const href = element.href.baseVal ? element.href.baseVal : element.href;
  434. const normalizedHref = DomUtil.normalizeURL(href);
  435. if (!normalizedHref || normalizedHref != this.baseURI) {
  436. element.setAttribute("href", href);
  437. }
  438. }
  439. });
  440. }
  441. removeUnusedStyles() {
  442. const stats = DOM.minifyCSS(this.doc);
  443. this.stats.set("processed", "CSS rules", stats.processed);
  444. this.stats.set("discarded", "CSS rules", stats.discarded);
  445. }
  446. removeAlternativeFonts() {
  447. DOM.minifyFonts(this.doc);
  448. }
  449. removeSrcSet() {
  450. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  451. const tagName = element.tagName.toLowerCase();
  452. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  453. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  454. element.removeAttribute(dataAttributeName);
  455. if (imageData) {
  456. if (tagName == "img") {
  457. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  458. element.removeAttribute("srcset");
  459. element.removeAttribute("sizes");
  460. element.src = imageData.source.src;
  461. }
  462. }
  463. if (tagName == "picture") {
  464. const imageElement = element.querySelector("img");
  465. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  466. imageElement.removeAttribute("srcset");
  467. imageElement.removeAttribute("sizes");
  468. imageElement.src = imageData.source.src;
  469. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  470. } else {
  471. if (imageData.sources) {
  472. element.querySelectorAll("source").forEach(sourceElement => {
  473. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  474. sourceElement.remove();
  475. }
  476. });
  477. const sourceElements = element.querySelectorAll("source");
  478. if (sourceElements.length) {
  479. const lastSourceElement = sourceElements[sourceElements.length - 1];
  480. if (imageElement) {
  481. if (lastSourceElement.src) {
  482. imageElement.src = lastSourceElement.src;
  483. } else {
  484. imageElement.removeAttribute("src");
  485. }
  486. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  487. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  488. } else {
  489. imageElement.removeAttribute("srcset");
  490. }
  491. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  492. }
  493. }
  494. }
  495. }
  496. }
  497. }
  498. });
  499. }
  500. postRemoveAlternativeFonts() {
  501. DOM.minifyFonts(this.doc, true);
  502. }
  503. removeHiddenElements() {
  504. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  505. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  506. this.stats.set("processed", "hidden elements", hiddenElements.length);
  507. hiddenElements.forEach(element => element.remove());
  508. }
  509. compressHTML() {
  510. let size;
  511. if (this.options.displayStats) {
  512. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  513. }
  514. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  515. if (this.options.displayStats) {
  516. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  517. }
  518. }
  519. postCompressHTML() {
  520. let size;
  521. if (this.options.displayStats) {
  522. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  523. }
  524. DOM.postMinifyHTML(this.doc);
  525. if (this.options.displayStats) {
  526. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  527. }
  528. }
  529. removeAlternativeMedias() {
  530. const stats = DOM.minifyMedias(this.doc);
  531. this.stats.set("processed", "medias", stats.processed);
  532. this.stats.set("discarded", "medias", stats.discarded);
  533. }
  534. compressCSS() {
  535. this.doc.querySelectorAll("style").forEach(styleElement => {
  536. if (styleElement) {
  537. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  538. }
  539. });
  540. this.doc.querySelectorAll("[style]").forEach(element => {
  541. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  542. });
  543. }
  544. insertSingleFileComment() {
  545. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  546. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  547. }
  548. replaceCanvasElements() {
  549. if (this.options.canvasData) {
  550. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  551. const canvasData = this.options.canvasData[indexCanvasElement];
  552. if (canvasData) {
  553. const imgElement = this.doc.createElement("img");
  554. imgElement.setAttribute("src", canvasData.dataURI);
  555. Array.from(canvasElement.attributes).forEach(attribute => {
  556. if (attribute.value) {
  557. imgElement.setAttribute(attribute.name, attribute.value);
  558. }
  559. });
  560. if (!imgElement.width && canvasData.width) {
  561. imgElement.style.pixelWidth = canvasData.width;
  562. }
  563. if (!imgElement.height && canvasData.height) {
  564. imgElement.style.pixelHeight = canvasData.height;
  565. }
  566. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  567. this.stats.add("processed", "canvas", 1);
  568. }
  569. });
  570. }
  571. }
  572. insertFonts() {
  573. if (this.options.fontsData && this.options.fontsData.length) {
  574. let stylesheetContent = "";
  575. this.options.fontsData.forEach(fontStyles => {
  576. if (fontStyles["font-family"] && fontStyles.src) {
  577. stylesheetContent += "@font-face{";
  578. let stylesContent = "";
  579. Object.keys(fontStyles).forEach(fontStyle => {
  580. if (stylesContent) {
  581. stylesContent += ";";
  582. }
  583. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  584. });
  585. stylesheetContent += stylesContent + "}";
  586. }
  587. });
  588. if (stylesheetContent) {
  589. const styleElement = this.doc.createElement("style");
  590. styleElement.textContent = stylesheetContent;
  591. const existingStyleElement = this.doc.querySelector("style");
  592. if (existingStyleElement) {
  593. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  594. } else {
  595. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  596. }
  597. }
  598. }
  599. }
  600. replaceStyleContents() {
  601. if (this.options.stylesheetContents) {
  602. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  603. if (this.options.stylesheetContents[styleIndex]) {
  604. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  605. }
  606. });
  607. }
  608. }
  609. async pageResources() {
  610. const resourcePromises = [
  611. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI, this.batchRequest),
  612. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", this.baseURI, this.batchRequest),
  613. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], embed[src*=\".svg\"]"), "src", this.baseURI, this.batchRequest),
  614. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI, this.batchRequest),
  615. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI, this.batchRequest),
  616. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", this.baseURI, this.batchRequest),
  617. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  618. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.batchRequest)
  619. ];
  620. if (!this.options.removeAudioSrc) {
  621. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.batchRequest));
  622. }
  623. if (!this.options.removeVideoSrc) {
  624. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.batchRequest));
  625. }
  626. if (this.options.lazyLoadImages) {
  627. const imageSelectors = DOM.lazyLoaderImageSelectors();
  628. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], this.baseURI, this.batchRequest)));
  629. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], this.baseURI, this.batchRequest)));
  630. }
  631. await resourcePromises;
  632. }
  633. async inlineStylesheets(initialization) {
  634. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  635. if (!initialization) {
  636. this.stats.add("processed", "stylesheets", 1);
  637. }
  638. let stylesheetContent = styleElement.textContent;
  639. if (initialization) {
  640. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  641. } else {
  642. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  643. }
  644. styleElement.textContent = stylesheetContent;
  645. }));
  646. }
  647. async scripts() {
  648. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  649. if (scriptElement.src) {
  650. this.stats.add("processed", "scripts", 1);
  651. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  652. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  653. }
  654. scriptElement.removeAttribute("src");
  655. }));
  656. }
  657. async frames(initialization) {
  658. if (this.options.framesData) {
  659. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  660. await Promise.all(frameElements.map(async frameElement => {
  661. DomProcessorHelper.setFrameEmptySrc(frameElement);
  662. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  663. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  664. if (frameWindowId) {
  665. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  666. if (frameData) {
  667. if (initialization) {
  668. const options = Object.create(this.options);
  669. options.insertSingleFileComment = false;
  670. options.insertFaviconLink = false;
  671. options.doc = null;
  672. options.win = null;
  673. options.url = frameData.baseURI;
  674. options.windowId = frameWindowId;
  675. if (frameData.content) {
  676. options.content = frameData.content;
  677. options.canvasData = frameData.canvasData;
  678. options.stylesheetContents = frameData.stylesheetContents;
  679. options.currentSrcImages = frameData.currentSrcImages;
  680. frameData.processor = new PageProcessor(options);
  681. frameData.frameElement = frameElement;
  682. await frameData.processor.loadPage();
  683. await frameData.processor.initialize();
  684. frameData.maxResources = this.batchRequest.getMaxResources();
  685. }
  686. } else {
  687. if (frameData.processor) {
  688. this.stats.add("processed", "frames", 1);
  689. await frameData.processor.preparePageData();
  690. const pageData = await frameData.processor.getPageData();
  691. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  692. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  693. this.stats.addAll(pageData);
  694. } else {
  695. this.stats.add("discarded", "frames", 1);
  696. }
  697. }
  698. }
  699. }
  700. }));
  701. }
  702. }
  703. async htmlImports(initialization) {
  704. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  705. if (!this.relImportProcessors) {
  706. this.relImportProcessors = new Map();
  707. }
  708. await Promise.all(linkElements.map(async linkElement => {
  709. if (initialization) {
  710. const resourceURL = linkElement.href;
  711. const options = Object.create(this.options);
  712. options.insertSingleFileComment = false;
  713. options.insertFaviconLink = false;
  714. options.doc = null;
  715. options.win = null;
  716. options.url = resourceURL;
  717. if (resourceURL) {
  718. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  719. const processor = new PageProcessor(options);
  720. this.relImportProcessors.set(linkElement, processor);
  721. await processor.loadPage();
  722. return processor.initialize();
  723. }
  724. }
  725. } else {
  726. linkElement.setAttribute("href", EMPTY_DATA_URI);
  727. const processor = this.relImportProcessors.get(linkElement);
  728. if (processor) {
  729. this.stats.add("processed", "HTML imports", 1);
  730. this.relImportProcessors.delete(linkElement);
  731. const pageData = await processor.getPageData();
  732. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  733. this.stats.addAll(pageData);
  734. } else {
  735. this.stats.add("discarded", "HTML imports", 1);
  736. }
  737. }
  738. }));
  739. }
  740. async attributeStyles(initialization) {
  741. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  742. let stylesheetContent = element.getAttribute("style");
  743. if (initialization) {
  744. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  745. } else {
  746. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  747. }
  748. element.setAttribute("style", stylesheetContent);
  749. }));
  750. }
  751. async linkStylesheets() {
  752. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  753. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  754. const styleElement = this.doc.createElement("style");
  755. styleElement.textContent = stylesheetContent;
  756. linkElement.parentElement.replaceChild(styleElement, linkElement);
  757. }));
  758. }
  759. }
  760. // ---------
  761. // DomHelper
  762. // ---------
  763. const REGEXP_AMP = /&/g;
  764. const REGEXP_NBSP = /\u00a0/g;
  765. const REGEXP_START_TAG = /</g;
  766. const REGEXP_END_TAG = />/g;
  767. const REGEXP_URL_HASH = /(#.+?)$/;
  768. const PREFIX_DATA_URI_IMAGE = "data:image/";
  769. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  770. class DomProcessorHelper {
  771. static async getFilename(options, content) {
  772. let filename = options.filenameTemplate;
  773. const date = new Date();
  774. const url = new URL(options.url);
  775. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  776. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  777. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  778. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  779. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  780. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  781. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  782. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  783. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  784. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  785. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  786. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  787. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  788. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  789. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  790. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  791. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  792. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  793. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  794. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  795. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  796. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  797. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  798. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  799. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  800. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  801. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  802. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  803. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  804. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  805. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  806. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  807. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  808. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  809. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  810. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  811. if (!options.backgroundSave) {
  812. filename = filename.replace(/\//g, "_");
  813. }
  814. if (filename.length > 192) {
  815. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  816. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  817. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  818. }
  819. if (!filename) {
  820. filename = "Unnamed page";
  821. }
  822. return filename;
  823. }
  824. static setFrameEmptySrc(frameElement) {
  825. if (frameElement.tagName == "OBJECT") {
  826. frameElement.setAttribute("data", "data:text/html,");
  827. } else {
  828. frameElement.setAttribute("srcdoc", "");
  829. frameElement.removeAttribute("src");
  830. }
  831. }
  832. static setFrameContent(frameElement, content) {
  833. if (frameElement.tagName == "OBJECT") {
  834. frameElement.setAttribute("data", "data:text/html," + content);
  835. } else {
  836. frameElement.setAttribute("srcdoc", content);
  837. frameElement.removeAttribute("src");
  838. }
  839. }
  840. static isolateElements(rootElement) {
  841. rootElement.querySelectorAll("*").forEach(element => {
  842. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  843. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  844. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  845. element.remove();
  846. }
  847. });
  848. isolateParentElements(rootElement.parentElement, rootElement);
  849. function isolateParentElements(parentElement, element) {
  850. if (parentElement) {
  851. Array.from(parentElement.childNodes).forEach(node => {
  852. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  853. node.remove();
  854. }
  855. });
  856. }
  857. element = element.parentElement;
  858. if (element && element.parentElement) {
  859. isolateParentElements(element.parentElement, element);
  860. }
  861. }
  862. }
  863. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  864. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  865. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  866. const imports = DomUtil.getImportFunctions(stylesheetContent);
  867. await Promise.all(imports.map(async cssImport => {
  868. const match = DomUtil.matchImport(cssImport);
  869. if (match) {
  870. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  871. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  872. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  873. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  874. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  875. if (stylesheetContent.includes(cssImport)) {
  876. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  877. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  878. }
  879. }
  880. }
  881. }));
  882. return stylesheetContent;
  883. }
  884. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  885. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  886. urlFunctions.map(urlFunction => {
  887. const originalResourceURL = DomUtil.matchURL(urlFunction);
  888. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  889. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  890. const resolvedURL = new URL(resourceURL, baseURI).href;
  891. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  892. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  893. }
  894. } else {
  895. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  896. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  897. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  898. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  899. }
  900. }
  901. }
  902. });
  903. return stylesheetContent;
  904. }
  905. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  906. resourceURL = DomUtil.normalizeURL(resourceURL);
  907. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  908. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  909. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  910. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  911. return stylesheetContent;
  912. }
  913. }
  914. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  915. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  916. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  917. let indexVariable = 0;
  918. await Promise.all(urlFunctions.map(async urlFunction => {
  919. const originalResourceURL = DomUtil.matchURL(urlFunction);
  920. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  921. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  922. const dataURI = await batchRequest.addURL(resourceURL);
  923. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  924. const functions = stylesheetContent.match(DomUtil.getRegExp(urlFunction));
  925. if (functions && functions.length > 1) {
  926. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  927. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  928. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), "var(" + variableName + ")");
  929. indexVariable++;
  930. } else {
  931. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  932. }
  933. } else {
  934. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  935. }
  936. }
  937. }));
  938. if (indexVariable) {
  939. stylesheetContent = ":root{" + stylesheetContent;
  940. }
  941. return stylesheetContent;
  942. }
  943. static async processAttribute(resourceElements, attributeName, baseURI, batchRequest) {
  944. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  945. let resourceURL = resourceElement.getAttribute(attributeName);
  946. if (resourceURL) {
  947. resourceURL = DomUtil.normalizeURL(resourceURL);
  948. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  949. try {
  950. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  951. resourceElement.setAttribute(attributeName, dataURI);
  952. } catch (error) {
  953. /* ignored */
  954. }
  955. }
  956. }
  957. }));
  958. }
  959. static async processXLinks(resourceElements, baseURI, batchRequest) {
  960. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  961. const originalResourceURL = resourceElement.getAttribute("xlink:href");
  962. if (originalResourceURL) {
  963. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  964. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  965. try {
  966. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  967. const DOMParser = DOM.getParser();
  968. if (DOMParser) {
  969. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  970. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  971. if (hashMatch && hashMatch[0]) {
  972. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  973. if (symbolElement) {
  974. resourceElement.setAttribute("xlink:href", hashMatch[0]);
  975. resourceElement.parentElement.appendChild(symbolElement);
  976. }
  977. } else {
  978. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  979. }
  980. } else {
  981. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  982. }
  983. } catch (error) {
  984. /* ignored */
  985. }
  986. }
  987. }
  988. }));
  989. }
  990. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  991. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  992. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  993. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  994. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  995. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  996. try {
  997. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  998. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  999. } catch (error) {
  1000. /* ignored */
  1001. }
  1002. }
  1003. }));
  1004. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1005. }));
  1006. }
  1007. }
  1008. // -------
  1009. // DomUtil
  1010. // -------
  1011. const DATA_URI_PREFIX = "data:";
  1012. const BLOB_URI_PREFIX = "blob:";
  1013. const HTTP_URI_PREFIX = /^https?:\/\//;
  1014. const ABOUT_BLANK_URI = "about:blank";
  1015. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1016. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1017. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1018. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1019. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1020. 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;
  1021. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1022. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1023. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1024. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1025. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1026. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1027. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1028. class DomUtil {
  1029. static normalizeURL(url) {
  1030. if (url.startsWith(DATA_URI_PREFIX)) {
  1031. return url;
  1032. } else {
  1033. return url.split("#")[0];
  1034. }
  1035. }
  1036. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1037. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1038. if (template.match(replaceRegExp)) {
  1039. let value = await valueGetter();
  1040. if (!dontReplaceSlash) {
  1041. value = value.replace(/\/+/g, "_");
  1042. }
  1043. return template.replace(replaceRegExp, value);
  1044. }
  1045. return template;
  1046. }
  1047. static getLastSegment(url) {
  1048. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1049. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1050. if (!lastSegment) {
  1051. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1052. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1053. }
  1054. if (!lastSegment) {
  1055. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1056. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1057. }
  1058. if (!lastSegment) {
  1059. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1060. }
  1061. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1062. return lastSegment;
  1063. }
  1064. static getRegExp(string) {
  1065. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1066. }
  1067. static getUrlFunctions(stylesheetContent) {
  1068. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1069. }
  1070. static getImportFunctions(stylesheetContent) {
  1071. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1072. }
  1073. static matchURL(stylesheetContent) {
  1074. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1075. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1076. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1077. return match && match[1];
  1078. }
  1079. static testValidPath(resourceURL) {
  1080. return !resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(BLOB_URI_PREFIX) && resourceURL != ABOUT_BLANK_URI && (!resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(NOT_EMPTY_URL));
  1081. }
  1082. static matchImport(stylesheetContent) {
  1083. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1084. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1085. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1086. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1087. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1088. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1089. if (match) {
  1090. const [, resourceURL, media] = match;
  1091. return { resourceURL, media };
  1092. }
  1093. }
  1094. static removeCssComments(stylesheetContent) {
  1095. let start, end;
  1096. do {
  1097. start = stylesheetContent.indexOf("/*");
  1098. end = stylesheetContent.indexOf("*/", start + 2);
  1099. if (start != -1 && end != -1) {
  1100. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1101. }
  1102. } while (start != -1 && end != -1);
  1103. return stylesheetContent;
  1104. }
  1105. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1106. if (mediaQuery) {
  1107. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1108. } else {
  1109. return stylesheetContent;
  1110. }
  1111. }
  1112. }
  1113. // -----
  1114. // Stats
  1115. // -----
  1116. const STATS_DEFAULT_VALUES = {
  1117. discarded: {
  1118. "HTML bytes": 0,
  1119. "hidden elements": 0,
  1120. "HTML imports": 0,
  1121. scripts: 0,
  1122. objects: 0,
  1123. "audio sources": 0,
  1124. "video sources": 0,
  1125. frames: 0,
  1126. "CSS rules": 0,
  1127. canvas: 0,
  1128. stylesheets: 0,
  1129. resources: 0,
  1130. medias: 0
  1131. },
  1132. processed: {
  1133. "HTML bytes": 0,
  1134. "hidden elements": 0,
  1135. "HTML imports": 0,
  1136. scripts: 0,
  1137. objects: 0,
  1138. "audio sources": 0,
  1139. "video sources": 0,
  1140. frames: 0,
  1141. "CSS rules": 0,
  1142. canvas: 0,
  1143. stylesheets: 0,
  1144. resources: 0,
  1145. medias: 0
  1146. }
  1147. };
  1148. class Stats {
  1149. constructor(options) {
  1150. this.options = options;
  1151. if (options.displayStats) {
  1152. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1153. }
  1154. }
  1155. set(type, subType, value) {
  1156. if (this.options.displayStats) {
  1157. this.data[type][subType] = value;
  1158. }
  1159. }
  1160. add(type, subType, value) {
  1161. if (this.options.displayStats) {
  1162. this.data[type][subType] += value;
  1163. }
  1164. }
  1165. addAll(pageData) {
  1166. if (this.options.displayStats) {
  1167. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1168. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1169. }
  1170. }
  1171. }
  1172. return { getClass };
  1173. })();