single-file-core.js 51 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223
  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. { option: "removeHiddenElements", action: "removeHiddenElements" }
  80. ],
  81. async: [
  82. { action: "inlineStylesheets" },
  83. { action: "linkStylesheets" },
  84. { action: "attributeStyles" },
  85. { option: "!removeFrames", action: "frames" },
  86. { option: "!removeImports", action: "htmlImports" }
  87. ]
  88. }, {
  89. sync: [
  90. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  91. { option: "compressHTML", action: "compressHTML" },
  92. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  93. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  94. ],
  95. async: [
  96. { action: "inlineStylesheets" },
  97. { action: "attributeStyles" },
  98. { action: "pageResources" },
  99. { option: "!removeScripts", action: "scripts" }
  100. ]
  101. }, {
  102. sync: [
  103. { option: "lazyLoadImages", action: "lazyLoadImages" },
  104. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  105. { option: "compressCSS", action: "compressCSS" }
  106. ],
  107. async: [
  108. { option: "!removeFrames", action: "frames" },
  109. { option: "!removeImports", action: "htmlImports" },
  110. ]
  111. }, {
  112. sync: [
  113. { option: "compressHTML", action: "postCompressHTML" },
  114. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  115. { action: "removeDefaultHeadTags" }
  116. ]
  117. }];
  118. class PageProcessor {
  119. constructor(options) {
  120. this.options = options;
  121. this.options.url = this.options.url || this.options.doc.location.href;
  122. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  123. this.batchRequest = new BatchRequest();
  124. this.processor = new DOMProcessor(options, this.batchRequest);
  125. if (this.options.doc) {
  126. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  127. this.options.canvasData = docData.canvasData;
  128. this.options.stylesheetContents = docData.stylesheetContents;
  129. this.options.imageData = docData.imageData;
  130. }
  131. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  132. this.onprogress = options.onprogress || (() => { });
  133. }
  134. async loadPage() {
  135. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  136. await this.processor.loadPage(this.options.content);
  137. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  138. }
  139. async initialize() {
  140. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  141. await this.executeStage(0);
  142. this.pendingPromises = this.executeStage(1);
  143. if (this.options.doc) {
  144. DOM.postProcessDoc(this.options.doc, this.options);
  145. this.options.doc = null;
  146. this.options.win = null;
  147. }
  148. }
  149. async preparePageData() {
  150. if (!this.options.windowId) {
  151. this.processor.initialize(this.batchRequest);
  152. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  153. }
  154. await this.batchRequest.run(details => {
  155. details.pageURL = this.options.url;
  156. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  157. }, this.options);
  158. await this.pendingPromises;
  159. await this.executeStage(2);
  160. await this.executeStage(3);
  161. }
  162. async getPageData() {
  163. if (!this.options.windowId) {
  164. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  165. }
  166. return this.processor.getPageData();
  167. }
  168. async executeStage(step) {
  169. STAGES[step].sync.forEach(task => {
  170. // const startTime = Date.now();
  171. this.executeTask(task, !step);
  172. // console.log(task.action, Date.now() - startTime);
  173. });
  174. if (STAGES[step].async) {
  175. return Promise.all(STAGES[step].async.map(task => {
  176. // const startTime = Date.now();
  177. const promise = this.executeTask(task, !step);
  178. // promise.then(() => console.log(task.action, Date.now() - startTime));
  179. return promise;
  180. }));
  181. }
  182. }
  183. executeTask(task, initialization) {
  184. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  185. return this.processor[task.action](initialization);
  186. }
  187. }
  188. }
  189. // --------
  190. // BatchRequest
  191. // --------
  192. class BatchRequest {
  193. constructor() {
  194. this.requests = new Map();
  195. }
  196. async addURL(resourceURL, asDataURI = true) {
  197. return new Promise((resolve, reject) => {
  198. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  199. const resourceRequests = this.requests.get(requestKey);
  200. if (resourceRequests) {
  201. resourceRequests.push({ resolve, reject });
  202. } else {
  203. this.requests.set(requestKey, [{ resolve, reject }]);
  204. }
  205. });
  206. }
  207. getMaxResources() {
  208. return Array.from(this.requests.keys()).length;
  209. }
  210. async run(onloadListener, options) {
  211. const resourceURLs = Array.from(this.requests.keys());
  212. let indexResource = 0;
  213. return Promise.all(resourceURLs.map(async requestKey => {
  214. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  215. const resourceRequests = this.requests.get(requestKey);
  216. try {
  217. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  218. indexResource = indexResource + 1;
  219. onloadListener({ index: indexResource, url: resourceURL });
  220. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  221. } catch (error) {
  222. indexResource = indexResource + 1;
  223. onloadListener({ index: indexResource, url: resourceURL });
  224. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  225. }
  226. this.requests.delete(requestKey);
  227. }));
  228. }
  229. }
  230. // ------------
  231. // DOMProcessor
  232. // ------------
  233. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  234. const EMPTY_DATA_URI = "data:base64,";
  235. class DOMProcessor {
  236. constructor(options, batchRequest) {
  237. this.options = options;
  238. this.stats = new Stats(options);
  239. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  240. this.batchRequest = batchRequest;
  241. }
  242. initialize() {
  243. this.maxResources = this.batchRequest.getMaxResources();
  244. if (!this.options.removeFrames) {
  245. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  246. }
  247. this.stats.set("processed", "resources", this.maxResources);
  248. }
  249. async loadPage(pageContent) {
  250. if (!pageContent || this.options.saveRawPage) {
  251. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  252. }
  253. this.doc = DOM.createDoc(pageContent, this.baseURI);
  254. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  255. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  256. await DOMProcessor.loadEscapedFragmentPage();
  257. }
  258. }
  259. async loadEscapedFragmentPage() {
  260. if (this.baseURI.includes("?")) {
  261. this.baseURI += "&";
  262. } else {
  263. this.baseURI += "?";
  264. }
  265. this.baseURI += ESCAPED_FRAGMENT;
  266. await this.loadPage();
  267. }
  268. async getPageData() {
  269. DOM.postProcessDoc(this.doc, this.options);
  270. if (this.options.selected) {
  271. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  272. if (rootElement) {
  273. DomProcessorHelper.isolateElements(rootElement);
  274. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  275. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  276. }
  277. }
  278. const titleElement = this.doc.querySelector("title");
  279. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  280. this.options.info = {};
  281. const descriptionElement = this.doc.querySelector("meta[name=description]");
  282. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  283. this.options.info.lang = this.doc.documentElement.lang;
  284. const authorElement = this.doc.querySelector("meta[name=author]");
  285. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  286. const creatorElement = this.doc.querySelector("meta[name=creator]");
  287. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  288. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  289. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  290. const url = new URL(this.baseURI);
  291. let size;
  292. if (this.options.displayStats) {
  293. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  294. }
  295. const content = DOM.serialize(this.doc, this.options.compressHTML);
  296. if (this.options.displayStats) {
  297. const contentSize = DOM.getContentSize(content);
  298. this.stats.set("processed", "HTML bytes", contentSize);
  299. this.stats.add("discarded", "HTML bytes", size - contentSize);
  300. }
  301. const filename = await DomProcessorHelper.getFilename(this.options, content);
  302. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  303. return {
  304. stats: this.stats.data,
  305. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  306. filename,
  307. content
  308. };
  309. }
  310. setInputValues() {
  311. this.doc.querySelectorAll("input").forEach(input => {
  312. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  313. if (value) {
  314. input.setAttribute("value", value);
  315. }
  316. });
  317. this.doc.querySelectorAll("textarea").forEach(textarea => {
  318. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  319. if (value) {
  320. textarea.textContent = value;
  321. }
  322. });
  323. this.doc.querySelectorAll("select").forEach(select => {
  324. select.querySelectorAll("option").forEach(option => {
  325. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  326. if (selected) {
  327. option.setAttribute("selected", "");
  328. }
  329. });
  330. });
  331. }
  332. lazyLoadImages() {
  333. DOM.lazyLoad(this.doc);
  334. }
  335. removeDiscardedResources() {
  336. 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\"])");
  337. this.stats.set("discarded", "objects", objectElements.length);
  338. this.stats.set("processed", "objects", objectElements.length);
  339. objectElements.forEach(element => element.remove());
  340. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  341. replacedAttributeValue.forEach(element => {
  342. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  343. if (relValue.length) {
  344. element.setAttribute("rel", relValue);
  345. } else {
  346. element.remove();
  347. }
  348. });
  349. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"], input[type=hidden]").forEach(element => element.remove());
  350. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  351. if (this.options.removeScripts) {
  352. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  353. this.doc.querySelectorAll("[href]").forEach(element => {
  354. if (element.href && element.href.match(/^\s*javascript:/)) {
  355. element.removeAttribute("href");
  356. }
  357. });
  358. this.doc.querySelectorAll("[src]").forEach(element => {
  359. if (element.src && element.src.match(/^\s*javascript:/)) {
  360. element.removeAttribute("src");
  361. }
  362. });
  363. }
  364. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  365. this.stats.set("processed", "audio sources", audioSourceElements.length);
  366. if (this.options.removeAudioSrc) {
  367. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  368. audioSourceElements.forEach(element => element.removeAttribute("src"));
  369. }
  370. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  371. this.stats.set("processed", "video sources", videoSourceElements.length);
  372. if (this.options.removeVideoSrc) {
  373. this.stats.set("discarded", "video sources", videoSourceElements.length);
  374. videoSourceElements.forEach(element => element.removeAttribute("src"));
  375. }
  376. }
  377. removeDefaultHeadTags() {
  378. this.doc.querySelectorAll("base").forEach(element => element.remove());
  379. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  380. this.doc.head.querySelector("meta[charset]").remove();
  381. }
  382. }
  383. removeUIElements() {
  384. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  385. }
  386. removeScripts() {
  387. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  388. this.stats.set("discarded", "scripts", scriptElements.length);
  389. this.stats.set("processed", "scripts", scriptElements.length);
  390. scriptElements.forEach(element => element.remove());
  391. }
  392. removeFrames() {
  393. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  394. this.stats.set("discarded", "frames", frameElements.length);
  395. this.stats.set("processed", "frames", frameElements.length);
  396. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  397. }
  398. removeImports() {
  399. const importElements = this.doc.querySelectorAll("link[rel=import]");
  400. this.stats.set("discarded", "HTML imports", importElements.length);
  401. this.stats.set("processed", "HTML imports", importElements.length);
  402. importElements.forEach(element => element.remove());
  403. }
  404. resetCharsetMeta() {
  405. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  406. const metaElement = this.doc.createElement("meta");
  407. metaElement.setAttribute("charset", "utf-8");
  408. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  409. }
  410. insertFaviconLink() {
  411. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  412. if (!faviconElement) {
  413. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  414. }
  415. if (!faviconElement) {
  416. faviconElement = this.doc.createElement("link");
  417. faviconElement.setAttribute("type", "image/x-icon");
  418. faviconElement.setAttribute("rel", "shortcut icon");
  419. faviconElement.setAttribute("href", "/favicon.ico");
  420. }
  421. this.doc.head.appendChild(faviconElement);
  422. }
  423. resolveHrefs() {
  424. this.doc.querySelectorAll("[href]").forEach(element => {
  425. if (element.href) {
  426. const href = element.href.baseVal ? element.href.baseVal : element.href;
  427. const normalizedHref = DomUtil.normalizeURL(href);
  428. if (!normalizedHref || normalizedHref != this.baseURI) {
  429. element.setAttribute("href", href);
  430. }
  431. }
  432. });
  433. }
  434. removeUnusedStyles() {
  435. const stats = DOM.minifyCSS(this.doc);
  436. this.stats.set("processed", "CSS rules", stats.processed);
  437. this.stats.set("discarded", "CSS rules", stats.discarded);
  438. }
  439. removeAlternativeFonts() {
  440. DOM.minifyFonts(this.doc);
  441. }
  442. removeSrcSet() {
  443. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  444. const tagName = element.tagName.toLowerCase();
  445. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  446. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  447. element.removeAttribute(dataAttributeName);
  448. if (imageData) {
  449. if (tagName == "img") {
  450. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  451. element.removeAttribute("srcset");
  452. element.removeAttribute("sizes");
  453. element.src = imageData.source.src;
  454. }
  455. }
  456. if (tagName == "picture") {
  457. const imageElement = element.querySelector("img");
  458. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  459. imageElement.removeAttribute("srcset");
  460. imageElement.removeAttribute("sizes");
  461. imageElement.src = imageData.source.src;
  462. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  463. } else {
  464. if (imageData.sources) {
  465. element.querySelectorAll("source").forEach(sourceElement => {
  466. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  467. sourceElement.remove();
  468. }
  469. });
  470. const sourceElements = element.querySelectorAll("source");
  471. if (sourceElements.length) {
  472. const lastSourceElement = sourceElements[sourceElements.length - 1];
  473. if (imageElement) {
  474. if (lastSourceElement.src) {
  475. imageElement.src = lastSourceElement.src;
  476. } else {
  477. imageElement.removeAttribute("src");
  478. }
  479. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  480. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  481. } else {
  482. imageElement.removeAttribute("srcset");
  483. }
  484. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  485. }
  486. }
  487. }
  488. }
  489. }
  490. }
  491. });
  492. }
  493. postRemoveAlternativeFonts() {
  494. DOM.minifyFonts(this.doc, true);
  495. }
  496. removeHiddenElements() {
  497. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  498. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  499. this.stats.set("processed", "hidden elements", hiddenElements.length);
  500. hiddenElements.forEach(element => element.remove());
  501. }
  502. compressHTML() {
  503. let size;
  504. if (this.options.displayStats) {
  505. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  506. }
  507. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  508. if (this.options.displayStats) {
  509. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  510. }
  511. }
  512. postCompressHTML() {
  513. let size;
  514. if (this.options.displayStats) {
  515. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  516. }
  517. DOM.postMinifyHTML(this.doc);
  518. if (this.options.displayStats) {
  519. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  520. }
  521. }
  522. removeAlternativeMedias() {
  523. const stats = DOM.minifyMedias(this.doc);
  524. this.stats.set("processed", "medias", stats.processed);
  525. this.stats.set("discarded", "medias", stats.discarded);
  526. }
  527. compressCSS() {
  528. this.doc.querySelectorAll("style").forEach(styleElement => {
  529. if (styleElement) {
  530. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  531. }
  532. });
  533. this.doc.querySelectorAll("[style]").forEach(element => {
  534. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  535. });
  536. }
  537. insertSingleFileComment() {
  538. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  539. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  540. }
  541. replaceCanvasElements() {
  542. if (this.options.canvasData) {
  543. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  544. const canvasData = this.options.canvasData[indexCanvasElement];
  545. if (canvasData) {
  546. const imgElement = this.doc.createElement("img");
  547. imgElement.setAttribute("src", canvasData.dataURI);
  548. Array.from(canvasElement.attributes).forEach(attribute => {
  549. if (attribute.value) {
  550. imgElement.setAttribute(attribute.name, attribute.value);
  551. }
  552. });
  553. if (!imgElement.width && canvasData.width) {
  554. imgElement.style.pixelWidth = canvasData.width;
  555. }
  556. if (!imgElement.height && canvasData.height) {
  557. imgElement.style.pixelHeight = canvasData.height;
  558. }
  559. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  560. this.stats.add("processed", "canvas", 1);
  561. }
  562. });
  563. }
  564. }
  565. replaceStyleContents() {
  566. if (this.options.stylesheetContents) {
  567. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  568. if (this.options.stylesheetContents[styleIndex]) {
  569. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  570. }
  571. });
  572. }
  573. }
  574. async pageResources() {
  575. const resourcePromises = [
  576. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI, this.batchRequest),
  577. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", this.baseURI, this.batchRequest),
  578. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], embed[src*=\".svg\"]"), "src", this.baseURI, this.batchRequest),
  579. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI, this.batchRequest),
  580. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI, this.batchRequest),
  581. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", this.baseURI, this.batchRequest),
  582. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  583. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.batchRequest)
  584. ];
  585. if (!this.options.removeAudioSrc) {
  586. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.batchRequest));
  587. }
  588. if (!this.options.removeVideoSrc) {
  589. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.batchRequest));
  590. }
  591. if (this.options.lazyLoadImages) {
  592. const imageSelectors = DOM.lazyLoaderImageSelectors();
  593. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], this.baseURI, this.batchRequest)));
  594. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], this.baseURI, this.batchRequest)));
  595. }
  596. await resourcePromises;
  597. }
  598. async inlineStylesheets(initialization) {
  599. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  600. if (!initialization) {
  601. this.stats.add("processed", "stylesheets", 1);
  602. }
  603. let stylesheetContent = styleElement.textContent;
  604. if (initialization) {
  605. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  606. } else {
  607. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  608. }
  609. styleElement.textContent = stylesheetContent;
  610. }));
  611. }
  612. async scripts() {
  613. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  614. if (scriptElement.src) {
  615. this.stats.add("processed", "scripts", 1);
  616. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  617. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  618. }
  619. scriptElement.removeAttribute("src");
  620. }));
  621. }
  622. async frames(initialization) {
  623. if (this.options.framesData) {
  624. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  625. await Promise.all(frameElements.map(async frameElement => {
  626. DomProcessorHelper.setFrameEmptySrc(frameElement);
  627. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  628. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  629. if (frameWindowId) {
  630. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  631. if (frameData) {
  632. if (initialization) {
  633. const options = Object.create(this.options);
  634. options.insertSingleFileComment = false;
  635. options.insertFaviconLink = false;
  636. options.doc = null;
  637. options.win = null;
  638. options.url = frameData.baseURI;
  639. options.windowId = frameWindowId;
  640. if (frameData.content) {
  641. options.content = frameData.content;
  642. options.canvasData = frameData.canvasData;
  643. options.stylesheetContents = frameData.stylesheetContents;
  644. options.currentSrcImages = frameData.currentSrcImages;
  645. frameData.processor = new PageProcessor(options);
  646. frameData.frameElement = frameElement;
  647. await frameData.processor.loadPage();
  648. await frameData.processor.initialize();
  649. frameData.maxResources = this.batchRequest.getMaxResources();
  650. }
  651. } else {
  652. if (frameData.processor) {
  653. this.stats.add("processed", "frames", 1);
  654. await frameData.processor.preparePageData();
  655. const pageData = await frameData.processor.getPageData();
  656. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  657. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  658. this.stats.addAll(pageData);
  659. } else {
  660. this.stats.add("discarded", "frames", 1);
  661. }
  662. }
  663. }
  664. }
  665. }));
  666. }
  667. }
  668. async htmlImports(initialization) {
  669. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  670. if (!this.relImportProcessors) {
  671. this.relImportProcessors = new Map();
  672. }
  673. await Promise.all(linkElements.map(async linkElement => {
  674. if (initialization) {
  675. const resourceURL = linkElement.href;
  676. const options = Object.create(this.options);
  677. options.insertSingleFileComment = false;
  678. options.insertFaviconLink = false;
  679. options.doc = null;
  680. options.win = null;
  681. options.url = resourceURL;
  682. if (resourceURL) {
  683. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  684. const processor = new PageProcessor(options);
  685. this.relImportProcessors.set(linkElement, processor);
  686. await processor.loadPage();
  687. return processor.initialize();
  688. }
  689. }
  690. } else {
  691. linkElement.setAttribute("href", EMPTY_DATA_URI);
  692. const processor = this.relImportProcessors.get(linkElement);
  693. if (processor) {
  694. this.stats.add("processed", "HTML imports", 1);
  695. this.relImportProcessors.delete(linkElement);
  696. const pageData = await processor.getPageData();
  697. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  698. this.stats.addAll(pageData);
  699. } else {
  700. this.stats.add("discarded", "HTML imports", 1);
  701. }
  702. }
  703. }));
  704. }
  705. async attributeStyles(initialization) {
  706. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  707. let stylesheetContent = element.getAttribute("style");
  708. if (initialization) {
  709. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  710. } else {
  711. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  712. }
  713. element.setAttribute("style", stylesheetContent);
  714. }));
  715. }
  716. async linkStylesheets() {
  717. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  718. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  719. const styleElement = this.doc.createElement("style");
  720. styleElement.textContent = stylesheetContent;
  721. linkElement.parentElement.replaceChild(styleElement, linkElement);
  722. }));
  723. }
  724. }
  725. // ---------
  726. // DomHelper
  727. // ---------
  728. const REGEXP_AMP = /&/g;
  729. const REGEXP_NBSP = /\u00a0/g;
  730. const REGEXP_START_TAG = /</g;
  731. const REGEXP_END_TAG = />/g;
  732. const REGEXP_URL_HASH = /(#.+?)$/;
  733. const PREFIX_DATA_URI_IMAGE = "data:image/";
  734. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  735. class DomProcessorHelper {
  736. static async getFilename(options, content) {
  737. let filename = options.filenameTemplate;
  738. const date = new Date();
  739. const url = new URL(options.url);
  740. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  741. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  742. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  743. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  744. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  745. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  746. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  747. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  748. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  749. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  750. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  751. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  752. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  753. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  754. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  755. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  756. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  757. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  758. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  759. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  760. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  761. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  762. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  763. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  764. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  765. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  766. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  767. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  768. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  769. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  770. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  771. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  772. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  773. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  774. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  775. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  776. if (filename.length > 192) {
  777. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  778. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  779. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  780. }
  781. if (!filename) {
  782. filename = "Unnamed page";
  783. }
  784. return filename;
  785. }
  786. static setFrameEmptySrc(frameElement) {
  787. if (frameElement.tagName == "OBJECT") {
  788. frameElement.setAttribute("data", "data:text/html,");
  789. } else {
  790. frameElement.setAttribute("srcdoc", "");
  791. frameElement.removeAttribute("src");
  792. }
  793. }
  794. static setFrameContent(frameElement, content) {
  795. if (frameElement.tagName == "OBJECT") {
  796. frameElement.setAttribute("data", "data:text/html," + content);
  797. } else {
  798. frameElement.setAttribute("srcdoc", content);
  799. frameElement.removeAttribute("src");
  800. }
  801. }
  802. static isolateElements(rootElement) {
  803. rootElement.querySelectorAll("*").forEach(element => {
  804. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  805. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  806. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  807. element.remove();
  808. }
  809. });
  810. isolateParentElements(rootElement.parentElement, rootElement);
  811. function isolateParentElements(parentElement, element) {
  812. if (parentElement) {
  813. Array.from(parentElement.childNodes).forEach(node => {
  814. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  815. node.remove();
  816. }
  817. });
  818. }
  819. element = element.parentElement;
  820. if (element && element.parentElement) {
  821. isolateParentElements(element.parentElement, element);
  822. }
  823. }
  824. }
  825. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  826. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  827. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  828. const imports = DomUtil.getImportFunctions(stylesheetContent);
  829. await Promise.all(imports.map(async cssImport => {
  830. const match = DomUtil.matchImport(cssImport);
  831. if (match) {
  832. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  833. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  834. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  835. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  836. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  837. if (stylesheetContent.includes(cssImport)) {
  838. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  839. if (cssImport.endsWith(";") && !stylesheetContent.endsWith(";")) {
  840. stylesheetContent += ";";
  841. }
  842. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  843. }
  844. }
  845. }
  846. }));
  847. return stylesheetContent;
  848. }
  849. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  850. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  851. urlFunctions.map(urlFunction => {
  852. const originalResourceURL = DomUtil.matchURL(urlFunction);
  853. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  854. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  855. const resolvedURL = new URL(resourceURL, baseURI).href;
  856. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  857. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  858. }
  859. } else {
  860. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  861. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  862. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  863. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  864. }
  865. }
  866. }
  867. });
  868. return stylesheetContent;
  869. }
  870. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  871. resourceURL = DomUtil.normalizeURL(resourceURL);
  872. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  873. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  874. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  875. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  876. return stylesheetContent;
  877. }
  878. }
  879. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  880. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  881. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  882. let indexVariable = 0;
  883. await Promise.all(urlFunctions.map(async urlFunction => {
  884. const originalResourceURL = DomUtil.matchURL(urlFunction);
  885. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  886. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  887. const dataURI = await batchRequest.addURL(resourceURL);
  888. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  889. const functions = stylesheetContent.match(DomUtil.getRegExp(urlFunction));
  890. if (functions.length > 1) {
  891. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  892. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  893. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), "var(" + variableName + ")");
  894. indexVariable++;
  895. } else {
  896. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  897. }
  898. } else {
  899. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  900. }
  901. }
  902. }));
  903. if (indexVariable) {
  904. stylesheetContent = ":root{" + stylesheetContent;
  905. }
  906. return stylesheetContent;
  907. }
  908. static async processAttribute(resourceElements, attributeName, baseURI, batchRequest) {
  909. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  910. let resourceURL = resourceElement.getAttribute(attributeName);
  911. if (resourceURL) {
  912. resourceURL = DomUtil.normalizeURL(resourceURL);
  913. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  914. try {
  915. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  916. resourceElement.setAttribute(attributeName, dataURI);
  917. } catch (error) {
  918. /* ignored */
  919. }
  920. }
  921. }
  922. }));
  923. }
  924. static async processXLinks(resourceElements, baseURI, batchRequest) {
  925. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  926. const originalResourceURL = resourceElement.getAttribute("xlink:href");
  927. if (originalResourceURL) {
  928. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  929. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  930. try {
  931. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  932. const DOMParser = DOM.getParser();
  933. if (DOMParser) {
  934. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  935. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  936. if (hashMatch && hashMatch[0]) {
  937. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  938. if (symbolElement) {
  939. resourceElement.setAttribute("xlink:href", hashMatch[0]);
  940. resourceElement.parentElement.appendChild(symbolElement);
  941. }
  942. } else {
  943. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  944. }
  945. } else {
  946. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  947. }
  948. } catch (error) {
  949. /* ignored */
  950. }
  951. }
  952. }
  953. }));
  954. }
  955. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  956. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  957. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  958. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  959. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  960. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  961. try {
  962. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  963. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  964. } catch (error) {
  965. /* ignored */
  966. }
  967. }
  968. }));
  969. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  970. }));
  971. }
  972. }
  973. // -------
  974. // DomUtil
  975. // -------
  976. const DATA_URI_PREFIX = "data:";
  977. const BLOB_URI_PREFIX = "blob:";
  978. const HTTP_URI_PREFIX = /^https?:\/\//;
  979. const ABOUT_BLANK_URI = "about:blank";
  980. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  981. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  982. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  983. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  984. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  985. 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;
  986. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  987. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  988. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  989. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  990. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  991. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  992. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  993. class DomUtil {
  994. static normalizeURL(url) {
  995. if (url.startsWith(DATA_URI_PREFIX)) {
  996. return url;
  997. } else {
  998. return url.split("#")[0];
  999. }
  1000. }
  1001. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1002. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1003. if (template.match(replaceRegExp)) {
  1004. let value = await valueGetter();
  1005. if (!dontReplaceSlash) {
  1006. value = value.replace(/\/+/g, "_");
  1007. }
  1008. return template.replace(replaceRegExp, value);
  1009. }
  1010. return template;
  1011. }
  1012. static getLastSegment(url) {
  1013. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1014. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1015. if (!lastSegment) {
  1016. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1017. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1018. }
  1019. if (!lastSegment) {
  1020. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1021. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1022. }
  1023. if (!lastSegment) {
  1024. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1025. }
  1026. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1027. return lastSegment;
  1028. }
  1029. static getRegExp(string) {
  1030. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1031. }
  1032. static getUrlFunctions(stylesheetContent) {
  1033. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1034. }
  1035. static getImportFunctions(stylesheetContent) {
  1036. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1037. }
  1038. static matchURL(stylesheetContent) {
  1039. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1040. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1041. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1042. return match && match[1];
  1043. }
  1044. static testValidPath(resourceURL) {
  1045. 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));
  1046. }
  1047. static matchImport(stylesheetContent) {
  1048. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1049. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1050. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1051. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1052. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1053. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1054. if (match) {
  1055. const [, resourceURL, media] = match;
  1056. return { resourceURL, media };
  1057. }
  1058. }
  1059. static removeCssComments(stylesheetContent) {
  1060. let start, end;
  1061. do {
  1062. start = stylesheetContent.indexOf("/*");
  1063. end = stylesheetContent.indexOf("*/", start + 2);
  1064. if (start != -1 && end != -1) {
  1065. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1066. }
  1067. } while (start != -1 && end != -1);
  1068. return stylesheetContent;
  1069. }
  1070. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1071. if (mediaQuery) {
  1072. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1073. } else {
  1074. return stylesheetContent;
  1075. }
  1076. }
  1077. }
  1078. // -----
  1079. // Stats
  1080. // -----
  1081. const STATS_DEFAULT_VALUES = {
  1082. discarded: {
  1083. "HTML bytes": 0,
  1084. "hidden elements": 0,
  1085. "HTML imports": 0,
  1086. scripts: 0,
  1087. objects: 0,
  1088. "audio sources": 0,
  1089. "video sources": 0,
  1090. frames: 0,
  1091. "CSS rules": 0,
  1092. canvas: 0,
  1093. stylesheets: 0,
  1094. resources: 0,
  1095. medias: 0
  1096. },
  1097. processed: {
  1098. "HTML bytes": 0,
  1099. "hidden elements": 0,
  1100. "HTML imports": 0,
  1101. scripts: 0,
  1102. objects: 0,
  1103. "audio sources": 0,
  1104. "video sources": 0,
  1105. frames: 0,
  1106. "CSS rules": 0,
  1107. canvas: 0,
  1108. stylesheets: 0,
  1109. resources: 0,
  1110. medias: 0
  1111. }
  1112. };
  1113. class Stats {
  1114. constructor(options) {
  1115. this.options = options;
  1116. if (options.displayStats) {
  1117. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1118. }
  1119. }
  1120. set(type, subType, value) {
  1121. if (this.options.displayStats) {
  1122. this.data[type][subType] = value;
  1123. }
  1124. }
  1125. add(type, subType, value) {
  1126. if (this.options.displayStats) {
  1127. this.data[type][subType] += value;
  1128. }
  1129. }
  1130. addAll(pageData) {
  1131. if (this.options.displayStats) {
  1132. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1133. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1134. }
  1135. }
  1136. }
  1137. return { getClass };
  1138. })();