single-file-core.js 55 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316
  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. const DEBUG = false;
  24. let Download, DOM, URL, sessionId = 0;
  25. function getClass(...args) {
  26. [Download, DOM, URL] = args;
  27. return class {
  28. constructor(options) {
  29. this.options = options;
  30. options.sessionId = sessionId;
  31. sessionId++;
  32. this.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  33. this.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  34. }
  35. async initialize() {
  36. this.processor = new PageProcessor(this.options);
  37. await this.processor.loadPage();
  38. await this.processor.initialize();
  39. }
  40. async preparePageData() {
  41. await this.processor.preparePageData();
  42. }
  43. async getPageData() {
  44. return this.processor.getPageData();
  45. }
  46. };
  47. }
  48. // -------------
  49. // ProgressEvent
  50. // -------------
  51. const PAGE_LOADING = "page-loading";
  52. const PAGE_LOADED = "page-loaded";
  53. const RESOURCES_INITIALIZING = "resource-initializing";
  54. const RESOURCES_INITIALIZED = "resources-initialized";
  55. const RESOURCE_LOADED = "resource-loaded";
  56. const PAGE_ENDED = "page-ended";
  57. class ProgressEvent {
  58. constructor(type, details) {
  59. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  60. }
  61. }
  62. // -------------
  63. // PageProcessor
  64. // -------------
  65. const STAGES = [{
  66. sequential: [
  67. { action: "removeUIElements" },
  68. { action: "replaceStyleContents" },
  69. { option: "removeVideoSrc", action: "insertVideoPosters" },
  70. { option: "removeSrcSet", action: "removeSrcSet" },
  71. { option: "removeFrames", action: "removeFrames" },
  72. { option: "removeImports", action: "removeImports" },
  73. { option: "removeScripts", action: "removeScripts" },
  74. { action: "removeDiscardedResources" },
  75. { action: "resetCharsetMeta" },
  76. { action: "setInputValues" },
  77. { option: "insertFaviconLink", action: "insertFaviconLink" },
  78. { action: "resolveHrefs" },
  79. { action: "replaceCanvasElements" },
  80. { action: "insertFonts" },
  81. { option: "removeHiddenElements", action: "removeHiddenElements" }
  82. ],
  83. parallel: [
  84. { action: "inlineStylesheets" },
  85. { action: "linkStylesheets" },
  86. { action: "attributeStyles" },
  87. { option: "!removeFrames", action: "frames" },
  88. { option: "!removeImports", action: "htmlImports" }
  89. ]
  90. }, {
  91. sequential: [
  92. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  93. { option: "groupDuplicateImages", action: "groupDuplicateImages" },
  94. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  95. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  96. ],
  97. parallel: [
  98. { action: "inlineStylesheets" },
  99. { action: "attributeStyles" },
  100. { action: "pageResources" },
  101. { option: "!removeScripts", action: "scripts" }
  102. ]
  103. }, {
  104. sequential: [
  105. { option: "groupDuplicateImages", action: "setPreservedAspectRatios" },
  106. { option: "lazyLoadImages", action: "lazyLoadImages" },
  107. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  108. { option: "compressCSS", action: "compressCSS" }
  109. ],
  110. parallel: [
  111. { option: "!removeFrames", action: "frames" },
  112. { option: "!removeImports", action: "htmlImports" },
  113. ]
  114. }, {
  115. sequential: [
  116. { option: "compressHTML", action: "compressHTML" },
  117. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  118. { action: "removeDefaultHeadTags" }
  119. ]
  120. }];
  121. class PageProcessor {
  122. constructor(options) {
  123. this.options = options;
  124. this.options.url = this.options.url || this.options.doc.location.href;
  125. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  126. this.batchRequest = new BatchRequest();
  127. this.processor = new DOMProcessor(options, this.batchRequest);
  128. if (this.options.doc) {
  129. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  130. this.options.canvasData = docData.canvasData;
  131. this.options.fontsData = docData.fontsData;
  132. this.options.stylesheetContents = docData.stylesheetContents;
  133. this.options.responsiveImageData = docData.responsiveImageData;
  134. this.options.imageData = docData.imageData;
  135. this.options.postersData = docData.postersData;
  136. }
  137. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  138. this.onprogress = options.onprogress || (() => { });
  139. }
  140. async loadPage() {
  141. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  142. await this.processor.loadPage(this.options.content);
  143. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  144. }
  145. async initialize() {
  146. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  147. await this.executeStage(0);
  148. this.pendingPromises = this.executeStage(1);
  149. if (this.options.doc) {
  150. DOM.postProcessDoc(this.options.doc, this.options);
  151. this.options.doc = null;
  152. this.options.win = null;
  153. }
  154. }
  155. async preparePageData() {
  156. if (!this.options.windowId) {
  157. this.processor.initialize(this.batchRequest);
  158. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  159. }
  160. await this.batchRequest.run(details => {
  161. details.pageURL = this.options.url;
  162. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  163. }, this.options);
  164. await this.pendingPromises;
  165. await this.executeStage(2);
  166. await this.executeStage(3);
  167. }
  168. async getPageData() {
  169. if (!this.options.windowId) {
  170. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  171. }
  172. return this.processor.getPageData();
  173. }
  174. async executeStage(step) {
  175. if (DEBUG) {
  176. log("**** STARTED STAGE", step, "****");
  177. }
  178. STAGES[step].sequential.forEach(task => {
  179. let startTime;
  180. if (DEBUG) {
  181. startTime = Date.now();
  182. log(" -- STARTED task =", task.action);
  183. }
  184. this.executeTask(task, !step);
  185. if (DEBUG) {
  186. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  187. }
  188. });
  189. if (STAGES[step].parallel) {
  190. return await Promise.all(STAGES[step].parallel.map(task => {
  191. let startTime;
  192. if (DEBUG) {
  193. startTime = Date.now();
  194. log(" // STARTED task =", task.action);
  195. }
  196. const promise = this.executeTask(task, !step);
  197. if (DEBUG) {
  198. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  199. }
  200. return promise;
  201. }));
  202. }
  203. if (DEBUG) {
  204. log("**** ENDED STAGE", step, "****");
  205. }
  206. }
  207. executeTask(task, initialization) {
  208. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  209. return this.processor[task.action](initialization);
  210. }
  211. }
  212. }
  213. // --------
  214. // BatchRequest
  215. // --------
  216. class BatchRequest {
  217. constructor() {
  218. this.requests = new Map();
  219. }
  220. async addURL(resourceURL, asDataURI = true) {
  221. return new Promise((resolve, reject) => {
  222. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  223. const resourceRequests = this.requests.get(requestKey);
  224. if (resourceRequests) {
  225. resourceRequests.push({ resolve, reject });
  226. } else {
  227. this.requests.set(requestKey, [{ resolve, reject }]);
  228. }
  229. });
  230. }
  231. getMaxResources() {
  232. return Array.from(this.requests.keys()).length;
  233. }
  234. async run(onloadListener, options) {
  235. const resourceURLs = Array.from(this.requests.keys());
  236. let indexResource = 0;
  237. return Promise.all(resourceURLs.map(async requestKey => {
  238. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  239. const resourceRequests = this.requests.get(requestKey);
  240. try {
  241. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  242. indexResource = indexResource + 1;
  243. onloadListener({ index: indexResource, url: resourceURL });
  244. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  245. } catch (error) {
  246. indexResource = indexResource + 1;
  247. onloadListener({ index: indexResource, url: resourceURL });
  248. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  249. }
  250. this.requests.delete(requestKey);
  251. }));
  252. }
  253. }
  254. // ------------
  255. // DOMProcessor
  256. // ------------
  257. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  258. const EMPTY_DATA_URI = "data:base64,";
  259. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  260. class DOMProcessor {
  261. constructor(options, batchRequest) {
  262. this.options = options;
  263. this.stats = new Stats(options);
  264. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  265. this.batchRequest = batchRequest;
  266. }
  267. initialize() {
  268. this.maxResources = this.batchRequest.getMaxResources();
  269. if (!this.options.removeFrames) {
  270. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  271. }
  272. this.stats.set("processed", "resources", this.maxResources);
  273. }
  274. async loadPage(pageContent) {
  275. if (!pageContent || this.options.saveRawPage) {
  276. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  277. }
  278. this.doc = DOM.createDoc(pageContent, this.baseURI);
  279. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  280. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  281. await DOMProcessor.loadEscapedFragmentPage();
  282. }
  283. }
  284. async loadEscapedFragmentPage() {
  285. if (this.baseURI.includes("?")) {
  286. this.baseURI += "&";
  287. } else {
  288. this.baseURI += "?";
  289. }
  290. this.baseURI += ESCAPED_FRAGMENT;
  291. await this.loadPage();
  292. }
  293. async getPageData() {
  294. DOM.postProcessDoc(this.doc, this.options);
  295. if (this.options.selected) {
  296. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  297. if (rootElement) {
  298. DomProcessorHelper.isolateElements(rootElement);
  299. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  300. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  301. }
  302. }
  303. const titleElement = this.doc.querySelector("title");
  304. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  305. this.options.info = {};
  306. const descriptionElement = this.doc.querySelector("meta[name=description]");
  307. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  308. this.options.info.lang = this.doc.documentElement.lang;
  309. const authorElement = this.doc.querySelector("meta[name=author]");
  310. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  311. const creatorElement = this.doc.querySelector("meta[name=creator]");
  312. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  313. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  314. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  315. const url = new URL(this.baseURI);
  316. let size;
  317. if (this.options.displayStats) {
  318. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  319. }
  320. const content = DOM.serialize(this.doc, this.options.compressHTML);
  321. if (this.options.displayStats) {
  322. const contentSize = DOM.getContentSize(content);
  323. this.stats.set("processed", "HTML bytes", contentSize);
  324. this.stats.add("discarded", "HTML bytes", size - contentSize);
  325. }
  326. const filename = await DomProcessorHelper.getFilename(this.options, content);
  327. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  328. return {
  329. stats: this.stats.data,
  330. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  331. filename,
  332. content
  333. };
  334. }
  335. setInputValues() {
  336. this.doc.querySelectorAll("input").forEach(input => {
  337. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  338. if (value) {
  339. input.setAttribute("value", value);
  340. }
  341. });
  342. this.doc.querySelectorAll("textarea").forEach(textarea => {
  343. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  344. if (value) {
  345. textarea.textContent = value;
  346. }
  347. });
  348. this.doc.querySelectorAll("select").forEach(select => {
  349. select.querySelectorAll("option").forEach(option => {
  350. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  351. if (selected) {
  352. option.setAttribute("selected", "");
  353. }
  354. });
  355. });
  356. }
  357. lazyLoadImages() {
  358. DOM.lazyLoad(this.doc);
  359. }
  360. removeDiscardedResources() {
  361. 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\"])");
  362. this.stats.set("discarded", "objects", objectElements.length);
  363. this.stats.set("processed", "objects", objectElements.length);
  364. objectElements.forEach(element => element.remove());
  365. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  366. replacedAttributeValue.forEach(element => {
  367. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  368. if (relValue.length) {
  369. element.setAttribute("rel", relValue);
  370. } else {
  371. element.remove();
  372. }
  373. });
  374. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  375. if (this.options.compressHTML) {
  376. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  377. }
  378. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  379. if (this.options.removeScripts) {
  380. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  381. this.doc.querySelectorAll("[href]").forEach(element => {
  382. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  383. element.removeAttribute("href");
  384. }
  385. });
  386. this.doc.querySelectorAll("[src]").forEach(element => {
  387. if (element.src && element.src.match(/^\s*javascript:/)) {
  388. element.removeAttribute("src");
  389. }
  390. });
  391. }
  392. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  393. this.stats.set("processed", "audio sources", audioSourceElements.length);
  394. if (this.options.removeAudioSrc) {
  395. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  396. audioSourceElements.forEach(element => element.removeAttribute("src"));
  397. }
  398. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  399. this.stats.set("processed", "video sources", videoSourceElements.length);
  400. if (this.options.removeVideoSrc) {
  401. this.stats.set("discarded", "video sources", videoSourceElements.length);
  402. videoSourceElements.forEach(element => element.removeAttribute("src"));
  403. }
  404. }
  405. removeDefaultHeadTags() {
  406. this.doc.querySelectorAll("base").forEach(element => element.remove());
  407. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  408. this.doc.head.querySelector("meta[charset]").remove();
  409. }
  410. }
  411. removeUIElements() {
  412. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  413. }
  414. removeScripts() {
  415. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  416. this.stats.set("discarded", "scripts", scriptElements.length);
  417. this.stats.set("processed", "scripts", scriptElements.length);
  418. scriptElements.forEach(element => element.remove());
  419. }
  420. removeFrames() {
  421. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  422. this.stats.set("discarded", "frames", frameElements.length);
  423. this.stats.set("processed", "frames", frameElements.length);
  424. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  425. }
  426. removeImports() {
  427. const importElements = this.doc.querySelectorAll("link[rel=import]");
  428. this.stats.set("discarded", "HTML imports", importElements.length);
  429. this.stats.set("processed", "HTML imports", importElements.length);
  430. importElements.forEach(element => element.remove());
  431. }
  432. resetCharsetMeta() {
  433. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  434. const metaElement = this.doc.createElement("meta");
  435. metaElement.setAttribute("charset", "utf-8");
  436. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  437. }
  438. insertFaviconLink() {
  439. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  440. if (!faviconElement) {
  441. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  442. }
  443. if (!faviconElement) {
  444. faviconElement = this.doc.createElement("link");
  445. faviconElement.setAttribute("type", "image/x-icon");
  446. faviconElement.setAttribute("rel", "shortcut icon");
  447. faviconElement.setAttribute("href", "/favicon.ico");
  448. }
  449. this.doc.head.appendChild(faviconElement);
  450. }
  451. resolveHrefs() {
  452. this.doc.querySelectorAll("[href]").forEach(element => {
  453. if (element.href) {
  454. const href = element.href.baseVal ? element.href.baseVal : element.href;
  455. const normalizedHref = DomUtil.normalizeURL(href);
  456. if (!normalizedHref || normalizedHref != this.baseURI) {
  457. element.setAttribute("href", href);
  458. }
  459. }
  460. });
  461. }
  462. removeUnusedStyles() {
  463. if (!this.mediaAllInfo) {
  464. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  465. }
  466. const stats = DOM.minifyCSS(this.doc, this.mediaAllInfo);
  467. this.stats.set("processed", "CSS rules", stats.processed);
  468. this.stats.set("discarded", "CSS rules", stats.discarded);
  469. }
  470. removeAlternativeFonts() {
  471. DOM.minifyFonts(this.doc);
  472. }
  473. removeSrcSet() {
  474. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  475. const tagName = element.tagName.toLowerCase();
  476. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  477. const responsiveImageData = this.options.responsiveImageData[Number(element.getAttribute(dataAttributeName))];
  478. element.removeAttribute(dataAttributeName);
  479. if (responsiveImageData) {
  480. if (tagName == "img") {
  481. if (responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  482. element.removeAttribute("srcset");
  483. element.removeAttribute("sizes");
  484. element.src = responsiveImageData.source.src;
  485. }
  486. }
  487. if (tagName == "picture") {
  488. const imageElement = element.querySelector("img");
  489. if (responsiveImageData.source && responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  490. imageElement.removeAttribute("srcset");
  491. imageElement.removeAttribute("sizes");
  492. imageElement.src = responsiveImageData.source.src;
  493. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  494. } else {
  495. if (responsiveImageData.sources) {
  496. element.querySelectorAll("source").forEach(sourceElement => {
  497. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  498. sourceElement.remove();
  499. }
  500. });
  501. const sourceElements = element.querySelectorAll("source");
  502. if (sourceElements.length) {
  503. const lastSourceElement = sourceElements[sourceElements.length - 1];
  504. if (imageElement) {
  505. if (lastSourceElement.src) {
  506. imageElement.src = lastSourceElement.src;
  507. } else {
  508. imageElement.removeAttribute("src");
  509. }
  510. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  511. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  512. } else {
  513. imageElement.removeAttribute("srcset");
  514. }
  515. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  516. }
  517. }
  518. }
  519. }
  520. }
  521. }
  522. });
  523. }
  524. postRemoveAlternativeFonts() {
  525. DOM.minifyFonts(this.doc, true);
  526. }
  527. removeHiddenElements() {
  528. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  529. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  530. this.stats.set("processed", "hidden elements", hiddenElements.length);
  531. hiddenElements.forEach(element => element.remove());
  532. }
  533. compressHTML() {
  534. let size;
  535. if (this.options.displayStats) {
  536. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  537. }
  538. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  539. if (this.options.displayStats) {
  540. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  541. }
  542. }
  543. groupDuplicateImages() {
  544. if (this.options.imageData) {
  545. if (!this.mediaAllInfo) {
  546. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  547. }
  548. DOM.minifyImages(this.doc, this.mediaAllInfo, this.options);
  549. }
  550. }
  551. setPreservedAspectRatios() {
  552. DOM.setPreservedAspectRatios(this.doc);
  553. }
  554. removeAlternativeMedias() {
  555. const stats = DOM.minifyMedias(this.doc);
  556. this.stats.set("processed", "medias", stats.processed);
  557. this.stats.set("discarded", "medias", stats.discarded);
  558. }
  559. compressCSS() {
  560. this.doc.querySelectorAll("style").forEach(styleElement => {
  561. if (styleElement) {
  562. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  563. }
  564. });
  565. this.doc.querySelectorAll("[style]").forEach(element => {
  566. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  567. });
  568. }
  569. insertSingleFileComment() {
  570. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  571. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  572. }
  573. replaceCanvasElements() {
  574. if (this.options.canvasData) {
  575. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  576. const canvasData = this.options.canvasData[indexCanvasElement];
  577. if (canvasData) {
  578. const imgElement = this.doc.createElement("img");
  579. imgElement.setAttribute("src", canvasData.dataURI);
  580. Array.from(canvasElement.attributes).forEach(attribute => {
  581. if (attribute.value) {
  582. imgElement.setAttribute(attribute.name, attribute.value);
  583. }
  584. });
  585. if (!imgElement.width && canvasData.width) {
  586. imgElement.style.pixelWidth = canvasData.width;
  587. }
  588. if (!imgElement.height && canvasData.height) {
  589. imgElement.style.pixelHeight = canvasData.height;
  590. }
  591. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  592. this.stats.add("processed", "canvas", 1);
  593. }
  594. });
  595. }
  596. }
  597. insertFonts() {
  598. if (this.options.fontsData && this.options.fontsData.length) {
  599. let stylesheetContent = "";
  600. this.options.fontsData.forEach(fontStyles => {
  601. if (fontStyles["font-family"] && fontStyles.src) {
  602. stylesheetContent += "@font-face{";
  603. let stylesContent = "";
  604. Object.keys(fontStyles).forEach(fontStyle => {
  605. if (stylesContent) {
  606. stylesContent += ";";
  607. }
  608. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  609. });
  610. stylesheetContent += stylesContent + "}";
  611. }
  612. });
  613. if (stylesheetContent) {
  614. const styleElement = this.doc.createElement("style");
  615. styleElement.textContent = stylesheetContent;
  616. const existingStyleElement = this.doc.querySelector("style");
  617. if (existingStyleElement) {
  618. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  619. } else {
  620. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  621. }
  622. }
  623. }
  624. }
  625. replaceStyleContents() {
  626. if (this.options.stylesheetContents) {
  627. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  628. if (this.options.stylesheetContents[styleIndex]) {
  629. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  630. }
  631. });
  632. }
  633. }
  634. insertVideoPosters() {
  635. if (this.options.postersData) {
  636. this.doc.querySelectorAll("video[src], video > source[src]").forEach((videoElement, videoIndex) => {
  637. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  638. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  639. }
  640. });
  641. }
  642. }
  643. async pageResources() {
  644. const resourcePromises = [
  645. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  646. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest),
  647. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image]"), "src", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  648. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest),
  649. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  650. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  651. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  652. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  653. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)
  654. ];
  655. if (!this.options.removeAudioSrc) {
  656. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", PREFIX_DATA_URI_AUDIO, this.baseURI, this.batchRequest));
  657. }
  658. if (!this.options.removeVideoSrc) {
  659. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", PREFIX_DATA_URI_VIDEO, this.baseURI, this.batchRequest));
  660. }
  661. if (this.options.lazyLoadImages) {
  662. const imageSelectors = DOM.lazyLoaderImageSelectors();
  663. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)));
  664. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)));
  665. }
  666. await resourcePromises;
  667. }
  668. async inlineStylesheets(initialization) {
  669. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  670. if (!initialization) {
  671. this.stats.add("processed", "stylesheets", 1);
  672. }
  673. let stylesheetContent = styleElement.textContent;
  674. if (initialization) {
  675. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  676. } else {
  677. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  678. }
  679. styleElement.textContent = stylesheetContent;
  680. }));
  681. }
  682. async scripts() {
  683. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  684. if (scriptElement.src) {
  685. this.stats.add("processed", "scripts", 1);
  686. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  687. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  688. }
  689. scriptElement.removeAttribute("src");
  690. }));
  691. }
  692. async frames(initialization) {
  693. if (this.options.framesData) {
  694. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  695. await Promise.all(frameElements.map(async frameElement => {
  696. DomProcessorHelper.setFrameEmptySrc(frameElement);
  697. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  698. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  699. if (frameWindowId) {
  700. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  701. if (frameData) {
  702. if (initialization) {
  703. const options = Object.create(this.options);
  704. options.insertSingleFileComment = false;
  705. options.insertFaviconLink = false;
  706. options.doc = null;
  707. options.win = null;
  708. options.url = frameData.baseURI;
  709. options.windowId = frameWindowId;
  710. if (frameData.content) {
  711. options.content = frameData.content;
  712. options.canvasData = frameData.canvasData;
  713. options.stylesheetContents = frameData.stylesheetContents;
  714. options.currentSrcImages = frameData.currentSrcImages;
  715. options.fontsData = frameData.fontsData;
  716. options.imageData = frameData.imageData;
  717. options.responsiveImageData = frameData.responsiveImageData;
  718. frameData.processor = new PageProcessor(options);
  719. frameData.frameElement = frameElement;
  720. await frameData.processor.loadPage();
  721. await frameData.processor.initialize();
  722. frameData.maxResources = this.batchRequest.getMaxResources();
  723. }
  724. } else {
  725. if (frameData.processor) {
  726. this.stats.add("processed", "frames", 1);
  727. await frameData.processor.preparePageData();
  728. const pageData = await frameData.processor.getPageData();
  729. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  730. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  731. this.stats.addAll(pageData);
  732. } else {
  733. this.stats.add("discarded", "frames", 1);
  734. }
  735. }
  736. }
  737. }
  738. }));
  739. }
  740. }
  741. async htmlImports(initialization) {
  742. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  743. if (!this.relImportProcessors) {
  744. this.relImportProcessors = new Map();
  745. }
  746. await Promise.all(linkElements.map(async linkElement => {
  747. if (initialization) {
  748. const resourceURL = linkElement.href;
  749. const options = Object.create(this.options);
  750. options.insertSingleFileComment = false;
  751. options.insertFaviconLink = false;
  752. options.doc = null;
  753. options.win = null;
  754. options.url = resourceURL;
  755. if (resourceURL) {
  756. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  757. const processor = new PageProcessor(options);
  758. this.relImportProcessors.set(linkElement, processor);
  759. await processor.loadPage();
  760. return processor.initialize();
  761. }
  762. }
  763. } else {
  764. linkElement.setAttribute("href", EMPTY_DATA_URI);
  765. const processor = this.relImportProcessors.get(linkElement);
  766. if (processor) {
  767. this.stats.add("processed", "HTML imports", 1);
  768. this.relImportProcessors.delete(linkElement);
  769. const pageData = await processor.getPageData();
  770. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  771. this.stats.addAll(pageData);
  772. } else {
  773. this.stats.add("discarded", "HTML imports", 1);
  774. }
  775. }
  776. }));
  777. }
  778. async attributeStyles(initialization) {
  779. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  780. let stylesheetContent = element.getAttribute("style");
  781. if (initialization) {
  782. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  783. } else {
  784. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  785. }
  786. element.setAttribute("style", stylesheetContent);
  787. }));
  788. }
  789. async linkStylesheets() {
  790. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  791. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  792. const styleElement = this.doc.createElement("style");
  793. styleElement.textContent = stylesheetContent;
  794. linkElement.parentElement.replaceChild(styleElement, linkElement);
  795. }));
  796. }
  797. }
  798. // ---------
  799. // DomHelper
  800. // ---------
  801. const REGEXP_AMP = /&/g;
  802. const REGEXP_NBSP = /\u00a0/g;
  803. const REGEXP_START_TAG = /</g;
  804. const REGEXP_END_TAG = />/g;
  805. const REGEXP_URL_HASH = /(#.+?)$/;
  806. const PREFIX_DATA_URI_IMAGE = "data:image/";
  807. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  808. const PREFIX_DATA_URI_VIDEO = "data:video/";
  809. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  810. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  811. const PREFIX_DATA_URI_OCTET_STREAM = "data:application/octet-stream";
  812. class DomProcessorHelper {
  813. static async getFilename(options, content) {
  814. let filename = options.filenameTemplate;
  815. const date = new Date();
  816. const url = new URL(options.url);
  817. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  818. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  819. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  820. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  821. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  822. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  823. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  824. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  825. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  826. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  827. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  828. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  829. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  830. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  831. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  832. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  833. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  834. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  835. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  836. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  837. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  838. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  839. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  840. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  841. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  842. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  843. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  844. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  845. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  846. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  847. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  848. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  849. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  850. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  851. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  852. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  853. if (!options.backgroundSave) {
  854. filename = filename.replace(/\//g, "_");
  855. }
  856. if (filename.length > 192) {
  857. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  858. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  859. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  860. }
  861. if (!filename) {
  862. filename = "Unnamed page";
  863. }
  864. return filename;
  865. }
  866. static setFrameEmptySrc(frameElement) {
  867. if (frameElement.tagName == "OBJECT") {
  868. frameElement.setAttribute("data", "data:text/html,");
  869. } else {
  870. frameElement.setAttribute("srcdoc", "");
  871. frameElement.removeAttribute("src");
  872. }
  873. }
  874. static setFrameContent(frameElement, content) {
  875. if (frameElement.tagName == "OBJECT") {
  876. frameElement.setAttribute("data", "data:text/html," + content);
  877. } else {
  878. frameElement.setAttribute("srcdoc", content);
  879. frameElement.removeAttribute("src");
  880. }
  881. }
  882. static isolateElements(rootElement) {
  883. rootElement.querySelectorAll("*").forEach(element => {
  884. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  885. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  886. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  887. element.remove();
  888. }
  889. });
  890. isolateParentElements(rootElement.parentElement, rootElement);
  891. function isolateParentElements(parentElement, element) {
  892. if (parentElement) {
  893. Array.from(parentElement.childNodes).forEach(node => {
  894. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  895. node.remove();
  896. }
  897. });
  898. }
  899. element = element.parentElement;
  900. if (element && element.parentElement) {
  901. isolateParentElements(element.parentElement, element);
  902. }
  903. }
  904. }
  905. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  906. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  907. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  908. const imports = DomUtil.getImportFunctions(stylesheetContent);
  909. await Promise.all(imports.map(async cssImport => {
  910. const match = DomUtil.matchImport(cssImport);
  911. if (match) {
  912. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  913. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  914. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  915. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  916. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  917. if (stylesheetContent.includes(cssImport)) {
  918. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  919. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  920. }
  921. }
  922. }
  923. }));
  924. return stylesheetContent;
  925. }
  926. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  927. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  928. urlFunctions.map(urlFunction => {
  929. const originalResourceURL = DomUtil.matchURL(urlFunction);
  930. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  931. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  932. const resolvedURL = new URL(resourceURL, baseURI).href;
  933. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  934. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  935. }
  936. } else {
  937. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  938. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  939. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  940. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  941. }
  942. }
  943. }
  944. });
  945. return stylesheetContent;
  946. }
  947. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  948. resourceURL = DomUtil.normalizeURL(resourceURL);
  949. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  950. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  951. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  952. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  953. return stylesheetContent;
  954. }
  955. }
  956. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  957. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  958. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  959. let indexVariable = 0;
  960. await Promise.all(urlFunctions.map(async urlFunction => {
  961. const originalResourceURL = DomUtil.matchURL(urlFunction);
  962. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  963. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  964. const dataURI = await batchRequest.addURL(resourceURL);
  965. const regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  966. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  967. const functions = stylesheetContent.match(regExpUrlFunction);
  968. if (options.groupDuplicateImages && functions && functions.length > 1) {
  969. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  970. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  971. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + variableName + ")");
  972. indexVariable++;
  973. } else {
  974. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  975. }
  976. } else {
  977. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  978. }
  979. }
  980. }));
  981. if (indexVariable) {
  982. stylesheetContent = ":root{" + stylesheetContent;
  983. }
  984. return stylesheetContent;
  985. }
  986. static async processAttribute(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  987. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  988. let resourceURL = resourceElement.getAttribute(attributeName);
  989. if (resourceURL) {
  990. resourceURL = DomUtil.normalizeURL(resourceURL);
  991. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  992. try {
  993. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  994. if (dataURI.startsWith(prefixDataURI) || dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) || dataURI.startsWith(PREFIX_DATA_URI_OCTET_STREAM)) {
  995. resourceElement.setAttribute(attributeName, dataURI);
  996. } else {
  997. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  998. }
  999. } catch (error) {
  1000. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1001. }
  1002. }
  1003. }
  1004. }));
  1005. }
  1006. static async processXLinks(resourceElements, baseURI, batchRequest) {
  1007. const attributeName = "xlink:href";
  1008. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1009. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1010. if (originalResourceURL) {
  1011. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1012. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1013. try {
  1014. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  1015. const DOMParser = DOM.getParser();
  1016. if (DOMParser) {
  1017. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  1018. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1019. if (hashMatch && hashMatch[0]) {
  1020. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1021. if (symbolElement) {
  1022. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1023. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1024. }
  1025. } else {
  1026. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1027. }
  1028. } else {
  1029. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1030. }
  1031. } catch (error) {
  1032. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1033. }
  1034. }
  1035. }
  1036. }));
  1037. }
  1038. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  1039. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1040. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1041. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1042. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1043. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1044. try {
  1045. let dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  1046. if (!dataURI.startsWith(prefixDataURI) && !dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) && !dataURI.startsWith(PREFIX_DATA_URI_OCTET_STREAM)) {
  1047. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1048. }
  1049. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1050. } catch (error) {
  1051. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1052. }
  1053. }
  1054. }));
  1055. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1056. }));
  1057. }
  1058. }
  1059. // -------
  1060. // DomUtil
  1061. // -------
  1062. const DATA_URI_PREFIX = "data:";
  1063. const BLOB_URI_PREFIX = "blob:";
  1064. const HTTP_URI_PREFIX = /^https?:\/\//;
  1065. const ABOUT_BLANK_URI = "about:blank";
  1066. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1067. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1068. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1069. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1070. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1071. 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;
  1072. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1073. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1074. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1075. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1076. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1077. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1078. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1079. class DomUtil {
  1080. static normalizeURL(url) {
  1081. if (url.startsWith(DATA_URI_PREFIX)) {
  1082. return url;
  1083. } else {
  1084. return url.split("#")[0];
  1085. }
  1086. }
  1087. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1088. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1089. if (template.match(replaceRegExp)) {
  1090. let value = await valueGetter();
  1091. if (!dontReplaceSlash) {
  1092. value = value.replace(/\/+/g, "_");
  1093. }
  1094. return template.replace(replaceRegExp, value);
  1095. }
  1096. return template;
  1097. }
  1098. static getLastSegment(url) {
  1099. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1100. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1101. if (!lastSegment) {
  1102. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1103. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1104. }
  1105. if (!lastSegment) {
  1106. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1107. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1108. }
  1109. if (!lastSegment) {
  1110. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1111. }
  1112. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1113. return lastSegment;
  1114. }
  1115. static getRegExp(string) {
  1116. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1117. }
  1118. static getUrlFunctions(stylesheetContent) {
  1119. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1120. }
  1121. static getImportFunctions(stylesheetContent) {
  1122. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1123. }
  1124. static matchURL(stylesheetContent) {
  1125. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1126. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1127. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1128. return match && match[1];
  1129. }
  1130. static testValidPath(resourceURL) {
  1131. 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));
  1132. }
  1133. static matchImport(stylesheetContent) {
  1134. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1135. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1136. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1137. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1138. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1139. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1140. if (match) {
  1141. const [, resourceURL, media] = match;
  1142. return { resourceURL, media };
  1143. }
  1144. }
  1145. static removeCssComments(stylesheetContent) {
  1146. let start, end;
  1147. do {
  1148. start = stylesheetContent.indexOf("/*");
  1149. end = stylesheetContent.indexOf("*/", start + 2);
  1150. if (start != -1 && end != -1) {
  1151. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1152. }
  1153. } while (start != -1 && end != -1);
  1154. return stylesheetContent;
  1155. }
  1156. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1157. if (mediaQuery) {
  1158. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1159. } else {
  1160. return stylesheetContent;
  1161. }
  1162. }
  1163. }
  1164. function log(...args) {
  1165. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1166. }
  1167. // -----
  1168. // Stats
  1169. // -----
  1170. const STATS_DEFAULT_VALUES = {
  1171. discarded: {
  1172. "HTML bytes": 0,
  1173. "hidden elements": 0,
  1174. "HTML imports": 0,
  1175. scripts: 0,
  1176. objects: 0,
  1177. "audio sources": 0,
  1178. "video sources": 0,
  1179. frames: 0,
  1180. "CSS rules": 0,
  1181. canvas: 0,
  1182. stylesheets: 0,
  1183. resources: 0,
  1184. medias: 0
  1185. },
  1186. processed: {
  1187. "HTML bytes": 0,
  1188. "hidden elements": 0,
  1189. "HTML imports": 0,
  1190. scripts: 0,
  1191. objects: 0,
  1192. "audio sources": 0,
  1193. "video sources": 0,
  1194. frames: 0,
  1195. "CSS rules": 0,
  1196. canvas: 0,
  1197. stylesheets: 0,
  1198. resources: 0,
  1199. medias: 0
  1200. }
  1201. };
  1202. class Stats {
  1203. constructor(options) {
  1204. this.options = options;
  1205. if (options.displayStats) {
  1206. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1207. }
  1208. }
  1209. set(type, subType, value) {
  1210. if (this.options.displayStats) {
  1211. this.data[type][subType] = value;
  1212. }
  1213. }
  1214. add(type, subType, value) {
  1215. if (this.options.displayStats) {
  1216. this.data[type][subType] += value;
  1217. }
  1218. }
  1219. addAll(pageData) {
  1220. if (this.options.displayStats) {
  1221. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1222. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1223. }
  1224. }
  1225. }
  1226. return { getClass };
  1227. })();