single-file-core.js 54 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307
  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: "lazyLoadImages", action: "lazyLoadImages" },
  106. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  107. { option: "compressCSS", action: "compressCSS" }
  108. ],
  109. parallel: [
  110. { option: "!removeFrames", action: "frames" },
  111. { option: "!removeImports", action: "htmlImports" },
  112. ]
  113. }, {
  114. sequential: [
  115. { option: "compressHTML", action: "compressHTML" },
  116. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  117. { action: "removeDefaultHeadTags" }
  118. ]
  119. }];
  120. class PageProcessor {
  121. constructor(options) {
  122. this.options = options;
  123. this.options.url = this.options.url || this.options.doc.location.href;
  124. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  125. this.batchRequest = new BatchRequest();
  126. this.processor = new DOMProcessor(options, this.batchRequest);
  127. if (this.options.doc) {
  128. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  129. this.options.canvasData = docData.canvasData;
  130. this.options.fontsData = docData.fontsData;
  131. this.options.stylesheetContents = docData.stylesheetContents;
  132. this.options.responsiveImageData = docData.responsiveImageData;
  133. this.options.imageData = docData.imageData;
  134. this.options.postersData = docData.postersData;
  135. }
  136. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  137. this.onprogress = options.onprogress || (() => { });
  138. }
  139. async loadPage() {
  140. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  141. await this.processor.loadPage(this.options.content);
  142. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  143. }
  144. async initialize() {
  145. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  146. await this.executeStage(0);
  147. this.pendingPromises = this.executeStage(1);
  148. if (this.options.doc) {
  149. DOM.postProcessDoc(this.options.doc, this.options);
  150. this.options.doc = null;
  151. this.options.win = null;
  152. }
  153. }
  154. async preparePageData() {
  155. if (!this.options.windowId) {
  156. this.processor.initialize(this.batchRequest);
  157. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  158. }
  159. await this.batchRequest.run(details => {
  160. details.pageURL = this.options.url;
  161. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  162. }, this.options);
  163. await this.pendingPromises;
  164. await this.executeStage(2);
  165. await this.executeStage(3);
  166. }
  167. async getPageData() {
  168. if (!this.options.windowId) {
  169. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  170. }
  171. return this.processor.getPageData();
  172. }
  173. async executeStage(step) {
  174. if (DEBUG) {
  175. log("**** STARTED STAGE", step, "****");
  176. }
  177. STAGES[step].sequential.forEach(task => {
  178. let startTime;
  179. if (DEBUG) {
  180. startTime = Date.now();
  181. log(" -- STARTED task =", task.action);
  182. }
  183. this.executeTask(task, !step);
  184. if (DEBUG) {
  185. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  186. }
  187. });
  188. if (STAGES[step].parallel) {
  189. return await Promise.all(STAGES[step].parallel.map(task => {
  190. let startTime;
  191. if (DEBUG) {
  192. startTime = Date.now();
  193. log(" // STARTED task =", task.action);
  194. }
  195. const promise = this.executeTask(task, !step);
  196. if (DEBUG) {
  197. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  198. }
  199. return promise;
  200. }));
  201. }
  202. if (DEBUG) {
  203. log("**** ENDED STAGE", step, "****");
  204. }
  205. }
  206. executeTask(task, initialization) {
  207. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  208. return this.processor[task.action](initialization);
  209. }
  210. }
  211. }
  212. // --------
  213. // BatchRequest
  214. // --------
  215. class BatchRequest {
  216. constructor() {
  217. this.requests = new Map();
  218. }
  219. async addURL(resourceURL, asDataURI = true) {
  220. return new Promise((resolve, reject) => {
  221. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  222. const resourceRequests = this.requests.get(requestKey);
  223. if (resourceRequests) {
  224. resourceRequests.push({ resolve, reject });
  225. } else {
  226. this.requests.set(requestKey, [{ resolve, reject }]);
  227. }
  228. });
  229. }
  230. getMaxResources() {
  231. return Array.from(this.requests.keys()).length;
  232. }
  233. async run(onloadListener, options) {
  234. const resourceURLs = Array.from(this.requests.keys());
  235. let indexResource = 0;
  236. return Promise.all(resourceURLs.map(async requestKey => {
  237. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  238. const resourceRequests = this.requests.get(requestKey);
  239. try {
  240. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  241. indexResource = indexResource + 1;
  242. onloadListener({ index: indexResource, url: resourceURL });
  243. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  244. } catch (error) {
  245. indexResource = indexResource + 1;
  246. onloadListener({ index: indexResource, url: resourceURL });
  247. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  248. }
  249. this.requests.delete(requestKey);
  250. }));
  251. }
  252. }
  253. // ------------
  254. // DOMProcessor
  255. // ------------
  256. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  257. const EMPTY_DATA_URI = "data:base64,";
  258. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  259. class DOMProcessor {
  260. constructor(options, batchRequest) {
  261. this.options = options;
  262. this.stats = new Stats(options);
  263. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  264. this.batchRequest = batchRequest;
  265. }
  266. initialize() {
  267. this.maxResources = this.batchRequest.getMaxResources();
  268. if (!this.options.removeFrames) {
  269. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  270. }
  271. this.stats.set("processed", "resources", this.maxResources);
  272. }
  273. async loadPage(pageContent) {
  274. if (!pageContent || this.options.saveRawPage) {
  275. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  276. }
  277. this.doc = DOM.createDoc(pageContent, this.baseURI);
  278. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  279. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  280. await DOMProcessor.loadEscapedFragmentPage();
  281. }
  282. }
  283. async loadEscapedFragmentPage() {
  284. if (this.baseURI.includes("?")) {
  285. this.baseURI += "&";
  286. } else {
  287. this.baseURI += "?";
  288. }
  289. this.baseURI += ESCAPED_FRAGMENT;
  290. await this.loadPage();
  291. }
  292. async getPageData() {
  293. DOM.postProcessDoc(this.doc, this.options);
  294. if (this.options.selected) {
  295. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  296. if (rootElement) {
  297. DomProcessorHelper.isolateElements(rootElement);
  298. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  299. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  300. }
  301. }
  302. const titleElement = this.doc.querySelector("title");
  303. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  304. this.options.info = {};
  305. const descriptionElement = this.doc.querySelector("meta[name=description]");
  306. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  307. this.options.info.lang = this.doc.documentElement.lang;
  308. const authorElement = this.doc.querySelector("meta[name=author]");
  309. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  310. const creatorElement = this.doc.querySelector("meta[name=creator]");
  311. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  312. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  313. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  314. const url = new URL(this.baseURI);
  315. let size;
  316. if (this.options.displayStats) {
  317. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  318. }
  319. const content = DOM.serialize(this.doc, this.options.compressHTML);
  320. if (this.options.displayStats) {
  321. const contentSize = DOM.getContentSize(content);
  322. this.stats.set("processed", "HTML bytes", contentSize);
  323. this.stats.add("discarded", "HTML bytes", size - contentSize);
  324. }
  325. const filename = await DomProcessorHelper.getFilename(this.options, content);
  326. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  327. return {
  328. stats: this.stats.data,
  329. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  330. filename,
  331. content
  332. };
  333. }
  334. setInputValues() {
  335. this.doc.querySelectorAll("input").forEach(input => {
  336. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  337. if (value) {
  338. input.setAttribute("value", value);
  339. }
  340. });
  341. this.doc.querySelectorAll("textarea").forEach(textarea => {
  342. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  343. if (value) {
  344. textarea.textContent = value;
  345. }
  346. });
  347. this.doc.querySelectorAll("select").forEach(select => {
  348. select.querySelectorAll("option").forEach(option => {
  349. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  350. if (selected) {
  351. option.setAttribute("selected", "");
  352. }
  353. });
  354. });
  355. }
  356. lazyLoadImages() {
  357. DOM.lazyLoad(this.doc);
  358. }
  359. removeDiscardedResources() {
  360. 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\"])");
  361. this.stats.set("discarded", "objects", objectElements.length);
  362. this.stats.set("processed", "objects", objectElements.length);
  363. objectElements.forEach(element => element.remove());
  364. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  365. replacedAttributeValue.forEach(element => {
  366. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  367. if (relValue.length) {
  368. element.setAttribute("rel", relValue);
  369. } else {
  370. element.remove();
  371. }
  372. });
  373. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  374. if (this.options.compressHTML) {
  375. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  376. }
  377. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  378. if (this.options.removeScripts) {
  379. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  380. this.doc.querySelectorAll("[href]").forEach(element => {
  381. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  382. element.removeAttribute("href");
  383. }
  384. });
  385. this.doc.querySelectorAll("[src]").forEach(element => {
  386. if (element.src && element.src.match(/^\s*javascript:/)) {
  387. element.removeAttribute("src");
  388. }
  389. });
  390. }
  391. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  392. this.stats.set("processed", "audio sources", audioSourceElements.length);
  393. if (this.options.removeAudioSrc) {
  394. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  395. audioSourceElements.forEach(element => element.removeAttribute("src"));
  396. }
  397. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  398. this.stats.set("processed", "video sources", videoSourceElements.length);
  399. if (this.options.removeVideoSrc) {
  400. this.stats.set("discarded", "video sources", videoSourceElements.length);
  401. videoSourceElements.forEach(element => element.removeAttribute("src"));
  402. }
  403. }
  404. removeDefaultHeadTags() {
  405. this.doc.querySelectorAll("base").forEach(element => element.remove());
  406. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  407. this.doc.head.querySelector("meta[charset]").remove();
  408. }
  409. }
  410. removeUIElements() {
  411. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  412. }
  413. removeScripts() {
  414. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  415. this.stats.set("discarded", "scripts", scriptElements.length);
  416. this.stats.set("processed", "scripts", scriptElements.length);
  417. scriptElements.forEach(element => element.remove());
  418. }
  419. removeFrames() {
  420. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  421. this.stats.set("discarded", "frames", frameElements.length);
  422. this.stats.set("processed", "frames", frameElements.length);
  423. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  424. }
  425. removeImports() {
  426. const importElements = this.doc.querySelectorAll("link[rel=import]");
  427. this.stats.set("discarded", "HTML imports", importElements.length);
  428. this.stats.set("processed", "HTML imports", importElements.length);
  429. importElements.forEach(element => element.remove());
  430. }
  431. resetCharsetMeta() {
  432. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  433. const metaElement = this.doc.createElement("meta");
  434. metaElement.setAttribute("charset", "utf-8");
  435. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  436. }
  437. insertFaviconLink() {
  438. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  439. if (!faviconElement) {
  440. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  441. }
  442. if (!faviconElement) {
  443. faviconElement = this.doc.createElement("link");
  444. faviconElement.setAttribute("type", "image/x-icon");
  445. faviconElement.setAttribute("rel", "shortcut icon");
  446. faviconElement.setAttribute("href", "/favicon.ico");
  447. }
  448. this.doc.head.appendChild(faviconElement);
  449. }
  450. resolveHrefs() {
  451. this.doc.querySelectorAll("[href]").forEach(element => {
  452. if (element.href) {
  453. const href = element.href.baseVal ? element.href.baseVal : element.href;
  454. const normalizedHref = DomUtil.normalizeURL(href);
  455. if (!normalizedHref || normalizedHref != this.baseURI) {
  456. element.setAttribute("href", href);
  457. }
  458. }
  459. });
  460. }
  461. removeUnusedStyles() {
  462. if (!this.mediaAllInfo) {
  463. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  464. }
  465. const stats = DOM.minifyCSS(this.doc, this.mediaAllInfo);
  466. this.stats.set("processed", "CSS rules", stats.processed);
  467. this.stats.set("discarded", "CSS rules", stats.discarded);
  468. }
  469. removeAlternativeFonts() {
  470. DOM.minifyFonts(this.doc);
  471. }
  472. removeSrcSet() {
  473. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  474. const tagName = element.tagName.toLowerCase();
  475. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  476. const responsiveImageData = this.options.responsiveImageData[Number(element.getAttribute(dataAttributeName))];
  477. element.removeAttribute(dataAttributeName);
  478. if (responsiveImageData) {
  479. if (tagName == "img") {
  480. if (responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  481. element.removeAttribute("srcset");
  482. element.removeAttribute("sizes");
  483. element.src = responsiveImageData.source.src;
  484. }
  485. }
  486. if (tagName == "picture") {
  487. const imageElement = element.querySelector("img");
  488. if (responsiveImageData.source && responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  489. imageElement.removeAttribute("srcset");
  490. imageElement.removeAttribute("sizes");
  491. imageElement.src = responsiveImageData.source.src;
  492. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  493. } else {
  494. if (responsiveImageData.sources) {
  495. element.querySelectorAll("source").forEach(sourceElement => {
  496. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  497. sourceElement.remove();
  498. }
  499. });
  500. const sourceElements = element.querySelectorAll("source");
  501. if (sourceElements.length) {
  502. const lastSourceElement = sourceElements[sourceElements.length - 1];
  503. if (imageElement) {
  504. if (lastSourceElement.src) {
  505. imageElement.src = lastSourceElement.src;
  506. } else {
  507. imageElement.removeAttribute("src");
  508. }
  509. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  510. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  511. } else {
  512. imageElement.removeAttribute("srcset");
  513. }
  514. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  515. }
  516. }
  517. }
  518. }
  519. }
  520. }
  521. });
  522. }
  523. postRemoveAlternativeFonts() {
  524. DOM.minifyFonts(this.doc, true);
  525. }
  526. removeHiddenElements() {
  527. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  528. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  529. this.stats.set("processed", "hidden elements", hiddenElements.length);
  530. hiddenElements.forEach(element => element.remove());
  531. }
  532. compressHTML() {
  533. let size;
  534. if (this.options.displayStats) {
  535. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  536. }
  537. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  538. if (this.options.displayStats) {
  539. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  540. }
  541. }
  542. groupDuplicateImages() {
  543. if (this.options.imageData) {
  544. if (!this.mediaAllInfo) {
  545. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  546. }
  547. DOM.minifyImages(this.doc, this.mediaAllInfo, this.options);
  548. }
  549. }
  550. removeAlternativeMedias() {
  551. const stats = DOM.minifyMedias(this.doc);
  552. this.stats.set("processed", "medias", stats.processed);
  553. this.stats.set("discarded", "medias", stats.discarded);
  554. }
  555. compressCSS() {
  556. this.doc.querySelectorAll("style").forEach(styleElement => {
  557. if (styleElement) {
  558. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  559. }
  560. });
  561. this.doc.querySelectorAll("[style]").forEach(element => {
  562. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  563. });
  564. }
  565. insertSingleFileComment() {
  566. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  567. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  568. }
  569. replaceCanvasElements() {
  570. if (this.options.canvasData) {
  571. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  572. const canvasData = this.options.canvasData[indexCanvasElement];
  573. if (canvasData) {
  574. const imgElement = this.doc.createElement("img");
  575. imgElement.setAttribute("src", canvasData.dataURI);
  576. Array.from(canvasElement.attributes).forEach(attribute => {
  577. if (attribute.value) {
  578. imgElement.setAttribute(attribute.name, attribute.value);
  579. }
  580. });
  581. if (!imgElement.width && canvasData.width) {
  582. imgElement.style.pixelWidth = canvasData.width;
  583. }
  584. if (!imgElement.height && canvasData.height) {
  585. imgElement.style.pixelHeight = canvasData.height;
  586. }
  587. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  588. this.stats.add("processed", "canvas", 1);
  589. }
  590. });
  591. }
  592. }
  593. insertFonts() {
  594. if (this.options.fontsData && this.options.fontsData.length) {
  595. let stylesheetContent = "";
  596. this.options.fontsData.forEach(fontStyles => {
  597. if (fontStyles["font-family"] && fontStyles.src) {
  598. stylesheetContent += "@font-face{";
  599. let stylesContent = "";
  600. Object.keys(fontStyles).forEach(fontStyle => {
  601. if (stylesContent) {
  602. stylesContent += ";";
  603. }
  604. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  605. });
  606. stylesheetContent += stylesContent + "}";
  607. }
  608. });
  609. if (stylesheetContent) {
  610. const styleElement = this.doc.createElement("style");
  611. styleElement.textContent = stylesheetContent;
  612. const existingStyleElement = this.doc.querySelector("style");
  613. if (existingStyleElement) {
  614. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  615. } else {
  616. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  617. }
  618. }
  619. }
  620. }
  621. replaceStyleContents() {
  622. if (this.options.stylesheetContents) {
  623. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  624. if (this.options.stylesheetContents[styleIndex]) {
  625. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  626. }
  627. });
  628. }
  629. }
  630. insertVideoPosters() {
  631. if (this.options.postersData) {
  632. this.doc.querySelectorAll("video[src], video > source[src]").forEach((videoElement, videoIndex) => {
  633. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  634. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  635. }
  636. });
  637. }
  638. }
  639. async pageResources() {
  640. const resourcePromises = [
  641. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  642. 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),
  643. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image]"), "src", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  644. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest),
  645. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  646. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  647. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  648. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  649. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)
  650. ];
  651. if (!this.options.removeAudioSrc) {
  652. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", PREFIX_DATA_URI_AUDIO, this.baseURI, this.batchRequest));
  653. }
  654. if (!this.options.removeVideoSrc) {
  655. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", PREFIX_DATA_URI_VIDEO, this.baseURI, this.batchRequest));
  656. }
  657. if (this.options.lazyLoadImages) {
  658. const imageSelectors = DOM.lazyLoaderImageSelectors();
  659. 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)));
  660. 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)));
  661. }
  662. await resourcePromises;
  663. }
  664. async inlineStylesheets(initialization) {
  665. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  666. if (!initialization) {
  667. this.stats.add("processed", "stylesheets", 1);
  668. }
  669. let stylesheetContent = styleElement.textContent;
  670. if (initialization) {
  671. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  672. } else {
  673. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  674. }
  675. styleElement.textContent = stylesheetContent;
  676. }));
  677. }
  678. async scripts() {
  679. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  680. if (scriptElement.src) {
  681. this.stats.add("processed", "scripts", 1);
  682. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  683. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  684. }
  685. scriptElement.removeAttribute("src");
  686. }));
  687. }
  688. async frames(initialization) {
  689. if (this.options.framesData) {
  690. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  691. await Promise.all(frameElements.map(async frameElement => {
  692. DomProcessorHelper.setFrameEmptySrc(frameElement);
  693. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  694. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  695. if (frameWindowId) {
  696. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  697. if (frameData) {
  698. if (initialization) {
  699. const options = Object.create(this.options);
  700. options.insertSingleFileComment = false;
  701. options.insertFaviconLink = false;
  702. options.doc = null;
  703. options.win = null;
  704. options.url = frameData.baseURI;
  705. options.windowId = frameWindowId;
  706. if (frameData.content) {
  707. options.content = frameData.content;
  708. options.canvasData = frameData.canvasData;
  709. options.stylesheetContents = frameData.stylesheetContents;
  710. options.currentSrcImages = frameData.currentSrcImages;
  711. frameData.processor = new PageProcessor(options);
  712. frameData.frameElement = frameElement;
  713. await frameData.processor.loadPage();
  714. await frameData.processor.initialize();
  715. frameData.maxResources = this.batchRequest.getMaxResources();
  716. }
  717. } else {
  718. if (frameData.processor) {
  719. this.stats.add("processed", "frames", 1);
  720. await frameData.processor.preparePageData();
  721. const pageData = await frameData.processor.getPageData();
  722. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  723. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  724. this.stats.addAll(pageData);
  725. } else {
  726. this.stats.add("discarded", "frames", 1);
  727. }
  728. }
  729. }
  730. }
  731. }));
  732. }
  733. }
  734. async htmlImports(initialization) {
  735. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  736. if (!this.relImportProcessors) {
  737. this.relImportProcessors = new Map();
  738. }
  739. await Promise.all(linkElements.map(async linkElement => {
  740. if (initialization) {
  741. const resourceURL = linkElement.href;
  742. const options = Object.create(this.options);
  743. options.insertSingleFileComment = false;
  744. options.insertFaviconLink = false;
  745. options.doc = null;
  746. options.win = null;
  747. options.url = resourceURL;
  748. if (resourceURL) {
  749. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  750. const processor = new PageProcessor(options);
  751. this.relImportProcessors.set(linkElement, processor);
  752. await processor.loadPage();
  753. return processor.initialize();
  754. }
  755. }
  756. } else {
  757. linkElement.setAttribute("href", EMPTY_DATA_URI);
  758. const processor = this.relImportProcessors.get(linkElement);
  759. if (processor) {
  760. this.stats.add("processed", "HTML imports", 1);
  761. this.relImportProcessors.delete(linkElement);
  762. const pageData = await processor.getPageData();
  763. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  764. this.stats.addAll(pageData);
  765. } else {
  766. this.stats.add("discarded", "HTML imports", 1);
  767. }
  768. }
  769. }));
  770. }
  771. async attributeStyles(initialization) {
  772. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  773. let stylesheetContent = element.getAttribute("style");
  774. if (initialization) {
  775. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  776. } else {
  777. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  778. }
  779. element.setAttribute("style", stylesheetContent);
  780. }));
  781. }
  782. async linkStylesheets() {
  783. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  784. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  785. const styleElement = this.doc.createElement("style");
  786. styleElement.textContent = stylesheetContent;
  787. linkElement.parentElement.replaceChild(styleElement, linkElement);
  788. }));
  789. }
  790. }
  791. // ---------
  792. // DomHelper
  793. // ---------
  794. const REGEXP_AMP = /&/g;
  795. const REGEXP_NBSP = /\u00a0/g;
  796. const REGEXP_START_TAG = /</g;
  797. const REGEXP_END_TAG = />/g;
  798. const REGEXP_URL_HASH = /(#.+?)$/;
  799. const PREFIX_DATA_URI_IMAGE = "data:image/";
  800. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  801. const PREFIX_DATA_URI_VIDEO = "data:video/";
  802. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  803. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  804. class DomProcessorHelper {
  805. static async getFilename(options, content) {
  806. let filename = options.filenameTemplate;
  807. const date = new Date();
  808. const url = new URL(options.url);
  809. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  810. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  811. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  812. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  813. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  814. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  815. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  816. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  817. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  818. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  819. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  820. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  821. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  822. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  823. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  824. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  825. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  826. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  827. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  828. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  829. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  830. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  831. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  832. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  833. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  834. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  835. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  836. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  837. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  838. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  839. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  840. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  841. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  842. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  843. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  844. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  845. if (!options.backgroundSave) {
  846. filename = filename.replace(/\//g, "_");
  847. }
  848. if (filename.length > 192) {
  849. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  850. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  851. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  852. }
  853. if (!filename) {
  854. filename = "Unnamed page";
  855. }
  856. return filename;
  857. }
  858. static setFrameEmptySrc(frameElement) {
  859. if (frameElement.tagName == "OBJECT") {
  860. frameElement.setAttribute("data", "data:text/html,");
  861. } else {
  862. frameElement.setAttribute("srcdoc", "");
  863. frameElement.removeAttribute("src");
  864. }
  865. }
  866. static setFrameContent(frameElement, content) {
  867. if (frameElement.tagName == "OBJECT") {
  868. frameElement.setAttribute("data", "data:text/html," + content);
  869. } else {
  870. frameElement.setAttribute("srcdoc", content);
  871. frameElement.removeAttribute("src");
  872. }
  873. }
  874. static isolateElements(rootElement) {
  875. rootElement.querySelectorAll("*").forEach(element => {
  876. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  877. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  878. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  879. element.remove();
  880. }
  881. });
  882. isolateParentElements(rootElement.parentElement, rootElement);
  883. function isolateParentElements(parentElement, element) {
  884. if (parentElement) {
  885. Array.from(parentElement.childNodes).forEach(node => {
  886. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  887. node.remove();
  888. }
  889. });
  890. }
  891. element = element.parentElement;
  892. if (element && element.parentElement) {
  893. isolateParentElements(element.parentElement, element);
  894. }
  895. }
  896. }
  897. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  898. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  899. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  900. const imports = DomUtil.getImportFunctions(stylesheetContent);
  901. await Promise.all(imports.map(async cssImport => {
  902. const match = DomUtil.matchImport(cssImport);
  903. if (match) {
  904. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  905. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  906. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  907. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  908. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  909. if (stylesheetContent.includes(cssImport)) {
  910. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  911. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  912. }
  913. }
  914. }
  915. }));
  916. return stylesheetContent;
  917. }
  918. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  919. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  920. urlFunctions.map(urlFunction => {
  921. const originalResourceURL = DomUtil.matchURL(urlFunction);
  922. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  923. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  924. const resolvedURL = new URL(resourceURL, baseURI).href;
  925. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  926. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  927. }
  928. } else {
  929. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  930. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  931. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  932. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  933. }
  934. }
  935. }
  936. });
  937. return stylesheetContent;
  938. }
  939. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  940. resourceURL = DomUtil.normalizeURL(resourceURL);
  941. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  942. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  943. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  944. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  945. return stylesheetContent;
  946. }
  947. }
  948. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  949. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  950. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  951. let indexVariable = 0;
  952. await Promise.all(urlFunctions.map(async urlFunction => {
  953. const originalResourceURL = DomUtil.matchURL(urlFunction);
  954. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  955. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  956. const dataURI = await batchRequest.addURL(resourceURL);
  957. const regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  958. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  959. const functions = stylesheetContent.match(regExpUrlFunction);
  960. if (options.groupDuplicateImages && functions && functions.length > 1) {
  961. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  962. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  963. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + variableName + ")");
  964. indexVariable++;
  965. } else {
  966. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  967. }
  968. } else {
  969. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  970. }
  971. }
  972. }));
  973. if (indexVariable) {
  974. stylesheetContent = ":root{" + stylesheetContent;
  975. }
  976. return stylesheetContent;
  977. }
  978. static async processAttribute(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  979. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  980. let resourceURL = resourceElement.getAttribute(attributeName);
  981. if (resourceURL) {
  982. resourceURL = DomUtil.normalizeURL(resourceURL);
  983. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  984. try {
  985. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  986. if (dataURI.startsWith(prefixDataURI) || dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE)) {
  987. resourceElement.setAttribute(attributeName, dataURI);
  988. } else {
  989. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  990. }
  991. } catch (error) {
  992. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  993. }
  994. }
  995. }
  996. }));
  997. }
  998. static async processXLinks(resourceElements, baseURI, batchRequest) {
  999. const attributeName = "xlink:href";
  1000. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1001. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1002. if (originalResourceURL) {
  1003. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1004. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1005. try {
  1006. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  1007. const DOMParser = DOM.getParser();
  1008. if (DOMParser) {
  1009. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  1010. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1011. if (hashMatch && hashMatch[0]) {
  1012. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1013. if (symbolElement) {
  1014. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1015. resourceElement.parentElement.appendChild(symbolElement);
  1016. }
  1017. } else {
  1018. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1019. }
  1020. } else {
  1021. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1022. }
  1023. } catch (error) {
  1024. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1025. }
  1026. }
  1027. }
  1028. }));
  1029. }
  1030. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  1031. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1032. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1033. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1034. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1035. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1036. try {
  1037. let dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  1038. if (!dataURI.startsWith(prefixDataURI) && !dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE)) {
  1039. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1040. }
  1041. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1042. } catch (error) {
  1043. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1044. }
  1045. }
  1046. }));
  1047. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1048. }));
  1049. }
  1050. }
  1051. // -------
  1052. // DomUtil
  1053. // -------
  1054. const DATA_URI_PREFIX = "data:";
  1055. const BLOB_URI_PREFIX = "blob:";
  1056. const HTTP_URI_PREFIX = /^https?:\/\//;
  1057. const ABOUT_BLANK_URI = "about:blank";
  1058. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1059. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1060. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1061. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1062. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1063. 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;
  1064. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1065. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1066. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1067. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1068. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1069. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1070. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1071. class DomUtil {
  1072. static normalizeURL(url) {
  1073. if (url.startsWith(DATA_URI_PREFIX)) {
  1074. return url;
  1075. } else {
  1076. return url.split("#")[0];
  1077. }
  1078. }
  1079. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1080. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1081. if (template.match(replaceRegExp)) {
  1082. let value = await valueGetter();
  1083. if (!dontReplaceSlash) {
  1084. value = value.replace(/\/+/g, "_");
  1085. }
  1086. return template.replace(replaceRegExp, value);
  1087. }
  1088. return template;
  1089. }
  1090. static getLastSegment(url) {
  1091. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1092. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1093. if (!lastSegment) {
  1094. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1095. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1096. }
  1097. if (!lastSegment) {
  1098. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1099. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1100. }
  1101. if (!lastSegment) {
  1102. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1103. }
  1104. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1105. return lastSegment;
  1106. }
  1107. static getRegExp(string) {
  1108. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1109. }
  1110. static getUrlFunctions(stylesheetContent) {
  1111. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1112. }
  1113. static getImportFunctions(stylesheetContent) {
  1114. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1115. }
  1116. static matchURL(stylesheetContent) {
  1117. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1118. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1119. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1120. return match && match[1];
  1121. }
  1122. static testValidPath(resourceURL) {
  1123. 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));
  1124. }
  1125. static matchImport(stylesheetContent) {
  1126. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1127. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1128. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1129. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1130. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1131. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1132. if (match) {
  1133. const [, resourceURL, media] = match;
  1134. return { resourceURL, media };
  1135. }
  1136. }
  1137. static removeCssComments(stylesheetContent) {
  1138. let start, end;
  1139. do {
  1140. start = stylesheetContent.indexOf("/*");
  1141. end = stylesheetContent.indexOf("*/", start + 2);
  1142. if (start != -1 && end != -1) {
  1143. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1144. }
  1145. } while (start != -1 && end != -1);
  1146. return stylesheetContent;
  1147. }
  1148. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1149. if (mediaQuery) {
  1150. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1151. } else {
  1152. return stylesheetContent;
  1153. }
  1154. }
  1155. }
  1156. function log(...args) {
  1157. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1158. }
  1159. // -----
  1160. // Stats
  1161. // -----
  1162. const STATS_DEFAULT_VALUES = {
  1163. discarded: {
  1164. "HTML bytes": 0,
  1165. "hidden elements": 0,
  1166. "HTML imports": 0,
  1167. scripts: 0,
  1168. objects: 0,
  1169. "audio sources": 0,
  1170. "video sources": 0,
  1171. frames: 0,
  1172. "CSS rules": 0,
  1173. canvas: 0,
  1174. stylesheets: 0,
  1175. resources: 0,
  1176. medias: 0
  1177. },
  1178. processed: {
  1179. "HTML bytes": 0,
  1180. "hidden elements": 0,
  1181. "HTML imports": 0,
  1182. scripts: 0,
  1183. objects: 0,
  1184. "audio sources": 0,
  1185. "video sources": 0,
  1186. frames: 0,
  1187. "CSS rules": 0,
  1188. canvas: 0,
  1189. stylesheets: 0,
  1190. resources: 0,
  1191. medias: 0
  1192. }
  1193. };
  1194. class Stats {
  1195. constructor(options) {
  1196. this.options = options;
  1197. if (options.displayStats) {
  1198. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1199. }
  1200. }
  1201. set(type, subType, value) {
  1202. if (this.options.displayStats) {
  1203. this.data[type][subType] = value;
  1204. }
  1205. }
  1206. add(type, subType, value) {
  1207. if (this.options.displayStats) {
  1208. this.data[type][subType] += value;
  1209. }
  1210. }
  1211. addAll(pageData) {
  1212. if (this.options.displayStats) {
  1213. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1214. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1215. }
  1216. }
  1217. }
  1218. return { getClass };
  1219. })();