single-file-core.js 52 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268
  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: "removeSrcSet", action: "removeSrcSet" },
  70. { option: "removeFrames", action: "removeFrames" },
  71. { option: "removeImports", action: "removeImports" },
  72. { option: "removeScripts", action: "removeScripts" },
  73. { action: "removeDiscardedResources" },
  74. { action: "resetCharsetMeta" },
  75. { action: "setInputValues" },
  76. { option: "insertFaviconLink", action: "insertFaviconLink" },
  77. { action: "resolveHrefs" },
  78. { action: "replaceCanvasElements" },
  79. { action: "insertFonts" },
  80. { option: "removeHiddenElements", action: "removeHiddenElements" }
  81. ],
  82. parallel: [
  83. { action: "inlineStylesheets" },
  84. { action: "linkStylesheets" },
  85. { action: "attributeStyles" },
  86. { option: "!removeFrames", action: "frames" },
  87. { option: "!removeImports", action: "htmlImports" }
  88. ]
  89. }, {
  90. sequential: [
  91. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  92. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  93. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  94. ],
  95. parallel: [
  96. { action: "inlineStylesheets" },
  97. { action: "attributeStyles" },
  98. { action: "pageResources" },
  99. { option: "!removeScripts", action: "scripts" }
  100. ]
  101. }, {
  102. sequential: [
  103. { option: "lazyLoadImages", action: "lazyLoadImages" },
  104. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  105. { option: "compressCSS", action: "compressCSS" }
  106. ],
  107. parallel: [
  108. { option: "!removeFrames", action: "frames" },
  109. { option: "!removeImports", action: "htmlImports" },
  110. ]
  111. }, {
  112. sequential: [
  113. { option: "compressHTML", action: "compressHTML" },
  114. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  115. { action: "removeDefaultHeadTags" }
  116. ]
  117. }];
  118. class PageProcessor {
  119. constructor(options) {
  120. this.options = options;
  121. this.options.url = this.options.url || this.options.doc.location.href;
  122. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  123. this.batchRequest = new BatchRequest();
  124. this.processor = new DOMProcessor(options, this.batchRequest);
  125. if (this.options.doc) {
  126. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  127. this.options.canvasData = docData.canvasData;
  128. this.options.fontsData = docData.fontsData;
  129. this.options.stylesheetContents = docData.stylesheetContents;
  130. this.options.imageData = docData.imageData;
  131. }
  132. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  133. this.onprogress = options.onprogress || (() => { });
  134. }
  135. async loadPage() {
  136. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  137. await this.processor.loadPage(this.options.content);
  138. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  139. }
  140. async initialize() {
  141. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  142. await this.executeStage(0);
  143. this.pendingPromises = this.executeStage(1);
  144. if (this.options.doc) {
  145. DOM.postProcessDoc(this.options.doc, this.options);
  146. this.options.doc = null;
  147. this.options.win = null;
  148. }
  149. }
  150. async preparePageData() {
  151. if (!this.options.windowId) {
  152. this.processor.initialize(this.batchRequest);
  153. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  154. }
  155. await this.batchRequest.run(details => {
  156. details.pageURL = this.options.url;
  157. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  158. }, this.options);
  159. await this.pendingPromises;
  160. await this.executeStage(2);
  161. await this.executeStage(3);
  162. }
  163. async getPageData() {
  164. if (!this.options.windowId) {
  165. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  166. }
  167. return this.processor.getPageData();
  168. }
  169. async executeStage(step) {
  170. if (DEBUG) {
  171. log("**** STARTED STAGE", step, "****");
  172. }
  173. STAGES[step].sequential.forEach(task => {
  174. let startTime;
  175. if (DEBUG) {
  176. startTime = Date.now();
  177. log(" -- STARTED task =", task.action);
  178. }
  179. this.executeTask(task, !step);
  180. if (DEBUG) {
  181. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  182. }
  183. });
  184. if (STAGES[step].parallel) {
  185. return await Promise.all(STAGES[step].parallel.map(task => {
  186. let startTime;
  187. if (DEBUG) {
  188. startTime = Date.now();
  189. log(" // STARTED task =", task.action);
  190. }
  191. const promise = this.executeTask(task, !step);
  192. if (DEBUG) {
  193. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  194. }
  195. return promise;
  196. }));
  197. }
  198. if (DEBUG) {
  199. log("**** ENDED STAGE", step, "****");
  200. }
  201. }
  202. executeTask(task, initialization) {
  203. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  204. return this.processor[task.action](initialization);
  205. }
  206. }
  207. }
  208. // --------
  209. // BatchRequest
  210. // --------
  211. class BatchRequest {
  212. constructor() {
  213. this.requests = new Map();
  214. }
  215. async addURL(resourceURL, asDataURI = true) {
  216. return new Promise((resolve, reject) => {
  217. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  218. const resourceRequests = this.requests.get(requestKey);
  219. if (resourceRequests) {
  220. resourceRequests.push({ resolve, reject });
  221. } else {
  222. this.requests.set(requestKey, [{ resolve, reject }]);
  223. }
  224. });
  225. }
  226. getMaxResources() {
  227. return Array.from(this.requests.keys()).length;
  228. }
  229. async run(onloadListener, options) {
  230. const resourceURLs = Array.from(this.requests.keys());
  231. let indexResource = 0;
  232. return Promise.all(resourceURLs.map(async requestKey => {
  233. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  234. const resourceRequests = this.requests.get(requestKey);
  235. try {
  236. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  237. indexResource = indexResource + 1;
  238. onloadListener({ index: indexResource, url: resourceURL });
  239. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  240. } catch (error) {
  241. indexResource = indexResource + 1;
  242. onloadListener({ index: indexResource, url: resourceURL });
  243. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  244. }
  245. this.requests.delete(requestKey);
  246. }));
  247. }
  248. }
  249. // ------------
  250. // DOMProcessor
  251. // ------------
  252. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  253. const EMPTY_DATA_URI = "data:base64,";
  254. class DOMProcessor {
  255. constructor(options, batchRequest) {
  256. this.options = options;
  257. this.stats = new Stats(options);
  258. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  259. this.batchRequest = batchRequest;
  260. }
  261. initialize() {
  262. this.maxResources = this.batchRequest.getMaxResources();
  263. if (!this.options.removeFrames) {
  264. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  265. }
  266. this.stats.set("processed", "resources", this.maxResources);
  267. }
  268. async loadPage(pageContent) {
  269. if (!pageContent || this.options.saveRawPage) {
  270. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  271. }
  272. this.doc = DOM.createDoc(pageContent, this.baseURI);
  273. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  274. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  275. await DOMProcessor.loadEscapedFragmentPage();
  276. }
  277. }
  278. async loadEscapedFragmentPage() {
  279. if (this.baseURI.includes("?")) {
  280. this.baseURI += "&";
  281. } else {
  282. this.baseURI += "?";
  283. }
  284. this.baseURI += ESCAPED_FRAGMENT;
  285. await this.loadPage();
  286. }
  287. async getPageData() {
  288. DOM.postProcessDoc(this.doc, this.options);
  289. if (this.options.selected) {
  290. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  291. if (rootElement) {
  292. DomProcessorHelper.isolateElements(rootElement);
  293. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  294. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  295. }
  296. }
  297. const titleElement = this.doc.querySelector("title");
  298. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  299. this.options.info = {};
  300. const descriptionElement = this.doc.querySelector("meta[name=description]");
  301. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  302. this.options.info.lang = this.doc.documentElement.lang;
  303. const authorElement = this.doc.querySelector("meta[name=author]");
  304. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  305. const creatorElement = this.doc.querySelector("meta[name=creator]");
  306. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  307. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  308. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  309. const url = new URL(this.baseURI);
  310. let size;
  311. if (this.options.displayStats) {
  312. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  313. }
  314. const content = DOM.serialize(this.doc, this.options.compressHTML);
  315. if (this.options.displayStats) {
  316. const contentSize = DOM.getContentSize(content);
  317. this.stats.set("processed", "HTML bytes", contentSize);
  318. this.stats.add("discarded", "HTML bytes", size - contentSize);
  319. }
  320. const filename = await DomProcessorHelper.getFilename(this.options, content);
  321. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  322. return {
  323. stats: this.stats.data,
  324. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  325. filename,
  326. content
  327. };
  328. }
  329. setInputValues() {
  330. this.doc.querySelectorAll("input").forEach(input => {
  331. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  332. if (value) {
  333. input.setAttribute("value", value);
  334. }
  335. });
  336. this.doc.querySelectorAll("textarea").forEach(textarea => {
  337. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  338. if (value) {
  339. textarea.textContent = value;
  340. }
  341. });
  342. this.doc.querySelectorAll("select").forEach(select => {
  343. select.querySelectorAll("option").forEach(option => {
  344. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  345. if (selected) {
  346. option.setAttribute("selected", "");
  347. }
  348. });
  349. });
  350. }
  351. lazyLoadImages() {
  352. DOM.lazyLoad(this.doc);
  353. }
  354. removeDiscardedResources() {
  355. 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\"])");
  356. this.stats.set("discarded", "objects", objectElements.length);
  357. this.stats.set("processed", "objects", objectElements.length);
  358. objectElements.forEach(element => element.remove());
  359. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  360. replacedAttributeValue.forEach(element => {
  361. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  362. if (relValue.length) {
  363. element.setAttribute("rel", relValue);
  364. } else {
  365. element.remove();
  366. }
  367. });
  368. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  369. if (this.options.compressHTML) {
  370. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  371. }
  372. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  373. if (this.options.removeScripts) {
  374. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  375. this.doc.querySelectorAll("[href]").forEach(element => {
  376. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  377. element.removeAttribute("href");
  378. }
  379. });
  380. this.doc.querySelectorAll("[src]").forEach(element => {
  381. if (element.src && element.src.match(/^\s*javascript:/)) {
  382. element.removeAttribute("src");
  383. }
  384. });
  385. }
  386. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  387. this.stats.set("processed", "audio sources", audioSourceElements.length);
  388. if (this.options.removeAudioSrc) {
  389. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  390. audioSourceElements.forEach(element => element.removeAttribute("src"));
  391. }
  392. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  393. this.stats.set("processed", "video sources", videoSourceElements.length);
  394. if (this.options.removeVideoSrc) {
  395. this.stats.set("discarded", "video sources", videoSourceElements.length);
  396. videoSourceElements.forEach(element => element.removeAttribute("src"));
  397. }
  398. }
  399. removeDefaultHeadTags() {
  400. this.doc.querySelectorAll("base").forEach(element => element.remove());
  401. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  402. this.doc.head.querySelector("meta[charset]").remove();
  403. }
  404. }
  405. removeUIElements() {
  406. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  407. }
  408. removeScripts() {
  409. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  410. this.stats.set("discarded", "scripts", scriptElements.length);
  411. this.stats.set("processed", "scripts", scriptElements.length);
  412. scriptElements.forEach(element => element.remove());
  413. }
  414. removeFrames() {
  415. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  416. this.stats.set("discarded", "frames", frameElements.length);
  417. this.stats.set("processed", "frames", frameElements.length);
  418. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  419. }
  420. removeImports() {
  421. const importElements = this.doc.querySelectorAll("link[rel=import]");
  422. this.stats.set("discarded", "HTML imports", importElements.length);
  423. this.stats.set("processed", "HTML imports", importElements.length);
  424. importElements.forEach(element => element.remove());
  425. }
  426. resetCharsetMeta() {
  427. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  428. const metaElement = this.doc.createElement("meta");
  429. metaElement.setAttribute("charset", "utf-8");
  430. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  431. }
  432. insertFaviconLink() {
  433. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  434. if (!faviconElement) {
  435. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  436. }
  437. if (!faviconElement) {
  438. faviconElement = this.doc.createElement("link");
  439. faviconElement.setAttribute("type", "image/x-icon");
  440. faviconElement.setAttribute("rel", "shortcut icon");
  441. faviconElement.setAttribute("href", "/favicon.ico");
  442. }
  443. this.doc.head.appendChild(faviconElement);
  444. }
  445. resolveHrefs() {
  446. this.doc.querySelectorAll("[href]").forEach(element => {
  447. if (element.href) {
  448. const href = element.href.baseVal ? element.href.baseVal : element.href;
  449. const normalizedHref = DomUtil.normalizeURL(href);
  450. if (!normalizedHref || normalizedHref != this.baseURI) {
  451. element.setAttribute("href", href);
  452. }
  453. }
  454. });
  455. }
  456. removeUnusedStyles() {
  457. const stats = DOM.minifyCSS(this.doc);
  458. this.stats.set("processed", "CSS rules", stats.processed);
  459. this.stats.set("discarded", "CSS rules", stats.discarded);
  460. }
  461. removeAlternativeFonts() {
  462. DOM.minifyFonts(this.doc);
  463. }
  464. removeSrcSet() {
  465. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  466. const tagName = element.tagName.toLowerCase();
  467. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  468. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  469. element.removeAttribute(dataAttributeName);
  470. if (imageData) {
  471. if (tagName == "img") {
  472. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  473. element.removeAttribute("srcset");
  474. element.removeAttribute("sizes");
  475. element.src = imageData.source.src;
  476. }
  477. }
  478. if (tagName == "picture") {
  479. const imageElement = element.querySelector("img");
  480. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  481. imageElement.removeAttribute("srcset");
  482. imageElement.removeAttribute("sizes");
  483. imageElement.src = imageData.source.src;
  484. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  485. } else {
  486. if (imageData.sources) {
  487. element.querySelectorAll("source").forEach(sourceElement => {
  488. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  489. sourceElement.remove();
  490. }
  491. });
  492. const sourceElements = element.querySelectorAll("source");
  493. if (sourceElements.length) {
  494. const lastSourceElement = sourceElements[sourceElements.length - 1];
  495. if (imageElement) {
  496. if (lastSourceElement.src) {
  497. imageElement.src = lastSourceElement.src;
  498. } else {
  499. imageElement.removeAttribute("src");
  500. }
  501. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  502. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  503. } else {
  504. imageElement.removeAttribute("srcset");
  505. }
  506. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  507. }
  508. }
  509. }
  510. }
  511. }
  512. }
  513. });
  514. }
  515. postRemoveAlternativeFonts() {
  516. DOM.minifyFonts(this.doc, true);
  517. }
  518. removeHiddenElements() {
  519. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  520. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  521. this.stats.set("processed", "hidden elements", hiddenElements.length);
  522. hiddenElements.forEach(element => element.remove());
  523. }
  524. compressHTML() {
  525. let size;
  526. if (this.options.displayStats) {
  527. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  528. }
  529. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  530. if (this.options.displayStats) {
  531. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  532. }
  533. }
  534. removeAlternativeMedias() {
  535. const stats = DOM.minifyMedias(this.doc);
  536. this.stats.set("processed", "medias", stats.processed);
  537. this.stats.set("discarded", "medias", stats.discarded);
  538. }
  539. compressCSS() {
  540. this.doc.querySelectorAll("style").forEach(styleElement => {
  541. if (styleElement) {
  542. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  543. }
  544. });
  545. this.doc.querySelectorAll("[style]").forEach(element => {
  546. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  547. });
  548. }
  549. insertSingleFileComment() {
  550. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  551. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  552. }
  553. replaceCanvasElements() {
  554. if (this.options.canvasData) {
  555. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  556. const canvasData = this.options.canvasData[indexCanvasElement];
  557. if (canvasData) {
  558. const imgElement = this.doc.createElement("img");
  559. imgElement.setAttribute("src", canvasData.dataURI);
  560. Array.from(canvasElement.attributes).forEach(attribute => {
  561. if (attribute.value) {
  562. imgElement.setAttribute(attribute.name, attribute.value);
  563. }
  564. });
  565. if (!imgElement.width && canvasData.width) {
  566. imgElement.style.pixelWidth = canvasData.width;
  567. }
  568. if (!imgElement.height && canvasData.height) {
  569. imgElement.style.pixelHeight = canvasData.height;
  570. }
  571. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  572. this.stats.add("processed", "canvas", 1);
  573. }
  574. });
  575. }
  576. }
  577. insertFonts() {
  578. if (this.options.fontsData && this.options.fontsData.length) {
  579. let stylesheetContent = "";
  580. this.options.fontsData.forEach(fontStyles => {
  581. if (fontStyles["font-family"] && fontStyles.src) {
  582. stylesheetContent += "@font-face{";
  583. let stylesContent = "";
  584. Object.keys(fontStyles).forEach(fontStyle => {
  585. if (stylesContent) {
  586. stylesContent += ";";
  587. }
  588. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  589. });
  590. stylesheetContent += stylesContent + "}";
  591. }
  592. });
  593. if (stylesheetContent) {
  594. const styleElement = this.doc.createElement("style");
  595. styleElement.textContent = stylesheetContent;
  596. const existingStyleElement = this.doc.querySelector("style");
  597. if (existingStyleElement) {
  598. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  599. } else {
  600. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  601. }
  602. }
  603. }
  604. }
  605. replaceStyleContents() {
  606. if (this.options.stylesheetContents) {
  607. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  608. if (this.options.stylesheetContents[styleIndex]) {
  609. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  610. }
  611. });
  612. }
  613. }
  614. async pageResources() {
  615. const resourcePromises = [
  616. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI, this.batchRequest),
  617. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", this.baseURI, this.batchRequest),
  618. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], embed[src*=\".svg\"]"), "src", this.baseURI, this.batchRequest),
  619. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI, this.batchRequest),
  620. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI, this.batchRequest),
  621. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", this.baseURI, this.batchRequest),
  622. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  623. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.batchRequest)
  624. ];
  625. if (!this.options.removeAudioSrc) {
  626. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.batchRequest));
  627. }
  628. if (!this.options.removeVideoSrc) {
  629. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.batchRequest));
  630. }
  631. if (this.options.lazyLoadImages) {
  632. const imageSelectors = DOM.lazyLoaderImageSelectors();
  633. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], this.baseURI, this.batchRequest)));
  634. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], this.baseURI, this.batchRequest)));
  635. }
  636. await resourcePromises;
  637. }
  638. async inlineStylesheets(initialization) {
  639. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  640. if (!initialization) {
  641. this.stats.add("processed", "stylesheets", 1);
  642. }
  643. let stylesheetContent = styleElement.textContent;
  644. if (initialization) {
  645. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  646. } else {
  647. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  648. }
  649. styleElement.textContent = stylesheetContent;
  650. }));
  651. }
  652. async scripts() {
  653. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  654. if (scriptElement.src) {
  655. this.stats.add("processed", "scripts", 1);
  656. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  657. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  658. }
  659. scriptElement.removeAttribute("src");
  660. }));
  661. }
  662. async frames(initialization) {
  663. if (this.options.framesData) {
  664. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  665. await Promise.all(frameElements.map(async frameElement => {
  666. DomProcessorHelper.setFrameEmptySrc(frameElement);
  667. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  668. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  669. if (frameWindowId) {
  670. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  671. if (frameData) {
  672. if (initialization) {
  673. const options = Object.create(this.options);
  674. options.insertSingleFileComment = false;
  675. options.insertFaviconLink = false;
  676. options.doc = null;
  677. options.win = null;
  678. options.url = frameData.baseURI;
  679. options.windowId = frameWindowId;
  680. if (frameData.content) {
  681. options.content = frameData.content;
  682. options.canvasData = frameData.canvasData;
  683. options.stylesheetContents = frameData.stylesheetContents;
  684. options.currentSrcImages = frameData.currentSrcImages;
  685. frameData.processor = new PageProcessor(options);
  686. frameData.frameElement = frameElement;
  687. await frameData.processor.loadPage();
  688. await frameData.processor.initialize();
  689. frameData.maxResources = this.batchRequest.getMaxResources();
  690. }
  691. } else {
  692. if (frameData.processor) {
  693. this.stats.add("processed", "frames", 1);
  694. await frameData.processor.preparePageData();
  695. const pageData = await frameData.processor.getPageData();
  696. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  697. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  698. this.stats.addAll(pageData);
  699. } else {
  700. this.stats.add("discarded", "frames", 1);
  701. }
  702. }
  703. }
  704. }
  705. }));
  706. }
  707. }
  708. async htmlImports(initialization) {
  709. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  710. if (!this.relImportProcessors) {
  711. this.relImportProcessors = new Map();
  712. }
  713. await Promise.all(linkElements.map(async linkElement => {
  714. if (initialization) {
  715. const resourceURL = linkElement.href;
  716. const options = Object.create(this.options);
  717. options.insertSingleFileComment = false;
  718. options.insertFaviconLink = false;
  719. options.doc = null;
  720. options.win = null;
  721. options.url = resourceURL;
  722. if (resourceURL) {
  723. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  724. const processor = new PageProcessor(options);
  725. this.relImportProcessors.set(linkElement, processor);
  726. await processor.loadPage();
  727. return processor.initialize();
  728. }
  729. }
  730. } else {
  731. linkElement.setAttribute("href", EMPTY_DATA_URI);
  732. const processor = this.relImportProcessors.get(linkElement);
  733. if (processor) {
  734. this.stats.add("processed", "HTML imports", 1);
  735. this.relImportProcessors.delete(linkElement);
  736. const pageData = await processor.getPageData();
  737. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  738. this.stats.addAll(pageData);
  739. } else {
  740. this.stats.add("discarded", "HTML imports", 1);
  741. }
  742. }
  743. }));
  744. }
  745. async attributeStyles(initialization) {
  746. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  747. let stylesheetContent = element.getAttribute("style");
  748. if (initialization) {
  749. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  750. } else {
  751. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  752. }
  753. element.setAttribute("style", stylesheetContent);
  754. }));
  755. }
  756. async linkStylesheets() {
  757. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  758. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  759. const styleElement = this.doc.createElement("style");
  760. styleElement.textContent = stylesheetContent;
  761. linkElement.parentElement.replaceChild(styleElement, linkElement);
  762. }));
  763. }
  764. }
  765. // ---------
  766. // DomHelper
  767. // ---------
  768. const REGEXP_AMP = /&/g;
  769. const REGEXP_NBSP = /\u00a0/g;
  770. const REGEXP_START_TAG = /</g;
  771. const REGEXP_END_TAG = />/g;
  772. const REGEXP_URL_HASH = /(#.+?)$/;
  773. const PREFIX_DATA_URI_IMAGE = "data:image/";
  774. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  775. class DomProcessorHelper {
  776. static async getFilename(options, content) {
  777. let filename = options.filenameTemplate;
  778. const date = new Date();
  779. const url = new URL(options.url);
  780. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  781. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  782. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  783. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  784. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  785. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  786. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  787. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  788. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  789. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  790. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  791. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  792. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  793. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  794. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  795. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  796. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  797. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  798. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  799. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  800. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  801. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  802. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  803. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  804. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  805. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  806. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  807. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  808. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  809. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  810. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  811. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  812. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  813. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  814. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  815. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  816. if (!options.backgroundSave) {
  817. filename = filename.replace(/\//g, "_");
  818. }
  819. if (filename.length > 192) {
  820. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  821. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  822. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  823. }
  824. if (!filename) {
  825. filename = "Unnamed page";
  826. }
  827. return filename;
  828. }
  829. static setFrameEmptySrc(frameElement) {
  830. if (frameElement.tagName == "OBJECT") {
  831. frameElement.setAttribute("data", "data:text/html,");
  832. } else {
  833. frameElement.setAttribute("srcdoc", "");
  834. frameElement.removeAttribute("src");
  835. }
  836. }
  837. static setFrameContent(frameElement, content) {
  838. if (frameElement.tagName == "OBJECT") {
  839. frameElement.setAttribute("data", "data:text/html," + content);
  840. } else {
  841. frameElement.setAttribute("srcdoc", content);
  842. frameElement.removeAttribute("src");
  843. }
  844. }
  845. static isolateElements(rootElement) {
  846. rootElement.querySelectorAll("*").forEach(element => {
  847. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  848. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  849. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  850. element.remove();
  851. }
  852. });
  853. isolateParentElements(rootElement.parentElement, rootElement);
  854. function isolateParentElements(parentElement, element) {
  855. if (parentElement) {
  856. Array.from(parentElement.childNodes).forEach(node => {
  857. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  858. node.remove();
  859. }
  860. });
  861. }
  862. element = element.parentElement;
  863. if (element && element.parentElement) {
  864. isolateParentElements(element.parentElement, element);
  865. }
  866. }
  867. }
  868. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  869. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  870. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  871. const imports = DomUtil.getImportFunctions(stylesheetContent);
  872. await Promise.all(imports.map(async cssImport => {
  873. const match = DomUtil.matchImport(cssImport);
  874. if (match) {
  875. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  876. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  877. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  878. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  879. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  880. if (stylesheetContent.includes(cssImport)) {
  881. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  882. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  883. }
  884. }
  885. }
  886. }));
  887. return stylesheetContent;
  888. }
  889. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  890. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  891. urlFunctions.map(urlFunction => {
  892. const originalResourceURL = DomUtil.matchURL(urlFunction);
  893. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  894. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  895. const resolvedURL = new URL(resourceURL, baseURI).href;
  896. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  897. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  898. }
  899. } else {
  900. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  901. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  902. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  903. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  904. }
  905. }
  906. }
  907. });
  908. return stylesheetContent;
  909. }
  910. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  911. resourceURL = DomUtil.normalizeURL(resourceURL);
  912. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  913. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  914. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  915. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  916. return stylesheetContent;
  917. }
  918. }
  919. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  920. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  921. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  922. let indexVariable = 0;
  923. await Promise.all(urlFunctions.map(async urlFunction => {
  924. const originalResourceURL = DomUtil.matchURL(urlFunction);
  925. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  926. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  927. const dataURI = await batchRequest.addURL(resourceURL);
  928. const regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  929. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  930. const functions = stylesheetContent.match(regExpUrlFunction);
  931. if (functions && functions.length > 1) {
  932. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  933. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  934. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + variableName + ")");
  935. indexVariable++;
  936. } else {
  937. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  938. }
  939. } else {
  940. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  941. }
  942. }
  943. }));
  944. if (indexVariable) {
  945. stylesheetContent = ":root{" + stylesheetContent;
  946. }
  947. return stylesheetContent;
  948. }
  949. static async processAttribute(resourceElements, attributeName, baseURI, batchRequest) {
  950. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  951. let resourceURL = resourceElement.getAttribute(attributeName);
  952. if (resourceURL) {
  953. resourceURL = DomUtil.normalizeURL(resourceURL);
  954. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  955. try {
  956. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  957. resourceElement.setAttribute(attributeName, dataURI);
  958. } catch (error) {
  959. /* ignored */
  960. }
  961. }
  962. }
  963. }));
  964. }
  965. static async processXLinks(resourceElements, baseURI, batchRequest) {
  966. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  967. const originalResourceURL = resourceElement.getAttribute("xlink:href");
  968. if (originalResourceURL) {
  969. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  970. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  971. try {
  972. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  973. const DOMParser = DOM.getParser();
  974. if (DOMParser) {
  975. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  976. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  977. if (hashMatch && hashMatch[0]) {
  978. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  979. if (symbolElement) {
  980. resourceElement.setAttribute("xlink:href", hashMatch[0]);
  981. resourceElement.parentElement.appendChild(symbolElement);
  982. }
  983. } else {
  984. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  985. }
  986. } else {
  987. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  988. }
  989. } catch (error) {
  990. /* ignored */
  991. }
  992. }
  993. }
  994. }));
  995. }
  996. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  997. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  998. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  999. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1000. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1001. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1002. try {
  1003. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  1004. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1005. } catch (error) {
  1006. /* ignored */
  1007. }
  1008. }
  1009. }));
  1010. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1011. }));
  1012. }
  1013. }
  1014. // -------
  1015. // DomUtil
  1016. // -------
  1017. const DATA_URI_PREFIX = "data:";
  1018. const BLOB_URI_PREFIX = "blob:";
  1019. const HTTP_URI_PREFIX = /^https?:\/\//;
  1020. const ABOUT_BLANK_URI = "about:blank";
  1021. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1022. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1023. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1024. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1025. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1026. 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;
  1027. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1028. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1029. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1030. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1031. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1032. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1033. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1034. class DomUtil {
  1035. static normalizeURL(url) {
  1036. if (url.startsWith(DATA_URI_PREFIX)) {
  1037. return url;
  1038. } else {
  1039. return url.split("#")[0];
  1040. }
  1041. }
  1042. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1043. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1044. if (template.match(replaceRegExp)) {
  1045. let value = await valueGetter();
  1046. if (!dontReplaceSlash) {
  1047. value = value.replace(/\/+/g, "_");
  1048. }
  1049. return template.replace(replaceRegExp, value);
  1050. }
  1051. return template;
  1052. }
  1053. static getLastSegment(url) {
  1054. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1055. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1056. if (!lastSegment) {
  1057. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1058. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1059. }
  1060. if (!lastSegment) {
  1061. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1062. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1063. }
  1064. if (!lastSegment) {
  1065. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1066. }
  1067. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1068. return lastSegment;
  1069. }
  1070. static getRegExp(string) {
  1071. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1072. }
  1073. static getUrlFunctions(stylesheetContent) {
  1074. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1075. }
  1076. static getImportFunctions(stylesheetContent) {
  1077. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1078. }
  1079. static matchURL(stylesheetContent) {
  1080. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1081. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1082. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1083. return match && match[1];
  1084. }
  1085. static testValidPath(resourceURL) {
  1086. 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));
  1087. }
  1088. static matchImport(stylesheetContent) {
  1089. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1090. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1091. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1092. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1093. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1094. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1095. if (match) {
  1096. const [, resourceURL, media] = match;
  1097. return { resourceURL, media };
  1098. }
  1099. }
  1100. static removeCssComments(stylesheetContent) {
  1101. let start, end;
  1102. do {
  1103. start = stylesheetContent.indexOf("/*");
  1104. end = stylesheetContent.indexOf("*/", start + 2);
  1105. if (start != -1 && end != -1) {
  1106. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1107. }
  1108. } while (start != -1 && end != -1);
  1109. return stylesheetContent;
  1110. }
  1111. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1112. if (mediaQuery) {
  1113. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1114. } else {
  1115. return stylesheetContent;
  1116. }
  1117. }
  1118. }
  1119. function log(...args) {
  1120. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1121. }
  1122. // -----
  1123. // Stats
  1124. // -----
  1125. const STATS_DEFAULT_VALUES = {
  1126. discarded: {
  1127. "HTML bytes": 0,
  1128. "hidden elements": 0,
  1129. "HTML imports": 0,
  1130. scripts: 0,
  1131. objects: 0,
  1132. "audio sources": 0,
  1133. "video sources": 0,
  1134. frames: 0,
  1135. "CSS rules": 0,
  1136. canvas: 0,
  1137. stylesheets: 0,
  1138. resources: 0,
  1139. medias: 0
  1140. },
  1141. processed: {
  1142. "HTML bytes": 0,
  1143. "hidden elements": 0,
  1144. "HTML imports": 0,
  1145. scripts: 0,
  1146. objects: 0,
  1147. "audio sources": 0,
  1148. "video sources": 0,
  1149. frames: 0,
  1150. "CSS rules": 0,
  1151. canvas: 0,
  1152. stylesheets: 0,
  1153. resources: 0,
  1154. medias: 0
  1155. }
  1156. };
  1157. class Stats {
  1158. constructor(options) {
  1159. this.options = options;
  1160. if (options.displayStats) {
  1161. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1162. }
  1163. }
  1164. set(type, subType, value) {
  1165. if (this.options.displayStats) {
  1166. this.data[type][subType] = value;
  1167. }
  1168. }
  1169. add(type, subType, value) {
  1170. if (this.options.displayStats) {
  1171. this.data[type][subType] += value;
  1172. }
  1173. }
  1174. addAll(pageData) {
  1175. if (this.options.displayStats) {
  1176. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1177. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1178. }
  1179. }
  1180. }
  1181. return { getClass };
  1182. })();