single-file-core.js 53 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293
  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: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  94. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  95. ],
  96. parallel: [
  97. { action: "inlineStylesheets" },
  98. { action: "attributeStyles" },
  99. { action: "pageResources" },
  100. { option: "!removeScripts", action: "scripts" }
  101. ]
  102. }, {
  103. sequential: [
  104. { option: "lazyLoadImages", action: "lazyLoadImages" },
  105. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  106. { option: "compressCSS", action: "compressCSS" }
  107. ],
  108. parallel: [
  109. { option: "!removeFrames", action: "frames" },
  110. { option: "!removeImports", action: "htmlImports" },
  111. ]
  112. }, {
  113. sequential: [
  114. { option: "compressHTML", action: "compressHTML" },
  115. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  116. { action: "removeDefaultHeadTags" }
  117. ]
  118. }];
  119. class PageProcessor {
  120. constructor(options) {
  121. this.options = options;
  122. this.options.url = this.options.url || this.options.doc.location.href;
  123. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  124. this.batchRequest = new BatchRequest();
  125. this.processor = new DOMProcessor(options, this.batchRequest);
  126. if (this.options.doc) {
  127. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  128. this.options.canvasData = docData.canvasData;
  129. this.options.fontsData = docData.fontsData;
  130. this.options.stylesheetContents = docData.stylesheetContents;
  131. this.options.imageData = docData.imageData;
  132. this.options.postersData = docData.postersData;
  133. }
  134. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  135. this.onprogress = options.onprogress || (() => { });
  136. }
  137. async loadPage() {
  138. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  139. await this.processor.loadPage(this.options.content);
  140. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  141. }
  142. async initialize() {
  143. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  144. await this.executeStage(0);
  145. this.pendingPromises = this.executeStage(1);
  146. if (this.options.doc) {
  147. DOM.postProcessDoc(this.options.doc, this.options);
  148. this.options.doc = null;
  149. this.options.win = null;
  150. }
  151. }
  152. async preparePageData() {
  153. if (!this.options.windowId) {
  154. this.processor.initialize(this.batchRequest);
  155. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  156. }
  157. await this.batchRequest.run(details => {
  158. details.pageURL = this.options.url;
  159. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  160. }, this.options);
  161. await this.pendingPromises;
  162. await this.executeStage(2);
  163. await this.executeStage(3);
  164. }
  165. async getPageData() {
  166. if (!this.options.windowId) {
  167. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  168. }
  169. return this.processor.getPageData();
  170. }
  171. async executeStage(step) {
  172. if (DEBUG) {
  173. log("**** STARTED STAGE", step, "****");
  174. }
  175. STAGES[step].sequential.forEach(task => {
  176. let startTime;
  177. if (DEBUG) {
  178. startTime = Date.now();
  179. log(" -- STARTED task =", task.action);
  180. }
  181. this.executeTask(task, !step);
  182. if (DEBUG) {
  183. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  184. }
  185. });
  186. if (STAGES[step].parallel) {
  187. return await Promise.all(STAGES[step].parallel.map(task => {
  188. let startTime;
  189. if (DEBUG) {
  190. startTime = Date.now();
  191. log(" // STARTED task =", task.action);
  192. }
  193. const promise = this.executeTask(task, !step);
  194. if (DEBUG) {
  195. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  196. }
  197. return promise;
  198. }));
  199. }
  200. if (DEBUG) {
  201. log("**** ENDED STAGE", step, "****");
  202. }
  203. }
  204. executeTask(task, initialization) {
  205. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  206. return this.processor[task.action](initialization);
  207. }
  208. }
  209. }
  210. // --------
  211. // BatchRequest
  212. // --------
  213. class BatchRequest {
  214. constructor() {
  215. this.requests = new Map();
  216. }
  217. async addURL(resourceURL, asDataURI = true) {
  218. return new Promise((resolve, reject) => {
  219. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  220. const resourceRequests = this.requests.get(requestKey);
  221. if (resourceRequests) {
  222. resourceRequests.push({ resolve, reject });
  223. } else {
  224. this.requests.set(requestKey, [{ resolve, reject }]);
  225. }
  226. });
  227. }
  228. getMaxResources() {
  229. return Array.from(this.requests.keys()).length;
  230. }
  231. async run(onloadListener, options) {
  232. const resourceURLs = Array.from(this.requests.keys());
  233. let indexResource = 0;
  234. return Promise.all(resourceURLs.map(async requestKey => {
  235. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  236. const resourceRequests = this.requests.get(requestKey);
  237. try {
  238. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  239. indexResource = indexResource + 1;
  240. onloadListener({ index: indexResource, url: resourceURL });
  241. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  242. } catch (error) {
  243. indexResource = indexResource + 1;
  244. onloadListener({ index: indexResource, url: resourceURL });
  245. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  246. }
  247. this.requests.delete(requestKey);
  248. }));
  249. }
  250. }
  251. // ------------
  252. // DOMProcessor
  253. // ------------
  254. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  255. const EMPTY_DATA_URI = "data:base64,";
  256. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  257. class DOMProcessor {
  258. constructor(options, batchRequest) {
  259. this.options = options;
  260. this.stats = new Stats(options);
  261. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  262. this.batchRequest = batchRequest;
  263. }
  264. initialize() {
  265. this.maxResources = this.batchRequest.getMaxResources();
  266. if (!this.options.removeFrames) {
  267. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  268. }
  269. this.stats.set("processed", "resources", this.maxResources);
  270. }
  271. async loadPage(pageContent) {
  272. if (!pageContent || this.options.saveRawPage) {
  273. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  274. }
  275. this.doc = DOM.createDoc(pageContent, this.baseURI);
  276. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  277. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  278. await DOMProcessor.loadEscapedFragmentPage();
  279. }
  280. }
  281. async loadEscapedFragmentPage() {
  282. if (this.baseURI.includes("?")) {
  283. this.baseURI += "&";
  284. } else {
  285. this.baseURI += "?";
  286. }
  287. this.baseURI += ESCAPED_FRAGMENT;
  288. await this.loadPage();
  289. }
  290. async getPageData() {
  291. DOM.postProcessDoc(this.doc, this.options);
  292. if (this.options.selected) {
  293. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  294. if (rootElement) {
  295. DomProcessorHelper.isolateElements(rootElement);
  296. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  297. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  298. }
  299. }
  300. const titleElement = this.doc.querySelector("title");
  301. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  302. this.options.info = {};
  303. const descriptionElement = this.doc.querySelector("meta[name=description]");
  304. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  305. this.options.info.lang = this.doc.documentElement.lang;
  306. const authorElement = this.doc.querySelector("meta[name=author]");
  307. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  308. const creatorElement = this.doc.querySelector("meta[name=creator]");
  309. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  310. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  311. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  312. const url = new URL(this.baseURI);
  313. let size;
  314. if (this.options.displayStats) {
  315. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  316. }
  317. const content = DOM.serialize(this.doc, this.options.compressHTML);
  318. if (this.options.displayStats) {
  319. const contentSize = DOM.getContentSize(content);
  320. this.stats.set("processed", "HTML bytes", contentSize);
  321. this.stats.add("discarded", "HTML bytes", size - contentSize);
  322. }
  323. const filename = await DomProcessorHelper.getFilename(this.options, content);
  324. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  325. return {
  326. stats: this.stats.data,
  327. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  328. filename,
  329. content
  330. };
  331. }
  332. setInputValues() {
  333. this.doc.querySelectorAll("input").forEach(input => {
  334. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  335. if (value) {
  336. input.setAttribute("value", value);
  337. }
  338. });
  339. this.doc.querySelectorAll("textarea").forEach(textarea => {
  340. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  341. if (value) {
  342. textarea.textContent = value;
  343. }
  344. });
  345. this.doc.querySelectorAll("select").forEach(select => {
  346. select.querySelectorAll("option").forEach(option => {
  347. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  348. if (selected) {
  349. option.setAttribute("selected", "");
  350. }
  351. });
  352. });
  353. }
  354. lazyLoadImages() {
  355. DOM.lazyLoad(this.doc);
  356. }
  357. removeDiscardedResources() {
  358. 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\"])");
  359. this.stats.set("discarded", "objects", objectElements.length);
  360. this.stats.set("processed", "objects", objectElements.length);
  361. objectElements.forEach(element => element.remove());
  362. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  363. replacedAttributeValue.forEach(element => {
  364. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  365. if (relValue.length) {
  366. element.setAttribute("rel", relValue);
  367. } else {
  368. element.remove();
  369. }
  370. });
  371. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  372. if (this.options.compressHTML) {
  373. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  374. }
  375. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  376. if (this.options.removeScripts) {
  377. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  378. this.doc.querySelectorAll("[href]").forEach(element => {
  379. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  380. element.removeAttribute("href");
  381. }
  382. });
  383. this.doc.querySelectorAll("[src]").forEach(element => {
  384. if (element.src && element.src.match(/^\s*javascript:/)) {
  385. element.removeAttribute("src");
  386. }
  387. });
  388. }
  389. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  390. this.stats.set("processed", "audio sources", audioSourceElements.length);
  391. if (this.options.removeAudioSrc) {
  392. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  393. audioSourceElements.forEach(element => element.removeAttribute("src"));
  394. }
  395. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  396. this.stats.set("processed", "video sources", videoSourceElements.length);
  397. if (this.options.removeVideoSrc) {
  398. this.stats.set("discarded", "video sources", videoSourceElements.length);
  399. videoSourceElements.forEach(element => element.removeAttribute("src"));
  400. }
  401. }
  402. removeDefaultHeadTags() {
  403. this.doc.querySelectorAll("base").forEach(element => element.remove());
  404. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  405. this.doc.head.querySelector("meta[charset]").remove();
  406. }
  407. }
  408. removeUIElements() {
  409. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  410. }
  411. removeScripts() {
  412. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  413. this.stats.set("discarded", "scripts", scriptElements.length);
  414. this.stats.set("processed", "scripts", scriptElements.length);
  415. scriptElements.forEach(element => element.remove());
  416. }
  417. removeFrames() {
  418. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  419. this.stats.set("discarded", "frames", frameElements.length);
  420. this.stats.set("processed", "frames", frameElements.length);
  421. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  422. }
  423. removeImports() {
  424. const importElements = this.doc.querySelectorAll("link[rel=import]");
  425. this.stats.set("discarded", "HTML imports", importElements.length);
  426. this.stats.set("processed", "HTML imports", importElements.length);
  427. importElements.forEach(element => element.remove());
  428. }
  429. resetCharsetMeta() {
  430. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  431. const metaElement = this.doc.createElement("meta");
  432. metaElement.setAttribute("charset", "utf-8");
  433. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  434. }
  435. insertFaviconLink() {
  436. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  437. if (!faviconElement) {
  438. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  439. }
  440. if (!faviconElement) {
  441. faviconElement = this.doc.createElement("link");
  442. faviconElement.setAttribute("type", "image/x-icon");
  443. faviconElement.setAttribute("rel", "shortcut icon");
  444. faviconElement.setAttribute("href", "/favicon.ico");
  445. }
  446. this.doc.head.appendChild(faviconElement);
  447. }
  448. resolveHrefs() {
  449. this.doc.querySelectorAll("[href]").forEach(element => {
  450. if (element.href) {
  451. const href = element.href.baseVal ? element.href.baseVal : element.href;
  452. const normalizedHref = DomUtil.normalizeURL(href);
  453. if (!normalizedHref || normalizedHref != this.baseURI) {
  454. element.setAttribute("href", href);
  455. }
  456. }
  457. });
  458. }
  459. removeUnusedStyles() {
  460. const stats = DOM.minifyCSS(this.doc);
  461. this.stats.set("processed", "CSS rules", stats.processed);
  462. this.stats.set("discarded", "CSS rules", stats.discarded);
  463. }
  464. removeAlternativeFonts() {
  465. DOM.minifyFonts(this.doc);
  466. }
  467. removeSrcSet() {
  468. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  469. const tagName = element.tagName.toLowerCase();
  470. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  471. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  472. element.removeAttribute(dataAttributeName);
  473. if (imageData) {
  474. if (tagName == "img") {
  475. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  476. element.removeAttribute("srcset");
  477. element.removeAttribute("sizes");
  478. element.src = imageData.source.src;
  479. }
  480. }
  481. if (tagName == "picture") {
  482. const imageElement = element.querySelector("img");
  483. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  484. imageElement.removeAttribute("srcset");
  485. imageElement.removeAttribute("sizes");
  486. imageElement.src = imageData.source.src;
  487. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  488. } else {
  489. if (imageData.sources) {
  490. element.querySelectorAll("source").forEach(sourceElement => {
  491. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  492. sourceElement.remove();
  493. }
  494. });
  495. const sourceElements = element.querySelectorAll("source");
  496. if (sourceElements.length) {
  497. const lastSourceElement = sourceElements[sourceElements.length - 1];
  498. if (imageElement) {
  499. if (lastSourceElement.src) {
  500. imageElement.src = lastSourceElement.src;
  501. } else {
  502. imageElement.removeAttribute("src");
  503. }
  504. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  505. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  506. } else {
  507. imageElement.removeAttribute("srcset");
  508. }
  509. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  510. }
  511. }
  512. }
  513. }
  514. }
  515. }
  516. });
  517. }
  518. postRemoveAlternativeFonts() {
  519. DOM.minifyFonts(this.doc, true);
  520. }
  521. removeHiddenElements() {
  522. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  523. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  524. this.stats.set("processed", "hidden elements", hiddenElements.length);
  525. hiddenElements.forEach(element => element.remove());
  526. }
  527. compressHTML() {
  528. let size;
  529. if (this.options.displayStats) {
  530. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  531. }
  532. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  533. if (this.options.displayStats) {
  534. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  535. }
  536. }
  537. removeAlternativeMedias() {
  538. const stats = DOM.minifyMedias(this.doc);
  539. this.stats.set("processed", "medias", stats.processed);
  540. this.stats.set("discarded", "medias", stats.discarded);
  541. }
  542. compressCSS() {
  543. this.doc.querySelectorAll("style").forEach(styleElement => {
  544. if (styleElement) {
  545. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  546. }
  547. });
  548. this.doc.querySelectorAll("[style]").forEach(element => {
  549. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  550. });
  551. }
  552. insertSingleFileComment() {
  553. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  554. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  555. }
  556. replaceCanvasElements() {
  557. if (this.options.canvasData) {
  558. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  559. const canvasData = this.options.canvasData[indexCanvasElement];
  560. if (canvasData) {
  561. const imgElement = this.doc.createElement("img");
  562. imgElement.setAttribute("src", canvasData.dataURI);
  563. Array.from(canvasElement.attributes).forEach(attribute => {
  564. if (attribute.value) {
  565. imgElement.setAttribute(attribute.name, attribute.value);
  566. }
  567. });
  568. if (!imgElement.width && canvasData.width) {
  569. imgElement.style.pixelWidth = canvasData.width;
  570. }
  571. if (!imgElement.height && canvasData.height) {
  572. imgElement.style.pixelHeight = canvasData.height;
  573. }
  574. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  575. this.stats.add("processed", "canvas", 1);
  576. }
  577. });
  578. }
  579. }
  580. insertFonts() {
  581. if (this.options.fontsData && this.options.fontsData.length) {
  582. let stylesheetContent = "";
  583. this.options.fontsData.forEach(fontStyles => {
  584. if (fontStyles["font-family"] && fontStyles.src) {
  585. stylesheetContent += "@font-face{";
  586. let stylesContent = "";
  587. Object.keys(fontStyles).forEach(fontStyle => {
  588. if (stylesContent) {
  589. stylesContent += ";";
  590. }
  591. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  592. });
  593. stylesheetContent += stylesContent + "}";
  594. }
  595. });
  596. if (stylesheetContent) {
  597. const styleElement = this.doc.createElement("style");
  598. styleElement.textContent = stylesheetContent;
  599. const existingStyleElement = this.doc.querySelector("style");
  600. if (existingStyleElement) {
  601. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  602. } else {
  603. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  604. }
  605. }
  606. }
  607. }
  608. replaceStyleContents() {
  609. if (this.options.stylesheetContents) {
  610. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  611. if (this.options.stylesheetContents[styleIndex]) {
  612. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  613. }
  614. });
  615. }
  616. }
  617. insertVideoPosters() {
  618. if (this.options.postersData) {
  619. this.doc.querySelectorAll("video[src], video > source[src]").forEach((videoElement, videoIndex) => {
  620. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  621. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  622. }
  623. });
  624. }
  625. }
  626. async pageResources() {
  627. const resourcePromises = [
  628. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  629. 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),
  630. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image]"), "src", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  631. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest),
  632. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  633. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  634. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  635. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  636. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)
  637. ];
  638. if (!this.options.removeAudioSrc) {
  639. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", PREFIX_DATA_URI_AUDIO, this.baseURI, this.batchRequest));
  640. }
  641. if (!this.options.removeVideoSrc) {
  642. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", PREFIX_DATA_URI_VIDEO, this.baseURI, this.batchRequest));
  643. }
  644. if (this.options.lazyLoadImages) {
  645. const imageSelectors = DOM.lazyLoaderImageSelectors();
  646. 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)));
  647. 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)));
  648. }
  649. await resourcePromises;
  650. }
  651. async inlineStylesheets(initialization) {
  652. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  653. if (!initialization) {
  654. this.stats.add("processed", "stylesheets", 1);
  655. }
  656. let stylesheetContent = styleElement.textContent;
  657. if (initialization) {
  658. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  659. } else {
  660. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  661. }
  662. styleElement.textContent = stylesheetContent;
  663. }));
  664. }
  665. async scripts() {
  666. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  667. if (scriptElement.src) {
  668. this.stats.add("processed", "scripts", 1);
  669. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  670. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  671. }
  672. scriptElement.removeAttribute("src");
  673. }));
  674. }
  675. async frames(initialization) {
  676. if (this.options.framesData) {
  677. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  678. await Promise.all(frameElements.map(async frameElement => {
  679. DomProcessorHelper.setFrameEmptySrc(frameElement);
  680. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  681. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  682. if (frameWindowId) {
  683. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  684. if (frameData) {
  685. if (initialization) {
  686. const options = Object.create(this.options);
  687. options.insertSingleFileComment = false;
  688. options.insertFaviconLink = false;
  689. options.doc = null;
  690. options.win = null;
  691. options.url = frameData.baseURI;
  692. options.windowId = frameWindowId;
  693. if (frameData.content) {
  694. options.content = frameData.content;
  695. options.canvasData = frameData.canvasData;
  696. options.stylesheetContents = frameData.stylesheetContents;
  697. options.currentSrcImages = frameData.currentSrcImages;
  698. frameData.processor = new PageProcessor(options);
  699. frameData.frameElement = frameElement;
  700. await frameData.processor.loadPage();
  701. await frameData.processor.initialize();
  702. frameData.maxResources = this.batchRequest.getMaxResources();
  703. }
  704. } else {
  705. if (frameData.processor) {
  706. this.stats.add("processed", "frames", 1);
  707. await frameData.processor.preparePageData();
  708. const pageData = await frameData.processor.getPageData();
  709. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  710. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  711. this.stats.addAll(pageData);
  712. } else {
  713. this.stats.add("discarded", "frames", 1);
  714. }
  715. }
  716. }
  717. }
  718. }));
  719. }
  720. }
  721. async htmlImports(initialization) {
  722. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  723. if (!this.relImportProcessors) {
  724. this.relImportProcessors = new Map();
  725. }
  726. await Promise.all(linkElements.map(async linkElement => {
  727. if (initialization) {
  728. const resourceURL = linkElement.href;
  729. const options = Object.create(this.options);
  730. options.insertSingleFileComment = false;
  731. options.insertFaviconLink = false;
  732. options.doc = null;
  733. options.win = null;
  734. options.url = resourceURL;
  735. if (resourceURL) {
  736. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  737. const processor = new PageProcessor(options);
  738. this.relImportProcessors.set(linkElement, processor);
  739. await processor.loadPage();
  740. return processor.initialize();
  741. }
  742. }
  743. } else {
  744. linkElement.setAttribute("href", EMPTY_DATA_URI);
  745. const processor = this.relImportProcessors.get(linkElement);
  746. if (processor) {
  747. this.stats.add("processed", "HTML imports", 1);
  748. this.relImportProcessors.delete(linkElement);
  749. const pageData = await processor.getPageData();
  750. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  751. this.stats.addAll(pageData);
  752. } else {
  753. this.stats.add("discarded", "HTML imports", 1);
  754. }
  755. }
  756. }));
  757. }
  758. async attributeStyles(initialization) {
  759. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  760. let stylesheetContent = element.getAttribute("style");
  761. if (initialization) {
  762. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  763. } else {
  764. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  765. }
  766. element.setAttribute("style", stylesheetContent);
  767. }));
  768. }
  769. async linkStylesheets() {
  770. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  771. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  772. const styleElement = this.doc.createElement("style");
  773. styleElement.textContent = stylesheetContent;
  774. linkElement.parentElement.replaceChild(styleElement, linkElement);
  775. }));
  776. }
  777. }
  778. // ---------
  779. // DomHelper
  780. // ---------
  781. const REGEXP_AMP = /&/g;
  782. const REGEXP_NBSP = /\u00a0/g;
  783. const REGEXP_START_TAG = /</g;
  784. const REGEXP_END_TAG = />/g;
  785. const REGEXP_URL_HASH = /(#.+?)$/;
  786. const PREFIX_DATA_URI_IMAGE = "data:image/";
  787. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  788. const PREFIX_DATA_URI_VIDEO = "data:video/";
  789. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  790. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  791. class DomProcessorHelper {
  792. static async getFilename(options, content) {
  793. let filename = options.filenameTemplate;
  794. const date = new Date();
  795. const url = new URL(options.url);
  796. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  797. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  798. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  799. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  800. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  801. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  802. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  803. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  804. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  805. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  806. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  807. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  808. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  809. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  810. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  811. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  812. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  813. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  814. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  815. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  816. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  817. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  818. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  819. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  820. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  821. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  822. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  823. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  824. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  825. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  826. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  827. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  828. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  829. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  830. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  831. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  832. if (!options.backgroundSave) {
  833. filename = filename.replace(/\//g, "_");
  834. }
  835. if (filename.length > 192) {
  836. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  837. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  838. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  839. }
  840. if (!filename) {
  841. filename = "Unnamed page";
  842. }
  843. return filename;
  844. }
  845. static setFrameEmptySrc(frameElement) {
  846. if (frameElement.tagName == "OBJECT") {
  847. frameElement.setAttribute("data", "data:text/html,");
  848. } else {
  849. frameElement.setAttribute("srcdoc", "");
  850. frameElement.removeAttribute("src");
  851. }
  852. }
  853. static setFrameContent(frameElement, content) {
  854. if (frameElement.tagName == "OBJECT") {
  855. frameElement.setAttribute("data", "data:text/html," + content);
  856. } else {
  857. frameElement.setAttribute("srcdoc", content);
  858. frameElement.removeAttribute("src");
  859. }
  860. }
  861. static isolateElements(rootElement) {
  862. rootElement.querySelectorAll("*").forEach(element => {
  863. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  864. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  865. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  866. element.remove();
  867. }
  868. });
  869. isolateParentElements(rootElement.parentElement, rootElement);
  870. function isolateParentElements(parentElement, element) {
  871. if (parentElement) {
  872. Array.from(parentElement.childNodes).forEach(node => {
  873. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  874. node.remove();
  875. }
  876. });
  877. }
  878. element = element.parentElement;
  879. if (element && element.parentElement) {
  880. isolateParentElements(element.parentElement, element);
  881. }
  882. }
  883. }
  884. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  885. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  886. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  887. const imports = DomUtil.getImportFunctions(stylesheetContent);
  888. await Promise.all(imports.map(async cssImport => {
  889. const match = DomUtil.matchImport(cssImport);
  890. if (match) {
  891. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  892. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  893. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  894. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  895. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  896. if (stylesheetContent.includes(cssImport)) {
  897. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  898. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  899. }
  900. }
  901. }
  902. }));
  903. return stylesheetContent;
  904. }
  905. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  906. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  907. urlFunctions.map(urlFunction => {
  908. const originalResourceURL = DomUtil.matchURL(urlFunction);
  909. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  910. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  911. const resolvedURL = new URL(resourceURL, baseURI).href;
  912. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  913. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  914. }
  915. } else {
  916. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  917. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  918. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  919. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  920. }
  921. }
  922. }
  923. });
  924. return stylesheetContent;
  925. }
  926. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  927. resourceURL = DomUtil.normalizeURL(resourceURL);
  928. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  929. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  930. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  931. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  932. return stylesheetContent;
  933. }
  934. }
  935. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  936. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  937. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  938. let indexVariable = 0;
  939. await Promise.all(urlFunctions.map(async urlFunction => {
  940. const originalResourceURL = DomUtil.matchURL(urlFunction);
  941. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  942. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  943. const dataURI = await batchRequest.addURL(resourceURL);
  944. const regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  945. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  946. const functions = stylesheetContent.match(regExpUrlFunction);
  947. if (functions && functions.length > 1) {
  948. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  949. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  950. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + variableName + ")");
  951. indexVariable++;
  952. } else {
  953. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  954. }
  955. } else {
  956. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  957. }
  958. }
  959. }));
  960. if (indexVariable) {
  961. stylesheetContent = ":root{" + stylesheetContent;
  962. }
  963. return stylesheetContent;
  964. }
  965. static async processAttribute(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  966. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  967. let resourceURL = resourceElement.getAttribute(attributeName);
  968. if (resourceURL) {
  969. resourceURL = DomUtil.normalizeURL(resourceURL);
  970. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  971. try {
  972. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  973. if (dataURI.startsWith(prefixDataURI) || dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE)) {
  974. resourceElement.setAttribute(attributeName, dataURI);
  975. } else {
  976. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  977. }
  978. } catch (error) {
  979. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  980. }
  981. }
  982. }
  983. }));
  984. }
  985. static async processXLinks(resourceElements, baseURI, batchRequest) {
  986. const attributeName = "xlink:href";
  987. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  988. const originalResourceURL = resourceElement.getAttribute(attributeName);
  989. if (originalResourceURL) {
  990. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  991. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  992. try {
  993. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  994. const DOMParser = DOM.getParser();
  995. if (DOMParser) {
  996. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  997. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  998. if (hashMatch && hashMatch[0]) {
  999. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1000. if (symbolElement) {
  1001. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1002. resourceElement.parentElement.appendChild(symbolElement);
  1003. }
  1004. } else {
  1005. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1006. }
  1007. } else {
  1008. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1009. }
  1010. } catch (error) {
  1011. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1012. }
  1013. }
  1014. }
  1015. }));
  1016. }
  1017. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  1018. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1019. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1020. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1021. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1022. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1023. try {
  1024. let dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  1025. if (!dataURI.startsWith(prefixDataURI) && !dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE)) {
  1026. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1027. }
  1028. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1029. } catch (error) {
  1030. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1031. }
  1032. }
  1033. }));
  1034. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1035. }));
  1036. }
  1037. }
  1038. // -------
  1039. // DomUtil
  1040. // -------
  1041. const DATA_URI_PREFIX = "data:";
  1042. const BLOB_URI_PREFIX = "blob:";
  1043. const HTTP_URI_PREFIX = /^https?:\/\//;
  1044. const ABOUT_BLANK_URI = "about:blank";
  1045. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1046. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1047. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1048. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1049. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1050. 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;
  1051. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1052. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1053. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1054. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1055. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1056. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1057. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1058. class DomUtil {
  1059. static normalizeURL(url) {
  1060. if (url.startsWith(DATA_URI_PREFIX)) {
  1061. return url;
  1062. } else {
  1063. return url.split("#")[0];
  1064. }
  1065. }
  1066. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1067. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1068. if (template.match(replaceRegExp)) {
  1069. let value = await valueGetter();
  1070. if (!dontReplaceSlash) {
  1071. value = value.replace(/\/+/g, "_");
  1072. }
  1073. return template.replace(replaceRegExp, value);
  1074. }
  1075. return template;
  1076. }
  1077. static getLastSegment(url) {
  1078. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1079. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1080. if (!lastSegment) {
  1081. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1082. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1083. }
  1084. if (!lastSegment) {
  1085. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1086. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1087. }
  1088. if (!lastSegment) {
  1089. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1090. }
  1091. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1092. return lastSegment;
  1093. }
  1094. static getRegExp(string) {
  1095. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1096. }
  1097. static getUrlFunctions(stylesheetContent) {
  1098. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1099. }
  1100. static getImportFunctions(stylesheetContent) {
  1101. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1102. }
  1103. static matchURL(stylesheetContent) {
  1104. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1105. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1106. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1107. return match && match[1];
  1108. }
  1109. static testValidPath(resourceURL) {
  1110. 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));
  1111. }
  1112. static matchImport(stylesheetContent) {
  1113. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1114. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1115. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1116. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1117. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1118. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1119. if (match) {
  1120. const [, resourceURL, media] = match;
  1121. return { resourceURL, media };
  1122. }
  1123. }
  1124. static removeCssComments(stylesheetContent) {
  1125. let start, end;
  1126. do {
  1127. start = stylesheetContent.indexOf("/*");
  1128. end = stylesheetContent.indexOf("*/", start + 2);
  1129. if (start != -1 && end != -1) {
  1130. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1131. }
  1132. } while (start != -1 && end != -1);
  1133. return stylesheetContent;
  1134. }
  1135. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1136. if (mediaQuery) {
  1137. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1138. } else {
  1139. return stylesheetContent;
  1140. }
  1141. }
  1142. }
  1143. function log(...args) {
  1144. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1145. }
  1146. // -----
  1147. // Stats
  1148. // -----
  1149. const STATS_DEFAULT_VALUES = {
  1150. discarded: {
  1151. "HTML bytes": 0,
  1152. "hidden elements": 0,
  1153. "HTML imports": 0,
  1154. scripts: 0,
  1155. objects: 0,
  1156. "audio sources": 0,
  1157. "video sources": 0,
  1158. frames: 0,
  1159. "CSS rules": 0,
  1160. canvas: 0,
  1161. stylesheets: 0,
  1162. resources: 0,
  1163. medias: 0
  1164. },
  1165. processed: {
  1166. "HTML bytes": 0,
  1167. "hidden elements": 0,
  1168. "HTML imports": 0,
  1169. scripts: 0,
  1170. objects: 0,
  1171. "audio sources": 0,
  1172. "video sources": 0,
  1173. frames: 0,
  1174. "CSS rules": 0,
  1175. canvas: 0,
  1176. stylesheets: 0,
  1177. resources: 0,
  1178. medias: 0
  1179. }
  1180. };
  1181. class Stats {
  1182. constructor(options) {
  1183. this.options = options;
  1184. if (options.displayStats) {
  1185. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1186. }
  1187. }
  1188. set(type, subType, value) {
  1189. if (this.options.displayStats) {
  1190. this.data[type][subType] = value;
  1191. }
  1192. }
  1193. add(type, subType, value) {
  1194. if (this.options.displayStats) {
  1195. this.data[type][subType] += value;
  1196. }
  1197. }
  1198. addAll(pageData) {
  1199. if (this.options.displayStats) {
  1200. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1201. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1202. }
  1203. }
  1204. }
  1205. return { getClass };
  1206. })();