single-file-core.js 52 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276
  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. sync: [
  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: "compressHTML", action: "compressHTML" },
  77. { option: "insertFaviconLink", action: "insertFaviconLink" },
  78. { action: "resolveHrefs" },
  79. { action: "replaceCanvasElements" },
  80. { action: "insertFonts" },
  81. { option: "removeHiddenElements", action: "removeHiddenElements" }
  82. ],
  83. async: [
  84. { action: "inlineStylesheets" },
  85. { action: "linkStylesheets" },
  86. { action: "attributeStyles" },
  87. { option: "!removeFrames", action: "frames" },
  88. { option: "!removeImports", action: "htmlImports" }
  89. ]
  90. }, {
  91. sync: [
  92. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  93. { option: "compressHTML", action: "compressHTML" },
  94. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  95. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  96. ],
  97. async: [
  98. { action: "inlineStylesheets" },
  99. { action: "attributeStyles" },
  100. { action: "pageResources" },
  101. { option: "!removeScripts", action: "scripts" }
  102. ]
  103. }, {
  104. sync: [
  105. { option: "lazyLoadImages", action: "lazyLoadImages" },
  106. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  107. { option: "compressCSS", action: "compressCSS" }
  108. ],
  109. async: [
  110. { option: "!removeFrames", action: "frames" },
  111. { option: "!removeImports", action: "htmlImports" },
  112. ]
  113. }, {
  114. sync: [
  115. { option: "compressHTML", action: "postCompressHTML" },
  116. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  117. { action: "removeDefaultHeadTags" }
  118. ]
  119. }];
  120. class PageProcessor {
  121. constructor(options) {
  122. this.options = options;
  123. this.options.url = this.options.url || this.options.doc.location.href;
  124. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  125. this.batchRequest = new BatchRequest();
  126. this.processor = new DOMProcessor(options, this.batchRequest);
  127. if (this.options.doc) {
  128. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  129. this.options.canvasData = docData.canvasData;
  130. this.options.fontsData = docData.fontsData;
  131. this.options.stylesheetContents = docData.stylesheetContents;
  132. this.options.imageData = docData.imageData;
  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. console.log("-- STARTED STAGE --", step); // eslint-disable-line no-console
  174. }
  175. STAGES[step].sync.forEach(task => {
  176. let startTime;
  177. if (DEBUG) {
  178. startTime = Date.now();
  179. console.log("STARTED", task.action); // eslint-disable-line no-console
  180. }
  181. this.executeTask(task, !step);
  182. if (DEBUG) {
  183. console.log("ENDED", task.action, Date.now() - startTime); // eslint-disable-line no-console
  184. }
  185. });
  186. if (STAGES[step].async) {
  187. return await Promise.all(STAGES[step].async.map(task => {
  188. let startTime;
  189. if (DEBUG) {
  190. startTime = Date.now();
  191. console.log("STARTED", task.action, "..."); // eslint-disable-line no-console
  192. }
  193. const promise = this.executeTask(task, !step);
  194. if (DEBUG) {
  195. promise.then(() => console.log("ENDED", task.action, Date.now() - startTime)); // eslint-disable-line no-console
  196. }
  197. return promise;
  198. }));
  199. }
  200. if (DEBUG) {
  201. console.log("-- ENDED STAGE --", step); // eslint-disable-line no-console
  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. class DOMProcessor {
  257. constructor(options, batchRequest) {
  258. this.options = options;
  259. this.stats = new Stats(options);
  260. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  261. this.batchRequest = batchRequest;
  262. }
  263. initialize() {
  264. this.maxResources = this.batchRequest.getMaxResources();
  265. if (!this.options.removeFrames) {
  266. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  267. }
  268. this.stats.set("processed", "resources", this.maxResources);
  269. }
  270. async loadPage(pageContent) {
  271. if (!pageContent || this.options.saveRawPage) {
  272. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  273. }
  274. this.doc = DOM.createDoc(pageContent, this.baseURI);
  275. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  276. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  277. await DOMProcessor.loadEscapedFragmentPage();
  278. }
  279. }
  280. async loadEscapedFragmentPage() {
  281. if (this.baseURI.includes("?")) {
  282. this.baseURI += "&";
  283. } else {
  284. this.baseURI += "?";
  285. }
  286. this.baseURI += ESCAPED_FRAGMENT;
  287. await this.loadPage();
  288. }
  289. async getPageData() {
  290. DOM.postProcessDoc(this.doc, this.options);
  291. if (this.options.selected) {
  292. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  293. if (rootElement) {
  294. DomProcessorHelper.isolateElements(rootElement);
  295. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  296. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  297. }
  298. }
  299. const titleElement = this.doc.querySelector("title");
  300. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  301. this.options.info = {};
  302. const descriptionElement = this.doc.querySelector("meta[name=description]");
  303. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  304. this.options.info.lang = this.doc.documentElement.lang;
  305. const authorElement = this.doc.querySelector("meta[name=author]");
  306. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  307. const creatorElement = this.doc.querySelector("meta[name=creator]");
  308. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  309. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  310. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  311. const url = new URL(this.baseURI);
  312. let size;
  313. if (this.options.displayStats) {
  314. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  315. }
  316. const content = DOM.serialize(this.doc, this.options.compressHTML);
  317. if (this.options.displayStats) {
  318. const contentSize = DOM.getContentSize(content);
  319. this.stats.set("processed", "HTML bytes", contentSize);
  320. this.stats.add("discarded", "HTML bytes", size - contentSize);
  321. }
  322. const filename = await DomProcessorHelper.getFilename(this.options, content);
  323. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  324. return {
  325. stats: this.stats.data,
  326. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  327. filename,
  328. content
  329. };
  330. }
  331. setInputValues() {
  332. this.doc.querySelectorAll("input").forEach(input => {
  333. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  334. if (value) {
  335. input.setAttribute("value", value);
  336. }
  337. });
  338. this.doc.querySelectorAll("textarea").forEach(textarea => {
  339. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  340. if (value) {
  341. textarea.textContent = value;
  342. }
  343. });
  344. this.doc.querySelectorAll("select").forEach(select => {
  345. select.querySelectorAll("option").forEach(option => {
  346. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  347. if (selected) {
  348. option.setAttribute("selected", "");
  349. }
  350. });
  351. });
  352. }
  353. lazyLoadImages() {
  354. DOM.lazyLoad(this.doc);
  355. }
  356. removeDiscardedResources() {
  357. 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\"])");
  358. this.stats.set("discarded", "objects", objectElements.length);
  359. this.stats.set("processed", "objects", objectElements.length);
  360. objectElements.forEach(element => element.remove());
  361. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  362. replacedAttributeValue.forEach(element => {
  363. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  364. if (relValue.length) {
  365. element.setAttribute("rel", relValue);
  366. } else {
  367. element.remove();
  368. }
  369. });
  370. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  371. if (this.options.compressHTML) {
  372. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  373. }
  374. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  375. if (this.options.removeScripts) {
  376. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  377. this.doc.querySelectorAll("[href]").forEach(element => {
  378. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  379. element.removeAttribute("href");
  380. }
  381. });
  382. this.doc.querySelectorAll("[src]").forEach(element => {
  383. if (element.src && element.src.match(/^\s*javascript:/)) {
  384. element.removeAttribute("src");
  385. }
  386. });
  387. }
  388. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  389. this.stats.set("processed", "audio sources", audioSourceElements.length);
  390. if (this.options.removeAudioSrc) {
  391. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  392. audioSourceElements.forEach(element => element.removeAttribute("src"));
  393. }
  394. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  395. this.stats.set("processed", "video sources", videoSourceElements.length);
  396. if (this.options.removeVideoSrc) {
  397. this.stats.set("discarded", "video sources", videoSourceElements.length);
  398. videoSourceElements.forEach(element => element.removeAttribute("src"));
  399. }
  400. }
  401. removeDefaultHeadTags() {
  402. this.doc.querySelectorAll("base").forEach(element => element.remove());
  403. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  404. this.doc.head.querySelector("meta[charset]").remove();
  405. }
  406. }
  407. removeUIElements() {
  408. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  409. }
  410. removeScripts() {
  411. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  412. this.stats.set("discarded", "scripts", scriptElements.length);
  413. this.stats.set("processed", "scripts", scriptElements.length);
  414. scriptElements.forEach(element => element.remove());
  415. }
  416. removeFrames() {
  417. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  418. this.stats.set("discarded", "frames", frameElements.length);
  419. this.stats.set("processed", "frames", frameElements.length);
  420. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  421. }
  422. removeImports() {
  423. const importElements = this.doc.querySelectorAll("link[rel=import]");
  424. this.stats.set("discarded", "HTML imports", importElements.length);
  425. this.stats.set("processed", "HTML imports", importElements.length);
  426. importElements.forEach(element => element.remove());
  427. }
  428. resetCharsetMeta() {
  429. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  430. const metaElement = this.doc.createElement("meta");
  431. metaElement.setAttribute("charset", "utf-8");
  432. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  433. }
  434. insertFaviconLink() {
  435. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  436. if (!faviconElement) {
  437. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  438. }
  439. if (!faviconElement) {
  440. faviconElement = this.doc.createElement("link");
  441. faviconElement.setAttribute("type", "image/x-icon");
  442. faviconElement.setAttribute("rel", "shortcut icon");
  443. faviconElement.setAttribute("href", "/favicon.ico");
  444. }
  445. this.doc.head.appendChild(faviconElement);
  446. }
  447. resolveHrefs() {
  448. this.doc.querySelectorAll("[href]").forEach(element => {
  449. if (element.href) {
  450. const href = element.href.baseVal ? element.href.baseVal : element.href;
  451. const normalizedHref = DomUtil.normalizeURL(href);
  452. if (!normalizedHref || normalizedHref != this.baseURI) {
  453. element.setAttribute("href", href);
  454. }
  455. }
  456. });
  457. }
  458. removeUnusedStyles() {
  459. const stats = DOM.minifyCSS(this.doc);
  460. this.stats.set("processed", "CSS rules", stats.processed);
  461. this.stats.set("discarded", "CSS rules", stats.discarded);
  462. }
  463. removeAlternativeFonts() {
  464. DOM.minifyFonts(this.doc);
  465. }
  466. removeSrcSet() {
  467. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  468. const tagName = element.tagName.toLowerCase();
  469. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  470. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  471. element.removeAttribute(dataAttributeName);
  472. if (imageData) {
  473. if (tagName == "img") {
  474. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  475. element.removeAttribute("srcset");
  476. element.removeAttribute("sizes");
  477. element.src = imageData.source.src;
  478. }
  479. }
  480. if (tagName == "picture") {
  481. const imageElement = element.querySelector("img");
  482. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  483. imageElement.removeAttribute("srcset");
  484. imageElement.removeAttribute("sizes");
  485. imageElement.src = imageData.source.src;
  486. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  487. } else {
  488. if (imageData.sources) {
  489. element.querySelectorAll("source").forEach(sourceElement => {
  490. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  491. sourceElement.remove();
  492. }
  493. });
  494. const sourceElements = element.querySelectorAll("source");
  495. if (sourceElements.length) {
  496. const lastSourceElement = sourceElements[sourceElements.length - 1];
  497. if (imageElement) {
  498. if (lastSourceElement.src) {
  499. imageElement.src = lastSourceElement.src;
  500. } else {
  501. imageElement.removeAttribute("src");
  502. }
  503. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  504. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  505. } else {
  506. imageElement.removeAttribute("srcset");
  507. }
  508. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  509. }
  510. }
  511. }
  512. }
  513. }
  514. }
  515. });
  516. }
  517. postRemoveAlternativeFonts() {
  518. DOM.minifyFonts(this.doc, true);
  519. }
  520. removeHiddenElements() {
  521. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  522. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  523. this.stats.set("processed", "hidden elements", hiddenElements.length);
  524. hiddenElements.forEach(element => element.remove());
  525. }
  526. compressHTML() {
  527. let size;
  528. if (this.options.displayStats) {
  529. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  530. }
  531. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  532. if (this.options.displayStats) {
  533. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  534. }
  535. }
  536. postCompressHTML() {
  537. let size;
  538. if (this.options.displayStats) {
  539. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  540. }
  541. DOM.postMinifyHTML(this.doc);
  542. if (this.options.displayStats) {
  543. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  544. }
  545. }
  546. removeAlternativeMedias() {
  547. const stats = DOM.minifyMedias(this.doc);
  548. this.stats.set("processed", "medias", stats.processed);
  549. this.stats.set("discarded", "medias", stats.discarded);
  550. }
  551. compressCSS() {
  552. this.doc.querySelectorAll("style").forEach(styleElement => {
  553. if (styleElement) {
  554. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  555. }
  556. });
  557. this.doc.querySelectorAll("[style]").forEach(element => {
  558. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  559. });
  560. }
  561. insertSingleFileComment() {
  562. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  563. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  564. }
  565. replaceCanvasElements() {
  566. if (this.options.canvasData) {
  567. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  568. const canvasData = this.options.canvasData[indexCanvasElement];
  569. if (canvasData) {
  570. const imgElement = this.doc.createElement("img");
  571. imgElement.setAttribute("src", canvasData.dataURI);
  572. Array.from(canvasElement.attributes).forEach(attribute => {
  573. if (attribute.value) {
  574. imgElement.setAttribute(attribute.name, attribute.value);
  575. }
  576. });
  577. if (!imgElement.width && canvasData.width) {
  578. imgElement.style.pixelWidth = canvasData.width;
  579. }
  580. if (!imgElement.height && canvasData.height) {
  581. imgElement.style.pixelHeight = canvasData.height;
  582. }
  583. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  584. this.stats.add("processed", "canvas", 1);
  585. }
  586. });
  587. }
  588. }
  589. insertFonts() {
  590. if (this.options.fontsData && this.options.fontsData.length) {
  591. let stylesheetContent = "";
  592. this.options.fontsData.forEach(fontStyles => {
  593. if (fontStyles["font-family"] && fontStyles.src) {
  594. stylesheetContent += "@font-face{";
  595. let stylesContent = "";
  596. Object.keys(fontStyles).forEach(fontStyle => {
  597. if (stylesContent) {
  598. stylesContent += ";";
  599. }
  600. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  601. });
  602. stylesheetContent += stylesContent + "}";
  603. }
  604. });
  605. if (stylesheetContent) {
  606. const styleElement = this.doc.createElement("style");
  607. styleElement.textContent = stylesheetContent;
  608. const existingStyleElement = this.doc.querySelector("style");
  609. if (existingStyleElement) {
  610. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  611. } else {
  612. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  613. }
  614. }
  615. }
  616. }
  617. replaceStyleContents() {
  618. if (this.options.stylesheetContents) {
  619. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  620. if (this.options.stylesheetContents[styleIndex]) {
  621. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  622. }
  623. });
  624. }
  625. }
  626. async pageResources() {
  627. const resourcePromises = [
  628. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI, this.batchRequest),
  629. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", this.baseURI, this.batchRequest),
  630. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], embed[src*=\".svg\"]"), "src", this.baseURI, this.batchRequest),
  631. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI, this.batchRequest),
  632. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI, this.batchRequest),
  633. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", this.baseURI, this.batchRequest),
  634. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  635. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.batchRequest)
  636. ];
  637. if (!this.options.removeAudioSrc) {
  638. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.batchRequest));
  639. }
  640. if (!this.options.removeVideoSrc) {
  641. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.batchRequest));
  642. }
  643. if (this.options.lazyLoadImages) {
  644. const imageSelectors = DOM.lazyLoaderImageSelectors();
  645. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], this.baseURI, this.batchRequest)));
  646. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], this.baseURI, this.batchRequest)));
  647. }
  648. await resourcePromises;
  649. }
  650. async inlineStylesheets(initialization) {
  651. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  652. if (!initialization) {
  653. this.stats.add("processed", "stylesheets", 1);
  654. }
  655. let stylesheetContent = styleElement.textContent;
  656. if (initialization) {
  657. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  658. } else {
  659. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  660. }
  661. styleElement.textContent = stylesheetContent;
  662. }));
  663. }
  664. async scripts() {
  665. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  666. if (scriptElement.src) {
  667. this.stats.add("processed", "scripts", 1);
  668. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  669. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  670. }
  671. scriptElement.removeAttribute("src");
  672. }));
  673. }
  674. async frames(initialization) {
  675. if (this.options.framesData) {
  676. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  677. await Promise.all(frameElements.map(async frameElement => {
  678. DomProcessorHelper.setFrameEmptySrc(frameElement);
  679. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  680. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  681. if (frameWindowId) {
  682. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  683. if (frameData) {
  684. if (initialization) {
  685. const options = Object.create(this.options);
  686. options.insertSingleFileComment = false;
  687. options.insertFaviconLink = false;
  688. options.doc = null;
  689. options.win = null;
  690. options.url = frameData.baseURI;
  691. options.windowId = frameWindowId;
  692. if (frameData.content) {
  693. options.content = frameData.content;
  694. options.canvasData = frameData.canvasData;
  695. options.stylesheetContents = frameData.stylesheetContents;
  696. options.currentSrcImages = frameData.currentSrcImages;
  697. frameData.processor = new PageProcessor(options);
  698. frameData.frameElement = frameElement;
  699. await frameData.processor.loadPage();
  700. await frameData.processor.initialize();
  701. frameData.maxResources = this.batchRequest.getMaxResources();
  702. }
  703. } else {
  704. if (frameData.processor) {
  705. this.stats.add("processed", "frames", 1);
  706. await frameData.processor.preparePageData();
  707. const pageData = await frameData.processor.getPageData();
  708. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  709. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  710. this.stats.addAll(pageData);
  711. } else {
  712. this.stats.add("discarded", "frames", 1);
  713. }
  714. }
  715. }
  716. }
  717. }));
  718. }
  719. }
  720. async htmlImports(initialization) {
  721. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  722. if (!this.relImportProcessors) {
  723. this.relImportProcessors = new Map();
  724. }
  725. await Promise.all(linkElements.map(async linkElement => {
  726. if (initialization) {
  727. const resourceURL = linkElement.href;
  728. const options = Object.create(this.options);
  729. options.insertSingleFileComment = false;
  730. options.insertFaviconLink = false;
  731. options.doc = null;
  732. options.win = null;
  733. options.url = resourceURL;
  734. if (resourceURL) {
  735. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  736. const processor = new PageProcessor(options);
  737. this.relImportProcessors.set(linkElement, processor);
  738. await processor.loadPage();
  739. return processor.initialize();
  740. }
  741. }
  742. } else {
  743. linkElement.setAttribute("href", EMPTY_DATA_URI);
  744. const processor = this.relImportProcessors.get(linkElement);
  745. if (processor) {
  746. this.stats.add("processed", "HTML imports", 1);
  747. this.relImportProcessors.delete(linkElement);
  748. const pageData = await processor.getPageData();
  749. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  750. this.stats.addAll(pageData);
  751. } else {
  752. this.stats.add("discarded", "HTML imports", 1);
  753. }
  754. }
  755. }));
  756. }
  757. async attributeStyles(initialization) {
  758. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  759. let stylesheetContent = element.getAttribute("style");
  760. if (initialization) {
  761. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  762. } else {
  763. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  764. }
  765. element.setAttribute("style", stylesheetContent);
  766. }));
  767. }
  768. async linkStylesheets() {
  769. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  770. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  771. const styleElement = this.doc.createElement("style");
  772. styleElement.textContent = stylesheetContent;
  773. linkElement.parentElement.replaceChild(styleElement, linkElement);
  774. }));
  775. }
  776. }
  777. // ---------
  778. // DomHelper
  779. // ---------
  780. const REGEXP_AMP = /&/g;
  781. const REGEXP_NBSP = /\u00a0/g;
  782. const REGEXP_START_TAG = /</g;
  783. const REGEXP_END_TAG = />/g;
  784. const REGEXP_URL_HASH = /(#.+?)$/;
  785. const PREFIX_DATA_URI_IMAGE = "data:image/";
  786. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  787. class DomProcessorHelper {
  788. static async getFilename(options, content) {
  789. let filename = options.filenameTemplate;
  790. const date = new Date();
  791. const url = new URL(options.url);
  792. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  793. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  794. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  795. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  796. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  797. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  798. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  799. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  800. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  801. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  802. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  803. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  804. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  805. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  806. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  807. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  808. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  809. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  810. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  811. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  812. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  813. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  814. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  815. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  816. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  817. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  818. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  819. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  820. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  821. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  822. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  823. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  824. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  825. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  826. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  827. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  828. if (!options.backgroundSave) {
  829. filename = filename.replace(/\//g, "_");
  830. }
  831. if (filename.length > 192) {
  832. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  833. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  834. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  835. }
  836. if (!filename) {
  837. filename = "Unnamed page";
  838. }
  839. return filename;
  840. }
  841. static setFrameEmptySrc(frameElement) {
  842. if (frameElement.tagName == "OBJECT") {
  843. frameElement.setAttribute("data", "data:text/html,");
  844. } else {
  845. frameElement.setAttribute("srcdoc", "");
  846. frameElement.removeAttribute("src");
  847. }
  848. }
  849. static setFrameContent(frameElement, content) {
  850. if (frameElement.tagName == "OBJECT") {
  851. frameElement.setAttribute("data", "data:text/html," + content);
  852. } else {
  853. frameElement.setAttribute("srcdoc", content);
  854. frameElement.removeAttribute("src");
  855. }
  856. }
  857. static isolateElements(rootElement) {
  858. rootElement.querySelectorAll("*").forEach(element => {
  859. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  860. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  861. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  862. element.remove();
  863. }
  864. });
  865. isolateParentElements(rootElement.parentElement, rootElement);
  866. function isolateParentElements(parentElement, element) {
  867. if (parentElement) {
  868. Array.from(parentElement.childNodes).forEach(node => {
  869. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  870. node.remove();
  871. }
  872. });
  873. }
  874. element = element.parentElement;
  875. if (element && element.parentElement) {
  876. isolateParentElements(element.parentElement, element);
  877. }
  878. }
  879. }
  880. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  881. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  882. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  883. const imports = DomUtil.getImportFunctions(stylesheetContent);
  884. await Promise.all(imports.map(async cssImport => {
  885. const match = DomUtil.matchImport(cssImport);
  886. if (match) {
  887. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  888. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  889. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  890. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  891. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  892. if (stylesheetContent.includes(cssImport)) {
  893. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  894. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  895. }
  896. }
  897. }
  898. }));
  899. return stylesheetContent;
  900. }
  901. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  902. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  903. urlFunctions.map(urlFunction => {
  904. const originalResourceURL = DomUtil.matchURL(urlFunction);
  905. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  906. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  907. const resolvedURL = new URL(resourceURL, baseURI).href;
  908. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  909. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  910. }
  911. } else {
  912. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  913. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  914. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  915. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  916. }
  917. }
  918. }
  919. });
  920. return stylesheetContent;
  921. }
  922. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  923. resourceURL = DomUtil.normalizeURL(resourceURL);
  924. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  925. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  926. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  927. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  928. return stylesheetContent;
  929. }
  930. }
  931. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  932. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  933. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  934. let indexVariable = 0;
  935. await Promise.all(urlFunctions.map(async urlFunction => {
  936. const originalResourceURL = DomUtil.matchURL(urlFunction);
  937. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  938. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  939. const dataURI = await batchRequest.addURL(resourceURL);
  940. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  941. const functions = stylesheetContent.match(DomUtil.getRegExp(urlFunction));
  942. if (functions && functions.length > 1) {
  943. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  944. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  945. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), "var(" + variableName + ")");
  946. indexVariable++;
  947. } else {
  948. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  949. }
  950. } else {
  951. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  952. }
  953. }
  954. }));
  955. if (indexVariable) {
  956. stylesheetContent = ":root{" + stylesheetContent;
  957. }
  958. return stylesheetContent;
  959. }
  960. static async processAttribute(resourceElements, attributeName, baseURI, batchRequest) {
  961. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  962. let resourceURL = resourceElement.getAttribute(attributeName);
  963. if (resourceURL) {
  964. resourceURL = DomUtil.normalizeURL(resourceURL);
  965. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  966. try {
  967. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  968. resourceElement.setAttribute(attributeName, dataURI);
  969. } catch (error) {
  970. /* ignored */
  971. }
  972. }
  973. }
  974. }));
  975. }
  976. static async processXLinks(resourceElements, baseURI, batchRequest) {
  977. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  978. const originalResourceURL = resourceElement.getAttribute("xlink:href");
  979. if (originalResourceURL) {
  980. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  981. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  982. try {
  983. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  984. const DOMParser = DOM.getParser();
  985. if (DOMParser) {
  986. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  987. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  988. if (hashMatch && hashMatch[0]) {
  989. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  990. if (symbolElement) {
  991. resourceElement.setAttribute("xlink:href", hashMatch[0]);
  992. resourceElement.parentElement.appendChild(symbolElement);
  993. }
  994. } else {
  995. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  996. }
  997. } else {
  998. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  999. }
  1000. } catch (error) {
  1001. /* ignored */
  1002. }
  1003. }
  1004. }
  1005. }));
  1006. }
  1007. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1008. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1009. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1010. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1011. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1012. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1013. try {
  1014. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  1015. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1016. } catch (error) {
  1017. /* ignored */
  1018. }
  1019. }
  1020. }));
  1021. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1022. }));
  1023. }
  1024. }
  1025. // -------
  1026. // DomUtil
  1027. // -------
  1028. const DATA_URI_PREFIX = "data:";
  1029. const BLOB_URI_PREFIX = "blob:";
  1030. const HTTP_URI_PREFIX = /^https?:\/\//;
  1031. const ABOUT_BLANK_URI = "about:blank";
  1032. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1033. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1034. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1035. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1036. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1037. 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;
  1038. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1039. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1040. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1041. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1042. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1043. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1044. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1045. class DomUtil {
  1046. static normalizeURL(url) {
  1047. if (url.startsWith(DATA_URI_PREFIX)) {
  1048. return url;
  1049. } else {
  1050. return url.split("#")[0];
  1051. }
  1052. }
  1053. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1054. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1055. if (template.match(replaceRegExp)) {
  1056. let value = await valueGetter();
  1057. if (!dontReplaceSlash) {
  1058. value = value.replace(/\/+/g, "_");
  1059. }
  1060. return template.replace(replaceRegExp, value);
  1061. }
  1062. return template;
  1063. }
  1064. static getLastSegment(url) {
  1065. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1066. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1067. if (!lastSegment) {
  1068. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1069. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1070. }
  1071. if (!lastSegment) {
  1072. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1073. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1074. }
  1075. if (!lastSegment) {
  1076. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1077. }
  1078. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1079. return lastSegment;
  1080. }
  1081. static getRegExp(string) {
  1082. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1083. }
  1084. static getUrlFunctions(stylesheetContent) {
  1085. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1086. }
  1087. static getImportFunctions(stylesheetContent) {
  1088. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1089. }
  1090. static matchURL(stylesheetContent) {
  1091. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1092. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1093. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1094. return match && match[1];
  1095. }
  1096. static testValidPath(resourceURL) {
  1097. 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));
  1098. }
  1099. static matchImport(stylesheetContent) {
  1100. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1101. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1102. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1103. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1104. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1105. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1106. if (match) {
  1107. const [, resourceURL, media] = match;
  1108. return { resourceURL, media };
  1109. }
  1110. }
  1111. static removeCssComments(stylesheetContent) {
  1112. let start, end;
  1113. do {
  1114. start = stylesheetContent.indexOf("/*");
  1115. end = stylesheetContent.indexOf("*/", start + 2);
  1116. if (start != -1 && end != -1) {
  1117. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1118. }
  1119. } while (start != -1 && end != -1);
  1120. return stylesheetContent;
  1121. }
  1122. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1123. if (mediaQuery) {
  1124. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1125. } else {
  1126. return stylesheetContent;
  1127. }
  1128. }
  1129. }
  1130. // -----
  1131. // Stats
  1132. // -----
  1133. const STATS_DEFAULT_VALUES = {
  1134. discarded: {
  1135. "HTML bytes": 0,
  1136. "hidden elements": 0,
  1137. "HTML imports": 0,
  1138. scripts: 0,
  1139. objects: 0,
  1140. "audio sources": 0,
  1141. "video sources": 0,
  1142. frames: 0,
  1143. "CSS rules": 0,
  1144. canvas: 0,
  1145. stylesheets: 0,
  1146. resources: 0,
  1147. medias: 0
  1148. },
  1149. processed: {
  1150. "HTML bytes": 0,
  1151. "hidden elements": 0,
  1152. "HTML imports": 0,
  1153. scripts: 0,
  1154. objects: 0,
  1155. "audio sources": 0,
  1156. "video sources": 0,
  1157. frames: 0,
  1158. "CSS rules": 0,
  1159. canvas: 0,
  1160. stylesheets: 0,
  1161. resources: 0,
  1162. medias: 0
  1163. }
  1164. };
  1165. class Stats {
  1166. constructor(options) {
  1167. this.options = options;
  1168. if (options.displayStats) {
  1169. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1170. }
  1171. }
  1172. set(type, subType, value) {
  1173. if (this.options.displayStats) {
  1174. this.data[type][subType] = value;
  1175. }
  1176. }
  1177. add(type, subType, value) {
  1178. if (this.options.displayStats) {
  1179. this.data[type][subType] += value;
  1180. }
  1181. }
  1182. addAll(pageData) {
  1183. if (this.options.displayStats) {
  1184. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1185. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1186. }
  1187. }
  1188. }
  1189. return { getClass };
  1190. })();