single-file-core.js 49 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201
  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. let Download, DOM, URL, sessionId = 0;
  24. function getClass(...args) {
  25. [Download, DOM, URL] = args;
  26. return class {
  27. constructor(options) {
  28. this.options = options;
  29. options.sessionId = sessionId;
  30. sessionId++;
  31. this.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  32. this.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  33. }
  34. async initialize() {
  35. this.processor = new PageProcessor(this.options);
  36. await this.processor.loadPage();
  37. await this.processor.initialize();
  38. }
  39. async preparePageData() {
  40. await this.processor.preparePageData();
  41. }
  42. async getPageData() {
  43. return this.processor.getPageData();
  44. }
  45. };
  46. }
  47. // -------------
  48. // ProgressEvent
  49. // -------------
  50. const PAGE_LOADING = "page-loading";
  51. const PAGE_LOADED = "page-loaded";
  52. const RESOURCES_INITIALIZING = "resource-initializing";
  53. const RESOURCES_INITIALIZED = "resources-initialized";
  54. const RESOURCE_LOADED = "resource-loaded";
  55. const PAGE_ENDED = "page-ended";
  56. class ProgressEvent {
  57. constructor(type, details) {
  58. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  59. }
  60. }
  61. // -------------
  62. // PageProcessor
  63. // -------------
  64. const STAGES = [{
  65. sync: [
  66. { action: "removeUIElements" },
  67. { action: "replaceStyleContents" },
  68. { option: "removeSrcSet", action: "removeSrcSet" },
  69. { option: "removeFrames", action: "removeFrames" },
  70. { option: "removeImports", action: "removeImports" },
  71. { option: "removeScripts", action: "removeScripts" },
  72. { action: "removeDiscardedResources" },
  73. { action: "resetCharsetMeta" },
  74. { action: "setInputValues" },
  75. { option: "compressHTML", action: "compressHTML" },
  76. { option: "insertFaviconLink", action: "insertFaviconLink" },
  77. { action: "resolveHrefs" },
  78. { action: "replaceCanvasElements" },
  79. { option: "removeHiddenElements", action: "removeHiddenElements" }
  80. ],
  81. async: [
  82. { action: "inlineStylesheets" },
  83. { action: "linkStylesheets" },
  84. { action: "attributeStyles" },
  85. { option: "!removeFrames", action: "frames" },
  86. { option: "!removeImports", action: "htmlImports" }
  87. ]
  88. }, {
  89. sync: [
  90. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  91. { option: "compressHTML", action: "compressHTML" },
  92. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  93. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  94. ],
  95. async: [
  96. { action: "inlineStylesheets" },
  97. { action: "attributeStyles" },
  98. { action: "pageResources" },
  99. { option: "!removeScripts", action: "scripts" }
  100. ]
  101. }, {
  102. sync: [
  103. { option: "lazyLoadImages", action: "lazyLoadImages" },
  104. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  105. { option: "compressCSS", action: "compressCSS" }
  106. ],
  107. async: [
  108. { option: "!removeFrames", action: "frames" },
  109. { option: "!removeImports", action: "htmlImports" },
  110. ]
  111. }, {
  112. sync: [
  113. { option: "compressHTML", action: "postCompressHTML" },
  114. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  115. { action: "removeDefaultHeadTags" }
  116. ]
  117. }];
  118. class PageProcessor {
  119. constructor(options) {
  120. this.options = options;
  121. this.options.url = this.options.url || this.options.doc.location.href;
  122. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  123. this.batchRequest = new BatchRequest();
  124. this.processor = new DOMProcessor(options, this.batchRequest);
  125. if (this.options.doc) {
  126. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  127. this.options.canvasData = docData.canvasData;
  128. this.options.stylesheetContents = docData.stylesheetContents;
  129. this.options.imageData = docData.imageData;
  130. }
  131. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  132. this.onprogress = options.onprogress || (() => { });
  133. }
  134. async loadPage() {
  135. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  136. await this.processor.loadPage(this.options.content);
  137. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  138. }
  139. async initialize() {
  140. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  141. await this.executeStage(0);
  142. this.pendingPromises = this.executeStage(1);
  143. if (this.options.doc) {
  144. DOM.postProcessDoc(this.options.doc, this.options);
  145. this.options.doc = null;
  146. this.options.win = null;
  147. }
  148. }
  149. async preparePageData() {
  150. if (!this.options.windowId) {
  151. this.processor.initialize(this.batchRequest);
  152. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  153. }
  154. await this.batchRequest.run(details => {
  155. details.pageURL = this.options.url;
  156. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  157. }, this.options);
  158. await this.pendingPromises;
  159. await this.executeStage(2);
  160. await this.executeStage(3);
  161. }
  162. async getPageData() {
  163. if (!this.options.windowId) {
  164. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  165. }
  166. return this.processor.getPageData();
  167. }
  168. async executeStage(step) {
  169. STAGES[step].sync.forEach(task => {
  170. // const startTime = Date.now();
  171. this.executeTask(task, !step);
  172. // console.log(task.action, Date.now() - startTime);
  173. });
  174. if (STAGES[step].async) {
  175. return Promise.all(STAGES[step].async.map(task => {
  176. // const startTime = Date.now();
  177. const promise = this.executeTask(task, !step);
  178. // promise.then(() => console.log(task.action, Date.now() - startTime));
  179. return promise;
  180. }));
  181. }
  182. }
  183. executeTask(task, initialization) {
  184. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  185. return this.processor[task.action](initialization);
  186. }
  187. }
  188. }
  189. // --------
  190. // BatchRequest
  191. // --------
  192. class BatchRequest {
  193. constructor() {
  194. this.requests = new Map();
  195. }
  196. async addURL(resourceURL, asDataURI = true) {
  197. return new Promise((resolve, reject) => {
  198. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  199. const resourceRequests = this.requests.get(requestKey);
  200. if (resourceRequests) {
  201. resourceRequests.push({ resolve, reject });
  202. } else {
  203. this.requests.set(requestKey, [{ resolve, reject }]);
  204. }
  205. });
  206. }
  207. getMaxResources() {
  208. return Array.from(this.requests.keys()).length;
  209. }
  210. async run(onloadListener, options) {
  211. const resourceURLs = Array.from(this.requests.keys());
  212. let indexResource = 0;
  213. return Promise.all(resourceURLs.map(async requestKey => {
  214. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  215. const resourceRequests = this.requests.get(requestKey);
  216. try {
  217. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  218. indexResource = indexResource + 1;
  219. onloadListener({ index: indexResource, url: resourceURL });
  220. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  221. } catch (error) {
  222. indexResource = indexResource + 1;
  223. onloadListener({ index: indexResource, url: resourceURL });
  224. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  225. }
  226. this.requests.delete(requestKey);
  227. }));
  228. }
  229. }
  230. // ------------
  231. // DOMProcessor
  232. // ------------
  233. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  234. const EMPTY_DATA_URI = "data:base64,";
  235. class DOMProcessor {
  236. constructor(options, batchRequest) {
  237. this.options = options;
  238. this.stats = new Stats(options);
  239. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  240. this.batchRequest = batchRequest;
  241. }
  242. initialize() {
  243. this.maxResources = this.batchRequest.getMaxResources();
  244. if (!this.options.removeFrames) {
  245. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  246. }
  247. this.stats.set("processed", "resources", this.maxResources);
  248. }
  249. async loadPage(pageContent) {
  250. if (!pageContent || this.options.saveRawPage) {
  251. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  252. }
  253. this.doc = DOM.createDoc(pageContent, this.baseURI);
  254. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  255. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  256. await DOMProcessor.loadEscapedFragmentPage();
  257. }
  258. }
  259. async loadEscapedFragmentPage() {
  260. if (this.baseURI.includes("?")) {
  261. this.baseURI += "&";
  262. } else {
  263. this.baseURI += "?";
  264. }
  265. this.baseURI += ESCAPED_FRAGMENT;
  266. await this.loadPage();
  267. }
  268. async getPageData() {
  269. DOM.postProcessDoc(this.doc, this.options);
  270. if (this.options.selected) {
  271. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  272. if (rootElement) {
  273. DomProcessorHelper.isolateElements(rootElement);
  274. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  275. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  276. }
  277. }
  278. const titleElement = this.doc.querySelector("title");
  279. if (titleElement) {
  280. this.options.title = titleElement.textContent.trim();
  281. } else {
  282. this.options.title = "";
  283. }
  284. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  285. const url = new URL(this.baseURI);
  286. let size;
  287. if (this.options.displayStats) {
  288. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  289. }
  290. const content = DOM.serialize(this.doc, this.options.compressHTML);
  291. if (this.options.displayStats) {
  292. const contentSize = DOM.getContentSize(content);
  293. this.stats.set("processed", "HTML bytes", contentSize);
  294. this.stats.add("discarded", "HTML bytes", size - contentSize);
  295. }
  296. const filename = await DomProcessorHelper.getFilename(this.options, content);
  297. return {
  298. stats: this.stats.data,
  299. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  300. filename,
  301. content
  302. };
  303. }
  304. setInputValues() {
  305. this.doc.querySelectorAll("input").forEach(input => {
  306. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  307. if (value) {
  308. input.setAttribute("value", value);
  309. }
  310. });
  311. this.doc.querySelectorAll("textarea").forEach(textarea => {
  312. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  313. if (value) {
  314. textarea.textContent = value;
  315. }
  316. });
  317. this.doc.querySelectorAll("select").forEach(select => {
  318. select.querySelectorAll("option").forEach(option => {
  319. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  320. if (selected) {
  321. option.setAttribute("selected", "");
  322. }
  323. });
  324. });
  325. }
  326. lazyLoadImages() {
  327. DOM.lazyLoad(this.doc);
  328. }
  329. removeDiscardedResources() {
  330. 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\"])");
  331. this.stats.set("discarded", "objects", objectElements.length);
  332. this.stats.set("processed", "objects", objectElements.length);
  333. objectElements.forEach(element => element.remove());
  334. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  335. replacedAttributeValue.forEach(element => {
  336. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  337. if (relValue.length) {
  338. element.setAttribute("rel", relValue);
  339. } else {
  340. element.remove();
  341. }
  342. });
  343. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"], input[type=hidden]").forEach(element => element.remove());
  344. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  345. if (this.options.removeScripts) {
  346. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  347. this.doc.querySelectorAll("[href]").forEach(element => {
  348. if (element.href && element.href.match(/^\s*javascript:/)) {
  349. element.removeAttribute("href");
  350. }
  351. });
  352. this.doc.querySelectorAll("[src]").forEach(element => {
  353. if (element.src && element.src.match(/^\s*javascript:/)) {
  354. element.removeAttribute("src");
  355. }
  356. });
  357. }
  358. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  359. this.stats.set("processed", "audio sources", audioSourceElements.length);
  360. if (this.options.removeAudioSrc) {
  361. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  362. audioSourceElements.forEach(element => element.removeAttribute("src"));
  363. }
  364. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  365. this.stats.set("processed", "video sources", videoSourceElements.length);
  366. if (this.options.removeVideoSrc) {
  367. this.stats.set("discarded", "video sources", videoSourceElements.length);
  368. videoSourceElements.forEach(element => element.removeAttribute("src"));
  369. }
  370. }
  371. removeDefaultHeadTags() {
  372. this.doc.querySelectorAll("base").forEach(element => element.remove());
  373. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  374. this.doc.head.querySelector("meta[charset]").remove();
  375. }
  376. }
  377. removeUIElements() {
  378. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  379. }
  380. removeScripts() {
  381. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  382. this.stats.set("discarded", "scripts", scriptElements.length);
  383. this.stats.set("processed", "scripts", scriptElements.length);
  384. scriptElements.forEach(element => element.remove());
  385. }
  386. removeFrames() {
  387. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  388. this.stats.set("discarded", "frames", frameElements.length);
  389. this.stats.set("processed", "frames", frameElements.length);
  390. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  391. }
  392. removeImports() {
  393. const importElements = this.doc.querySelectorAll("link[rel=import]");
  394. this.stats.set("discarded", "HTML imports", importElements.length);
  395. this.stats.set("processed", "HTML imports", importElements.length);
  396. importElements.forEach(element => element.remove());
  397. }
  398. resetCharsetMeta() {
  399. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  400. const metaElement = this.doc.createElement("meta");
  401. metaElement.setAttribute("charset", "utf-8");
  402. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  403. }
  404. insertFaviconLink() {
  405. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  406. if (!faviconElement) {
  407. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  408. }
  409. if (!faviconElement) {
  410. faviconElement = this.doc.createElement("link");
  411. faviconElement.setAttribute("type", "image/x-icon");
  412. faviconElement.setAttribute("rel", "shortcut icon");
  413. faviconElement.setAttribute("href", "/favicon.ico");
  414. }
  415. this.doc.head.appendChild(faviconElement);
  416. }
  417. resolveHrefs() {
  418. this.doc.querySelectorAll("[href]").forEach(element => {
  419. if (element.href) {
  420. const href = element.href.baseVal ? element.href.baseVal : element.href;
  421. const normalizedHref = DomUtil.normalizeURL(href);
  422. if (!normalizedHref || normalizedHref != this.baseURI) {
  423. element.setAttribute("href", href);
  424. }
  425. }
  426. });
  427. }
  428. removeUnusedStyles() {
  429. const stats = DOM.minifyCSS(this.doc);
  430. this.stats.set("processed", "CSS rules", stats.processed);
  431. this.stats.set("discarded", "CSS rules", stats.discarded);
  432. }
  433. removeAlternativeFonts() {
  434. DOM.minifyFonts(this.doc);
  435. }
  436. removeSrcSet() {
  437. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  438. const tagName = element.tagName.toLowerCase();
  439. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  440. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  441. element.removeAttribute(dataAttributeName);
  442. if (imageData) {
  443. if (tagName == "img") {
  444. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  445. element.removeAttribute("srcset");
  446. element.removeAttribute("sizes");
  447. element.src = imageData.source.src;
  448. }
  449. }
  450. if (tagName == "picture") {
  451. const imageElement = element.querySelector("img");
  452. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  453. imageElement.removeAttribute("srcset");
  454. imageElement.removeAttribute("sizes");
  455. imageElement.src = imageData.source.src;
  456. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  457. } else {
  458. if (imageData.sources) {
  459. element.querySelectorAll("source").forEach(sourceElement => {
  460. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  461. sourceElement.remove();
  462. }
  463. });
  464. const sourceElements = element.querySelectorAll("source");
  465. if (sourceElements.length) {
  466. const lastSourceElement = sourceElements[sourceElements.length - 1];
  467. if (imageElement) {
  468. if (lastSourceElement.src) {
  469. imageElement.src = lastSourceElement.src;
  470. } else {
  471. imageElement.removeAttribute("src");
  472. }
  473. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  474. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  475. } else {
  476. imageElement.removeAttribute("srcset");
  477. }
  478. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  479. }
  480. }
  481. }
  482. }
  483. }
  484. }
  485. });
  486. }
  487. postRemoveAlternativeFonts() {
  488. DOM.minifyFonts(this.doc, true);
  489. }
  490. removeHiddenElements() {
  491. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  492. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  493. this.stats.set("processed", "hidden elements", hiddenElements.length);
  494. hiddenElements.forEach(element => element.remove());
  495. }
  496. compressHTML() {
  497. let size;
  498. if (this.options.displayStats) {
  499. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  500. }
  501. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  502. if (this.options.displayStats) {
  503. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  504. }
  505. }
  506. postCompressHTML() {
  507. let size;
  508. if (this.options.displayStats) {
  509. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  510. }
  511. DOM.postMinifyHTML(this.doc);
  512. if (this.options.displayStats) {
  513. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  514. }
  515. }
  516. removeAlternativeMedias() {
  517. const stats = DOM.minifyMedias(this.doc);
  518. this.stats.set("processed", "medias", stats.processed);
  519. this.stats.set("discarded", "medias", stats.discarded);
  520. }
  521. compressCSS() {
  522. this.doc.querySelectorAll("style").forEach(styleElement => {
  523. if (styleElement) {
  524. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  525. }
  526. });
  527. this.doc.querySelectorAll("[style]").forEach(element => {
  528. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  529. });
  530. }
  531. insertSingleFileComment() {
  532. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  533. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  534. }
  535. replaceCanvasElements() {
  536. if (this.options.canvasData) {
  537. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  538. const canvasData = this.options.canvasData[indexCanvasElement];
  539. if (canvasData) {
  540. const imgElement = this.doc.createElement("img");
  541. imgElement.setAttribute("src", canvasData.dataURI);
  542. Array.from(canvasElement.attributes).forEach(attribute => {
  543. if (attribute.value) {
  544. imgElement.setAttribute(attribute.name, attribute.value);
  545. }
  546. });
  547. if (!imgElement.width && canvasData.width) {
  548. imgElement.style.pixelWidth = canvasData.width;
  549. }
  550. if (!imgElement.height && canvasData.height) {
  551. imgElement.style.pixelHeight = canvasData.height;
  552. }
  553. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  554. this.stats.add("processed", "canvas", 1);
  555. }
  556. });
  557. }
  558. }
  559. replaceStyleContents() {
  560. if (this.options.stylesheetContents) {
  561. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  562. if (this.options.stylesheetContents[styleIndex]) {
  563. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  564. }
  565. });
  566. }
  567. }
  568. async pageResources() {
  569. const resourcePromises = [
  570. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI, this.batchRequest),
  571. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", this.baseURI, this.batchRequest),
  572. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], embed[src*=\".svg\"]"), "src", this.baseURI, this.batchRequest),
  573. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI, this.batchRequest),
  574. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI, this.batchRequest),
  575. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", this.baseURI, this.batchRequest),
  576. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  577. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.batchRequest)
  578. ];
  579. if (!this.options.removeAudioSrc) {
  580. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.batchRequest));
  581. }
  582. if (!this.options.removeVideoSrc) {
  583. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.batchRequest));
  584. }
  585. if (this.options.lazyLoadImages) {
  586. const imageSelectors = DOM.lazyLoaderImageSelectors();
  587. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], this.baseURI, this.batchRequest)));
  588. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], this.baseURI, this.batchRequest)));
  589. }
  590. await resourcePromises;
  591. }
  592. async inlineStylesheets(initialization) {
  593. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  594. if (!initialization) {
  595. this.stats.add("processed", "stylesheets", 1);
  596. }
  597. let stylesheetContent = styleElement.textContent;
  598. if (initialization) {
  599. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  600. } else {
  601. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  602. }
  603. styleElement.textContent = stylesheetContent;
  604. }));
  605. }
  606. async scripts() {
  607. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  608. if (scriptElement.src) {
  609. this.stats.add("processed", "scripts", 1);
  610. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  611. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  612. }
  613. scriptElement.removeAttribute("src");
  614. }));
  615. }
  616. async frames(initialization) {
  617. if (this.options.framesData) {
  618. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  619. await Promise.all(frameElements.map(async frameElement => {
  620. DomProcessorHelper.setFrameEmptySrc(frameElement);
  621. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  622. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  623. if (frameWindowId) {
  624. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  625. if (frameData) {
  626. if (initialization) {
  627. const options = Object.create(this.options);
  628. options.insertSingleFileComment = false;
  629. options.insertFaviconLink = false;
  630. options.doc = null;
  631. options.win = null;
  632. options.url = frameData.baseURI;
  633. options.windowId = frameWindowId;
  634. if (frameData.content) {
  635. options.content = frameData.content;
  636. options.canvasData = frameData.canvasData;
  637. options.stylesheetContents = frameData.stylesheetContents;
  638. options.currentSrcImages = frameData.currentSrcImages;
  639. frameData.processor = new PageProcessor(options);
  640. frameData.frameElement = frameElement;
  641. await frameData.processor.loadPage();
  642. await frameData.processor.initialize();
  643. frameData.maxResources = this.batchRequest.getMaxResources();
  644. }
  645. } else {
  646. if (frameData.processor) {
  647. this.stats.add("processed", "frames", 1);
  648. await frameData.processor.preparePageData();
  649. const pageData = await frameData.processor.getPageData();
  650. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  651. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  652. this.stats.addAll(pageData);
  653. } else {
  654. this.stats.add("discarded", "frames", 1);
  655. }
  656. }
  657. }
  658. }
  659. }));
  660. }
  661. }
  662. async htmlImports(initialization) {
  663. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  664. if (!this.relImportProcessors) {
  665. this.relImportProcessors = new Map();
  666. }
  667. await Promise.all(linkElements.map(async linkElement => {
  668. if (initialization) {
  669. const resourceURL = linkElement.href;
  670. const options = Object.create(this.options);
  671. options.insertSingleFileComment = false;
  672. options.insertFaviconLink = false;
  673. options.doc = null;
  674. options.win = null;
  675. options.url = resourceURL;
  676. if (resourceURL) {
  677. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  678. const processor = new PageProcessor(options);
  679. this.relImportProcessors.set(linkElement, processor);
  680. await processor.loadPage();
  681. return processor.initialize();
  682. }
  683. }
  684. } else {
  685. linkElement.setAttribute("href", EMPTY_DATA_URI);
  686. const processor = this.relImportProcessors.get(linkElement);
  687. if (processor) {
  688. this.stats.add("processed", "HTML imports", 1);
  689. this.relImportProcessors.delete(linkElement);
  690. const pageData = await processor.getPageData();
  691. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  692. this.stats.addAll(pageData);
  693. } else {
  694. this.stats.add("discarded", "HTML imports", 1);
  695. }
  696. }
  697. }));
  698. }
  699. async attributeStyles(initialization) {
  700. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  701. let stylesheetContent = element.getAttribute("style");
  702. if (initialization) {
  703. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  704. } else {
  705. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  706. }
  707. element.setAttribute("style", stylesheetContent);
  708. }));
  709. }
  710. async linkStylesheets() {
  711. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  712. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  713. const styleElement = this.doc.createElement("style");
  714. styleElement.textContent = stylesheetContent;
  715. linkElement.parentElement.replaceChild(styleElement, linkElement);
  716. }));
  717. }
  718. }
  719. // ---------
  720. // DomHelper
  721. // ---------
  722. const REGEXP_AMP = /&/g;
  723. const REGEXP_NBSP = /\u00a0/g;
  724. const REGEXP_START_TAG = /</g;
  725. const REGEXP_END_TAG = />/g;
  726. const REGEXP_URL_HASH = /(#.+?)$/;
  727. const PREFIX_DATA_URI_IMAGE = "data:image/";
  728. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  729. class DomProcessorHelper {
  730. static async getFilename(options, content) {
  731. let filename = options.filenameTemplate;
  732. const date = new Date();
  733. const url = new URL(options.url);
  734. filename = filename.replace(/{\s*title\s*}/g, options.title.replace(/\/+/g, "_"));
  735. filename = filename.replace(/{\s*iso-datetime\s*}/g, date.toISOString().replace(/\/+/g, "_"));
  736. filename = filename.replace(/{\s*iso-date\s*}/g, date.toISOString().split("T")[0].replace(/\/+/g, "_"));
  737. filename = filename.replace(/{\s*iso-time\s*}/g, date.toISOString().split("T")[1].split("Z")[0].replace(/\/+/g, "_"));
  738. filename = filename.replace(/{\s*locale-date\s*}/g, date.toLocaleDateString().replace(/\/+/g, "_"));
  739. filename = filename.replace(/{\s*locale-time\s*}/g, date.toLocaleTimeString().replace(/\/+/g, "_"));
  740. filename = filename.replace(/{\s*locale-day\s*}/g, String(date.getDate()).replace(/\/+/g, "_"));
  741. filename = filename.replace(/{\s*locale-month\s*}/g, String(date.getMonth() + 1).replace(/\/+/g, "_"));
  742. filename = filename.replace(/{\s*locale-year\s*}/g, String(date.getFullYear()).replace(/\/+/g, "_"));
  743. filename = filename.replace(/{\s*locale-datetime\s*}/g, date.toLocaleString().replace(/\/+/g, "_"));
  744. filename = filename.replace(/{\s*utc-datetime\s*}/g, date.toUTCString().replace(/\/+/g, "_"));
  745. filename = filename.replace(/{\s*utc-day\s*}/g, String(date.getUTCDate()).replace(/\/+/g, "_"));
  746. filename = filename.replace(/{\s*utc-month\s*}/g, String(date.getUTCMonth() + 1).replace(/\/+/g, "_"));
  747. filename = filename.replace(/{\s*utc-year\s*}/g, String(date.getUTCFullYear()).replace(/\/+/g, "_"));
  748. filename = filename.replace(/{\s*url-hash\s*}/g, url.hash.substring(1).replace(/\/+/g, "_"));
  749. filename = filename.replace(/{\s*url-host\s*}/g, url.host.replace(/\/+/g, "_").replace(/\/$/, ""));
  750. filename = filename.replace(/{\s*url-hostname\s*}/g, url.hostname.replace(/\/+/g, "_").replace(/\/$/, ""));
  751. filename = filename.replace(/{\s*url-href\s*}/g, url.href.replace(/\/+/g, "_"));
  752. filename = filename.replace(/{\s*url-password\s*}/g, url.password.replace(/\/+/g, "_"));
  753. filename = filename.replace(/{\s*url-pathname\s*}/g, url.pathname.replace(/\/+/g, "_").replace(/\/$/, "").replace(/^\//, ""));
  754. filename = filename.replace(/{\s*url-port\s*}/g, url.port.replace(/\/+/g, "_"));
  755. filename = filename.replace(/{\s*url-protocol\s*}/g, url.protocol.replace(/\/+/g, "_").replace(/\/$/, ""));
  756. filename = filename.replace(/{\s*url-search\s*}/g, url.search.substring(1).replace(/\/+/g, "_"));
  757. filename = filename.replace(/{\s*url-username\s*}/g, url.username.replace(/\/+/g, "_"));
  758. filename = filename.replace(/{\s*tab-id\s*}/g, String(options.tabId || "unknown"));
  759. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  760. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  761. if (!lastSegment) {
  762. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  763. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  764. }
  765. if (!lastSegment) {
  766. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  767. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  768. }
  769. if (!lastSegment) {
  770. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  771. }
  772. filename = filename.replace(/{\s*url-last-segment\s*}/g, lastSegment.replace(/\/+/g, "").replace(/\/$/, "").replace(/^\//, ""));
  773. if (filename.match(/{\s*digest-sha-256\s*}/g)) {
  774. filename = filename.replace(/{\s*digest-sha-256\s*}/g, await DOM.digest("SHA-256", content));
  775. }
  776. if (filename.match(/{\s*digest-sha-384\s*}/g)) {
  777. filename = filename.replace(/{\s*digest-sha-384\s*}/g, await DOM.digest("SHA-384", content));
  778. }
  779. if (filename.match(/{\s*digest-sha-512\s*}/g)) {
  780. filename = filename.replace(/{\s*digest-sha-512\s*}/g, await DOM.digest("SHA-512", content));
  781. }
  782. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  783. filename = filename.replace(/\.\.\/?/g, "").replace(/^\/+/, "");
  784. if (filename.length > 192) {
  785. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  786. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  787. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  788. }
  789. if (!filename) {
  790. filename = "Untitled page";
  791. }
  792. return filename;
  793. }
  794. static setFrameEmptySrc(frameElement) {
  795. if (frameElement.tagName == "OBJECT") {
  796. frameElement.setAttribute("data", "data:text/html,");
  797. } else {
  798. frameElement.setAttribute("srcdoc", "");
  799. frameElement.removeAttribute("src");
  800. }
  801. }
  802. static setFrameContent(frameElement, content) {
  803. if (frameElement.tagName == "OBJECT") {
  804. frameElement.setAttribute("data", "data:text/html," + content);
  805. } else {
  806. frameElement.setAttribute("srcdoc", content);
  807. frameElement.removeAttribute("src");
  808. }
  809. }
  810. static isolateElements(rootElement) {
  811. rootElement.querySelectorAll("*").forEach(element => {
  812. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  813. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  814. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  815. element.remove();
  816. }
  817. });
  818. isolateParentElements(rootElement.parentElement, rootElement);
  819. function isolateParentElements(parentElement, element) {
  820. if (parentElement) {
  821. Array.from(parentElement.childNodes).forEach(node => {
  822. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  823. node.remove();
  824. }
  825. });
  826. }
  827. element = element.parentElement;
  828. if (element && element.parentElement) {
  829. isolateParentElements(element.parentElement, element);
  830. }
  831. }
  832. }
  833. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  834. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  835. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  836. const imports = DomUtil.getImportFunctions(stylesheetContent);
  837. await Promise.all(imports.map(async cssImport => {
  838. const match = DomUtil.matchImport(cssImport);
  839. if (match) {
  840. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  841. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  842. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  843. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  844. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  845. if (stylesheetContent.includes(cssImport)) {
  846. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  847. if (cssImport.endsWith(";") && !stylesheetContent.endsWith(";")) {
  848. stylesheetContent += ";";
  849. }
  850. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  851. }
  852. }
  853. }
  854. }));
  855. return stylesheetContent;
  856. }
  857. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  858. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  859. urlFunctions.map(urlFunction => {
  860. const originalResourceURL = DomUtil.matchURL(urlFunction);
  861. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  862. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  863. const resolvedURL = new URL(resourceURL, baseURI).href;
  864. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  865. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  866. }
  867. } else {
  868. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  869. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  870. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  871. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  872. }
  873. }
  874. }
  875. });
  876. return stylesheetContent;
  877. }
  878. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  879. resourceURL = DomUtil.normalizeURL(resourceURL);
  880. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  881. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  882. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  883. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  884. return stylesheetContent;
  885. }
  886. }
  887. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  888. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  889. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  890. let indexVariable = 0;
  891. await Promise.all(urlFunctions.map(async urlFunction => {
  892. const originalResourceURL = DomUtil.matchURL(urlFunction);
  893. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  894. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  895. const dataURI = await batchRequest.addURL(resourceURL);
  896. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  897. const functions = stylesheetContent.match(DomUtil.getRegExp(urlFunction));
  898. if (functions.length > 1) {
  899. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  900. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  901. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), "var(" + variableName + ")");
  902. indexVariable++;
  903. } else {
  904. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  905. }
  906. } else {
  907. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  908. }
  909. }
  910. }));
  911. if (indexVariable) {
  912. stylesheetContent = ":root{" + stylesheetContent;
  913. }
  914. return stylesheetContent;
  915. }
  916. static async processAttribute(resourceElements, attributeName, baseURI, batchRequest) {
  917. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  918. let resourceURL = resourceElement.getAttribute(attributeName);
  919. if (resourceURL) {
  920. resourceURL = DomUtil.normalizeURL(resourceURL);
  921. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  922. try {
  923. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  924. resourceElement.setAttribute(attributeName, dataURI);
  925. } catch (error) {
  926. /* ignored */
  927. }
  928. }
  929. }
  930. }));
  931. }
  932. static async processXLinks(resourceElements, baseURI, batchRequest) {
  933. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  934. const originalResourceURL = resourceElement.getAttribute("xlink:href");
  935. if (originalResourceURL) {
  936. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  937. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  938. try {
  939. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  940. const DOMParser = DOM.getParser();
  941. if (DOMParser) {
  942. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  943. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  944. if (hashMatch && hashMatch[0]) {
  945. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  946. if (symbolElement) {
  947. resourceElement.setAttribute("xlink:href", hashMatch[0]);
  948. resourceElement.parentElement.appendChild(symbolElement);
  949. }
  950. } else {
  951. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  952. }
  953. } else {
  954. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  955. }
  956. } catch (error) {
  957. /* ignored */
  958. }
  959. }
  960. }
  961. }));
  962. }
  963. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  964. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  965. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  966. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  967. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  968. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  969. try {
  970. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  971. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  972. } catch (error) {
  973. /* ignored */
  974. }
  975. }
  976. }));
  977. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  978. }));
  979. }
  980. }
  981. // -------
  982. // DomUtil
  983. // -------
  984. const DATA_URI_PREFIX = "data:";
  985. const BLOB_URI_PREFIX = "blob:";
  986. const HTTP_URI_PREFIX = /^https?:\/\//;
  987. const ABOUT_BLANK_URI = "about:blank";
  988. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  989. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  990. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  991. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  992. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  993. 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;
  994. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  995. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  996. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  997. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  998. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  999. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1000. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1001. class DomUtil {
  1002. static normalizeURL(url) {
  1003. if (url.startsWith(DATA_URI_PREFIX)) {
  1004. return url;
  1005. } else {
  1006. return url.split("#")[0];
  1007. }
  1008. }
  1009. static getRegExp(string) {
  1010. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1011. }
  1012. static getUrlFunctions(stylesheetContent) {
  1013. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1014. }
  1015. static getImportFunctions(stylesheetContent) {
  1016. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1017. }
  1018. static matchURL(stylesheetContent) {
  1019. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1020. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1021. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1022. return match && match[1];
  1023. }
  1024. static testValidPath(resourceURL) {
  1025. 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));
  1026. }
  1027. static matchImport(stylesheetContent) {
  1028. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1029. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1030. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1031. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1032. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1033. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1034. if (match) {
  1035. const [, resourceURL, media] = match;
  1036. return { resourceURL, media };
  1037. }
  1038. }
  1039. static removeCssComments(stylesheetContent) {
  1040. let start, end;
  1041. do {
  1042. start = stylesheetContent.indexOf("/*");
  1043. end = stylesheetContent.indexOf("*/", start + 2);
  1044. if (start != -1 && end != -1) {
  1045. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1046. }
  1047. } while (start != -1 && end != -1);
  1048. return stylesheetContent;
  1049. }
  1050. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1051. if (mediaQuery) {
  1052. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1053. } else {
  1054. return stylesheetContent;
  1055. }
  1056. }
  1057. }
  1058. // -----
  1059. // Stats
  1060. // -----
  1061. const STATS_DEFAULT_VALUES = {
  1062. discarded: {
  1063. "HTML bytes": 0,
  1064. "hidden elements": 0,
  1065. "HTML imports": 0,
  1066. scripts: 0,
  1067. objects: 0,
  1068. "audio sources": 0,
  1069. "video sources": 0,
  1070. frames: 0,
  1071. "CSS rules": 0,
  1072. canvas: 0,
  1073. stylesheets: 0,
  1074. resources: 0,
  1075. medias: 0
  1076. },
  1077. processed: {
  1078. "HTML bytes": 0,
  1079. "hidden elements": 0,
  1080. "HTML imports": 0,
  1081. scripts: 0,
  1082. objects: 0,
  1083. "audio sources": 0,
  1084. "video sources": 0,
  1085. frames: 0,
  1086. "CSS rules": 0,
  1087. canvas: 0,
  1088. stylesheets: 0,
  1089. resources: 0,
  1090. medias: 0
  1091. }
  1092. };
  1093. class Stats {
  1094. constructor(options) {
  1095. this.options = options;
  1096. if (options.displayStats) {
  1097. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1098. }
  1099. }
  1100. set(type, subType, value) {
  1101. if (this.options.displayStats) {
  1102. this.data[type][subType] = value;
  1103. }
  1104. }
  1105. add(type, subType, value) {
  1106. if (this.options.displayStats) {
  1107. this.data[type][subType] += value;
  1108. }
  1109. }
  1110. addAll(pageData) {
  1111. if (this.options.displayStats) {
  1112. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1113. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1114. }
  1115. }
  1116. }
  1117. return { getClass };
  1118. })();