single-file-core.js 51 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228
  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. // console.log(task.action, "...");
  172. this.executeTask(task, !step);
  173. // console.log(task.action, Date.now() - startTime);
  174. });
  175. if (STAGES[step].async) {
  176. return Promise.all(STAGES[step].async.map(task => {
  177. // const startTime = Date.now();
  178. // console.log(task.action, "...");
  179. const promise = this.executeTask(task, !step);
  180. // promise.then(() => console.log(task.action, Date.now() - startTime));
  181. return promise;
  182. }));
  183. }
  184. }
  185. executeTask(task, initialization) {
  186. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  187. return this.processor[task.action](initialization);
  188. }
  189. }
  190. }
  191. // --------
  192. // BatchRequest
  193. // --------
  194. class BatchRequest {
  195. constructor() {
  196. this.requests = new Map();
  197. }
  198. async addURL(resourceURL, asDataURI = true) {
  199. return new Promise((resolve, reject) => {
  200. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  201. const resourceRequests = this.requests.get(requestKey);
  202. if (resourceRequests) {
  203. resourceRequests.push({ resolve, reject });
  204. } else {
  205. this.requests.set(requestKey, [{ resolve, reject }]);
  206. }
  207. });
  208. }
  209. getMaxResources() {
  210. return Array.from(this.requests.keys()).length;
  211. }
  212. async run(onloadListener, options) {
  213. const resourceURLs = Array.from(this.requests.keys());
  214. let indexResource = 0;
  215. return Promise.all(resourceURLs.map(async requestKey => {
  216. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  217. const resourceRequests = this.requests.get(requestKey);
  218. try {
  219. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  220. indexResource = indexResource + 1;
  221. onloadListener({ index: indexResource, url: resourceURL });
  222. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  223. } catch (error) {
  224. indexResource = indexResource + 1;
  225. onloadListener({ index: indexResource, url: resourceURL });
  226. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  227. }
  228. this.requests.delete(requestKey);
  229. }));
  230. }
  231. }
  232. // ------------
  233. // DOMProcessor
  234. // ------------
  235. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  236. const EMPTY_DATA_URI = "data:base64,";
  237. class DOMProcessor {
  238. constructor(options, batchRequest) {
  239. this.options = options;
  240. this.stats = new Stats(options);
  241. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  242. this.batchRequest = batchRequest;
  243. }
  244. initialize() {
  245. this.maxResources = this.batchRequest.getMaxResources();
  246. if (!this.options.removeFrames) {
  247. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  248. }
  249. this.stats.set("processed", "resources", this.maxResources);
  250. }
  251. async loadPage(pageContent) {
  252. if (!pageContent || this.options.saveRawPage) {
  253. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  254. }
  255. this.doc = DOM.createDoc(pageContent, this.baseURI);
  256. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  257. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  258. await DOMProcessor.loadEscapedFragmentPage();
  259. }
  260. }
  261. async loadEscapedFragmentPage() {
  262. if (this.baseURI.includes("?")) {
  263. this.baseURI += "&";
  264. } else {
  265. this.baseURI += "?";
  266. }
  267. this.baseURI += ESCAPED_FRAGMENT;
  268. await this.loadPage();
  269. }
  270. async getPageData() {
  271. DOM.postProcessDoc(this.doc, this.options);
  272. if (this.options.selected) {
  273. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  274. if (rootElement) {
  275. DomProcessorHelper.isolateElements(rootElement);
  276. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  277. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  278. }
  279. }
  280. const titleElement = this.doc.querySelector("title");
  281. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  282. this.options.info = {};
  283. const descriptionElement = this.doc.querySelector("meta[name=description]");
  284. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  285. this.options.info.lang = this.doc.documentElement.lang;
  286. const authorElement = this.doc.querySelector("meta[name=author]");
  287. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  288. const creatorElement = this.doc.querySelector("meta[name=creator]");
  289. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  290. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  291. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  292. const url = new URL(this.baseURI);
  293. let size;
  294. if (this.options.displayStats) {
  295. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  296. }
  297. const content = DOM.serialize(this.doc, this.options.compressHTML);
  298. if (this.options.displayStats) {
  299. const contentSize = DOM.getContentSize(content);
  300. this.stats.set("processed", "HTML bytes", contentSize);
  301. this.stats.add("discarded", "HTML bytes", size - contentSize);
  302. }
  303. const filename = await DomProcessorHelper.getFilename(this.options, content);
  304. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  305. return {
  306. stats: this.stats.data,
  307. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  308. filename,
  309. content
  310. };
  311. }
  312. setInputValues() {
  313. this.doc.querySelectorAll("input").forEach(input => {
  314. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  315. if (value) {
  316. input.setAttribute("value", value);
  317. }
  318. });
  319. this.doc.querySelectorAll("textarea").forEach(textarea => {
  320. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  321. if (value) {
  322. textarea.textContent = value;
  323. }
  324. });
  325. this.doc.querySelectorAll("select").forEach(select => {
  326. select.querySelectorAll("option").forEach(option => {
  327. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  328. if (selected) {
  329. option.setAttribute("selected", "");
  330. }
  331. });
  332. });
  333. }
  334. lazyLoadImages() {
  335. DOM.lazyLoad(this.doc);
  336. }
  337. removeDiscardedResources() {
  338. 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\"])");
  339. this.stats.set("discarded", "objects", objectElements.length);
  340. this.stats.set("processed", "objects", objectElements.length);
  341. objectElements.forEach(element => element.remove());
  342. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  343. replacedAttributeValue.forEach(element => {
  344. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  345. if (relValue.length) {
  346. element.setAttribute("rel", relValue);
  347. } else {
  348. element.remove();
  349. }
  350. });
  351. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"], input[type=hidden]").forEach(element => element.remove());
  352. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  353. if (this.options.removeScripts) {
  354. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  355. this.doc.querySelectorAll("[href]").forEach(element => {
  356. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  357. element.removeAttribute("href");
  358. }
  359. });
  360. this.doc.querySelectorAll("[src]").forEach(element => {
  361. if (element.src && element.src.match(/^\s*javascript:/)) {
  362. element.removeAttribute("src");
  363. }
  364. });
  365. }
  366. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  367. this.stats.set("processed", "audio sources", audioSourceElements.length);
  368. if (this.options.removeAudioSrc) {
  369. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  370. audioSourceElements.forEach(element => element.removeAttribute("src"));
  371. }
  372. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  373. this.stats.set("processed", "video sources", videoSourceElements.length);
  374. if (this.options.removeVideoSrc) {
  375. this.stats.set("discarded", "video sources", videoSourceElements.length);
  376. videoSourceElements.forEach(element => element.removeAttribute("src"));
  377. }
  378. }
  379. removeDefaultHeadTags() {
  380. this.doc.querySelectorAll("base").forEach(element => element.remove());
  381. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  382. this.doc.head.querySelector("meta[charset]").remove();
  383. }
  384. }
  385. removeUIElements() {
  386. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  387. }
  388. removeScripts() {
  389. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  390. this.stats.set("discarded", "scripts", scriptElements.length);
  391. this.stats.set("processed", "scripts", scriptElements.length);
  392. scriptElements.forEach(element => element.remove());
  393. }
  394. removeFrames() {
  395. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  396. this.stats.set("discarded", "frames", frameElements.length);
  397. this.stats.set("processed", "frames", frameElements.length);
  398. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  399. }
  400. removeImports() {
  401. const importElements = this.doc.querySelectorAll("link[rel=import]");
  402. this.stats.set("discarded", "HTML imports", importElements.length);
  403. this.stats.set("processed", "HTML imports", importElements.length);
  404. importElements.forEach(element => element.remove());
  405. }
  406. resetCharsetMeta() {
  407. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  408. const metaElement = this.doc.createElement("meta");
  409. metaElement.setAttribute("charset", "utf-8");
  410. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  411. }
  412. insertFaviconLink() {
  413. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  414. if (!faviconElement) {
  415. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  416. }
  417. if (!faviconElement) {
  418. faviconElement = this.doc.createElement("link");
  419. faviconElement.setAttribute("type", "image/x-icon");
  420. faviconElement.setAttribute("rel", "shortcut icon");
  421. faviconElement.setAttribute("href", "/favicon.ico");
  422. }
  423. this.doc.head.appendChild(faviconElement);
  424. }
  425. resolveHrefs() {
  426. this.doc.querySelectorAll("[href]").forEach(element => {
  427. if (element.href) {
  428. const href = element.href.baseVal ? element.href.baseVal : element.href;
  429. const normalizedHref = DomUtil.normalizeURL(href);
  430. if (!normalizedHref || normalizedHref != this.baseURI) {
  431. element.setAttribute("href", href);
  432. }
  433. }
  434. });
  435. }
  436. removeUnusedStyles() {
  437. const stats = DOM.minifyCSS(this.doc);
  438. this.stats.set("processed", "CSS rules", stats.processed);
  439. this.stats.set("discarded", "CSS rules", stats.discarded);
  440. }
  441. removeAlternativeFonts() {
  442. DOM.minifyFonts(this.doc);
  443. }
  444. removeSrcSet() {
  445. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  446. const tagName = element.tagName.toLowerCase();
  447. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  448. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  449. element.removeAttribute(dataAttributeName);
  450. if (imageData) {
  451. if (tagName == "img") {
  452. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  453. element.removeAttribute("srcset");
  454. element.removeAttribute("sizes");
  455. element.src = imageData.source.src;
  456. }
  457. }
  458. if (tagName == "picture") {
  459. const imageElement = element.querySelector("img");
  460. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  461. imageElement.removeAttribute("srcset");
  462. imageElement.removeAttribute("sizes");
  463. imageElement.src = imageData.source.src;
  464. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  465. } else {
  466. if (imageData.sources) {
  467. element.querySelectorAll("source").forEach(sourceElement => {
  468. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  469. sourceElement.remove();
  470. }
  471. });
  472. const sourceElements = element.querySelectorAll("source");
  473. if (sourceElements.length) {
  474. const lastSourceElement = sourceElements[sourceElements.length - 1];
  475. if (imageElement) {
  476. if (lastSourceElement.src) {
  477. imageElement.src = lastSourceElement.src;
  478. } else {
  479. imageElement.removeAttribute("src");
  480. }
  481. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  482. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  483. } else {
  484. imageElement.removeAttribute("srcset");
  485. }
  486. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  487. }
  488. }
  489. }
  490. }
  491. }
  492. }
  493. });
  494. }
  495. postRemoveAlternativeFonts() {
  496. DOM.minifyFonts(this.doc, true);
  497. }
  498. removeHiddenElements() {
  499. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  500. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  501. this.stats.set("processed", "hidden elements", hiddenElements.length);
  502. hiddenElements.forEach(element => element.remove());
  503. }
  504. compressHTML() {
  505. let size;
  506. if (this.options.displayStats) {
  507. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  508. }
  509. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  510. if (this.options.displayStats) {
  511. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  512. }
  513. }
  514. postCompressHTML() {
  515. let size;
  516. if (this.options.displayStats) {
  517. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  518. }
  519. DOM.postMinifyHTML(this.doc);
  520. if (this.options.displayStats) {
  521. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  522. }
  523. }
  524. removeAlternativeMedias() {
  525. const stats = DOM.minifyMedias(this.doc);
  526. this.stats.set("processed", "medias", stats.processed);
  527. this.stats.set("discarded", "medias", stats.discarded);
  528. }
  529. compressCSS() {
  530. this.doc.querySelectorAll("style").forEach(styleElement => {
  531. if (styleElement) {
  532. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  533. }
  534. });
  535. this.doc.querySelectorAll("[style]").forEach(element => {
  536. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  537. });
  538. }
  539. insertSingleFileComment() {
  540. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  541. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  542. }
  543. replaceCanvasElements() {
  544. if (this.options.canvasData) {
  545. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  546. const canvasData = this.options.canvasData[indexCanvasElement];
  547. if (canvasData) {
  548. const imgElement = this.doc.createElement("img");
  549. imgElement.setAttribute("src", canvasData.dataURI);
  550. Array.from(canvasElement.attributes).forEach(attribute => {
  551. if (attribute.value) {
  552. imgElement.setAttribute(attribute.name, attribute.value);
  553. }
  554. });
  555. if (!imgElement.width && canvasData.width) {
  556. imgElement.style.pixelWidth = canvasData.width;
  557. }
  558. if (!imgElement.height && canvasData.height) {
  559. imgElement.style.pixelHeight = canvasData.height;
  560. }
  561. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  562. this.stats.add("processed", "canvas", 1);
  563. }
  564. });
  565. }
  566. }
  567. replaceStyleContents() {
  568. if (this.options.stylesheetContents) {
  569. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  570. if (this.options.stylesheetContents[styleIndex]) {
  571. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  572. }
  573. });
  574. }
  575. }
  576. async pageResources() {
  577. const resourcePromises = [
  578. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI, this.batchRequest),
  579. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", this.baseURI, this.batchRequest),
  580. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], embed[src*=\".svg\"]"), "src", this.baseURI, this.batchRequest),
  581. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI, this.batchRequest),
  582. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI, this.batchRequest),
  583. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", this.baseURI, this.batchRequest),
  584. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  585. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.batchRequest)
  586. ];
  587. if (!this.options.removeAudioSrc) {
  588. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.batchRequest));
  589. }
  590. if (!this.options.removeVideoSrc) {
  591. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.batchRequest));
  592. }
  593. if (this.options.lazyLoadImages) {
  594. const imageSelectors = DOM.lazyLoaderImageSelectors();
  595. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], this.baseURI, this.batchRequest)));
  596. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], this.baseURI, this.batchRequest)));
  597. }
  598. await resourcePromises;
  599. }
  600. async inlineStylesheets(initialization) {
  601. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  602. if (!initialization) {
  603. this.stats.add("processed", "stylesheets", 1);
  604. }
  605. let stylesheetContent = styleElement.textContent;
  606. if (initialization) {
  607. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  608. } else {
  609. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  610. }
  611. styleElement.textContent = stylesheetContent;
  612. }));
  613. }
  614. async scripts() {
  615. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  616. if (scriptElement.src) {
  617. this.stats.add("processed", "scripts", 1);
  618. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  619. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  620. }
  621. scriptElement.removeAttribute("src");
  622. }));
  623. }
  624. async frames(initialization) {
  625. if (this.options.framesData) {
  626. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  627. await Promise.all(frameElements.map(async frameElement => {
  628. DomProcessorHelper.setFrameEmptySrc(frameElement);
  629. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  630. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  631. if (frameWindowId) {
  632. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  633. if (frameData) {
  634. if (initialization) {
  635. const options = Object.create(this.options);
  636. options.insertSingleFileComment = false;
  637. options.insertFaviconLink = false;
  638. options.doc = null;
  639. options.win = null;
  640. options.url = frameData.baseURI;
  641. options.windowId = frameWindowId;
  642. if (frameData.content) {
  643. options.content = frameData.content;
  644. options.canvasData = frameData.canvasData;
  645. options.stylesheetContents = frameData.stylesheetContents;
  646. options.currentSrcImages = frameData.currentSrcImages;
  647. frameData.processor = new PageProcessor(options);
  648. frameData.frameElement = frameElement;
  649. await frameData.processor.loadPage();
  650. await frameData.processor.initialize();
  651. frameData.maxResources = this.batchRequest.getMaxResources();
  652. }
  653. } else {
  654. if (frameData.processor) {
  655. this.stats.add("processed", "frames", 1);
  656. await frameData.processor.preparePageData();
  657. const pageData = await frameData.processor.getPageData();
  658. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  659. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  660. this.stats.addAll(pageData);
  661. } else {
  662. this.stats.add("discarded", "frames", 1);
  663. }
  664. }
  665. }
  666. }
  667. }));
  668. }
  669. }
  670. async htmlImports(initialization) {
  671. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  672. if (!this.relImportProcessors) {
  673. this.relImportProcessors = new Map();
  674. }
  675. await Promise.all(linkElements.map(async linkElement => {
  676. if (initialization) {
  677. const resourceURL = linkElement.href;
  678. const options = Object.create(this.options);
  679. options.insertSingleFileComment = false;
  680. options.insertFaviconLink = false;
  681. options.doc = null;
  682. options.win = null;
  683. options.url = resourceURL;
  684. if (resourceURL) {
  685. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  686. const processor = new PageProcessor(options);
  687. this.relImportProcessors.set(linkElement, processor);
  688. await processor.loadPage();
  689. return processor.initialize();
  690. }
  691. }
  692. } else {
  693. linkElement.setAttribute("href", EMPTY_DATA_URI);
  694. const processor = this.relImportProcessors.get(linkElement);
  695. if (processor) {
  696. this.stats.add("processed", "HTML imports", 1);
  697. this.relImportProcessors.delete(linkElement);
  698. const pageData = await processor.getPageData();
  699. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  700. this.stats.addAll(pageData);
  701. } else {
  702. this.stats.add("discarded", "HTML imports", 1);
  703. }
  704. }
  705. }));
  706. }
  707. async attributeStyles(initialization) {
  708. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  709. let stylesheetContent = element.getAttribute("style");
  710. if (initialization) {
  711. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  712. } else {
  713. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  714. }
  715. element.setAttribute("style", stylesheetContent);
  716. }));
  717. }
  718. async linkStylesheets() {
  719. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  720. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  721. const styleElement = this.doc.createElement("style");
  722. styleElement.textContent = stylesheetContent;
  723. linkElement.parentElement.replaceChild(styleElement, linkElement);
  724. }));
  725. }
  726. }
  727. // ---------
  728. // DomHelper
  729. // ---------
  730. const REGEXP_AMP = /&/g;
  731. const REGEXP_NBSP = /\u00a0/g;
  732. const REGEXP_START_TAG = /</g;
  733. const REGEXP_END_TAG = />/g;
  734. const REGEXP_URL_HASH = /(#.+?)$/;
  735. const PREFIX_DATA_URI_IMAGE = "data:image/";
  736. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  737. class DomProcessorHelper {
  738. static async getFilename(options, content) {
  739. let filename = options.filenameTemplate;
  740. const date = new Date();
  741. const url = new URL(options.url);
  742. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  743. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  744. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  745. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  746. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  747. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  748. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  749. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  750. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  751. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  752. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  753. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  754. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  755. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  756. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  757. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  758. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  759. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  760. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  761. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  762. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  763. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  764. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  765. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  766. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  767. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  768. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  769. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  770. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  771. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  772. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  773. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  774. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  775. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  776. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  777. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  778. if (!options.backgroundSave) {
  779. filename = filename.replace(/\//g, "_");
  780. }
  781. if (filename.length > 192) {
  782. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  783. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  784. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  785. }
  786. if (!filename) {
  787. filename = "Unnamed page";
  788. }
  789. return filename;
  790. }
  791. static setFrameEmptySrc(frameElement) {
  792. if (frameElement.tagName == "OBJECT") {
  793. frameElement.setAttribute("data", "data:text/html,");
  794. } else {
  795. frameElement.setAttribute("srcdoc", "");
  796. frameElement.removeAttribute("src");
  797. }
  798. }
  799. static setFrameContent(frameElement, content) {
  800. if (frameElement.tagName == "OBJECT") {
  801. frameElement.setAttribute("data", "data:text/html," + content);
  802. } else {
  803. frameElement.setAttribute("srcdoc", content);
  804. frameElement.removeAttribute("src");
  805. }
  806. }
  807. static isolateElements(rootElement) {
  808. rootElement.querySelectorAll("*").forEach(element => {
  809. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  810. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  811. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  812. element.remove();
  813. }
  814. });
  815. isolateParentElements(rootElement.parentElement, rootElement);
  816. function isolateParentElements(parentElement, element) {
  817. if (parentElement) {
  818. Array.from(parentElement.childNodes).forEach(node => {
  819. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  820. node.remove();
  821. }
  822. });
  823. }
  824. element = element.parentElement;
  825. if (element && element.parentElement) {
  826. isolateParentElements(element.parentElement, element);
  827. }
  828. }
  829. }
  830. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  831. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  832. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  833. const imports = DomUtil.getImportFunctions(stylesheetContent);
  834. await Promise.all(imports.map(async cssImport => {
  835. const match = DomUtil.matchImport(cssImport);
  836. if (match) {
  837. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  838. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  839. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  840. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  841. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  842. if (stylesheetContent.includes(cssImport)) {
  843. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  844. if (cssImport.endsWith(";") && !stylesheetContent.endsWith(";")) {
  845. stylesheetContent += ";";
  846. }
  847. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  848. }
  849. }
  850. }
  851. }));
  852. return stylesheetContent;
  853. }
  854. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  855. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  856. urlFunctions.map(urlFunction => {
  857. const originalResourceURL = DomUtil.matchURL(urlFunction);
  858. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  859. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  860. const resolvedURL = new URL(resourceURL, baseURI).href;
  861. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  862. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  863. }
  864. } else {
  865. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  866. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  867. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  868. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  869. }
  870. }
  871. }
  872. });
  873. return stylesheetContent;
  874. }
  875. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  876. resourceURL = DomUtil.normalizeURL(resourceURL);
  877. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  878. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  879. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  880. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  881. return stylesheetContent;
  882. }
  883. }
  884. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  885. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  886. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  887. let indexVariable = 0;
  888. await Promise.all(urlFunctions.map(async urlFunction => {
  889. const originalResourceURL = DomUtil.matchURL(urlFunction);
  890. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  891. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  892. const dataURI = await batchRequest.addURL(resourceURL);
  893. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  894. const functions = stylesheetContent.match(DomUtil.getRegExp(urlFunction));
  895. if (functions.length > 1) {
  896. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  897. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  898. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), "var(" + variableName + ")");
  899. indexVariable++;
  900. } else {
  901. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  902. }
  903. } else {
  904. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, dataURI));
  905. }
  906. }
  907. }));
  908. if (indexVariable) {
  909. stylesheetContent = ":root{" + stylesheetContent;
  910. }
  911. return stylesheetContent;
  912. }
  913. static async processAttribute(resourceElements, attributeName, baseURI, batchRequest) {
  914. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  915. let resourceURL = resourceElement.getAttribute(attributeName);
  916. if (resourceURL) {
  917. resourceURL = DomUtil.normalizeURL(resourceURL);
  918. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  919. try {
  920. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  921. resourceElement.setAttribute(attributeName, dataURI);
  922. } catch (error) {
  923. /* ignored */
  924. }
  925. }
  926. }
  927. }));
  928. }
  929. static async processXLinks(resourceElements, baseURI, batchRequest) {
  930. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  931. const originalResourceURL = resourceElement.getAttribute("xlink:href");
  932. if (originalResourceURL) {
  933. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  934. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  935. try {
  936. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  937. const DOMParser = DOM.getParser();
  938. if (DOMParser) {
  939. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  940. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  941. if (hashMatch && hashMatch[0]) {
  942. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  943. if (symbolElement) {
  944. resourceElement.setAttribute("xlink:href", hashMatch[0]);
  945. resourceElement.parentElement.appendChild(symbolElement);
  946. }
  947. } else {
  948. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  949. }
  950. } else {
  951. resourceElement.setAttribute("xlink:href", "data:image/svg+xml," + content);
  952. }
  953. } catch (error) {
  954. /* ignored */
  955. }
  956. }
  957. }
  958. }));
  959. }
  960. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  961. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  962. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  963. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  964. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  965. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  966. try {
  967. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  968. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  969. } catch (error) {
  970. /* ignored */
  971. }
  972. }
  973. }));
  974. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  975. }));
  976. }
  977. }
  978. // -------
  979. // DomUtil
  980. // -------
  981. const DATA_URI_PREFIX = "data:";
  982. const BLOB_URI_PREFIX = "blob:";
  983. const HTTP_URI_PREFIX = /^https?:\/\//;
  984. const ABOUT_BLANK_URI = "about:blank";
  985. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  986. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  987. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  988. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  989. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  990. 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;
  991. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  992. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  993. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  994. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  995. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  996. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  997. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  998. class DomUtil {
  999. static normalizeURL(url) {
  1000. if (url.startsWith(DATA_URI_PREFIX)) {
  1001. return url;
  1002. } else {
  1003. return url.split("#")[0];
  1004. }
  1005. }
  1006. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1007. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1008. if (template.match(replaceRegExp)) {
  1009. let value = await valueGetter();
  1010. if (!dontReplaceSlash) {
  1011. value = value.replace(/\/+/g, "_");
  1012. }
  1013. return template.replace(replaceRegExp, value);
  1014. }
  1015. return template;
  1016. }
  1017. static getLastSegment(url) {
  1018. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1019. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1020. if (!lastSegment) {
  1021. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1022. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1023. }
  1024. if (!lastSegment) {
  1025. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1026. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1027. }
  1028. if (!lastSegment) {
  1029. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1030. }
  1031. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1032. return lastSegment;
  1033. }
  1034. static getRegExp(string) {
  1035. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1036. }
  1037. static getUrlFunctions(stylesheetContent) {
  1038. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1039. }
  1040. static getImportFunctions(stylesheetContent) {
  1041. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1042. }
  1043. static matchURL(stylesheetContent) {
  1044. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1045. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1046. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1047. return match && match[1];
  1048. }
  1049. static testValidPath(resourceURL) {
  1050. 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));
  1051. }
  1052. static matchImport(stylesheetContent) {
  1053. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1054. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1055. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1056. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1057. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1058. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1059. if (match) {
  1060. const [, resourceURL, media] = match;
  1061. return { resourceURL, media };
  1062. }
  1063. }
  1064. static removeCssComments(stylesheetContent) {
  1065. let start, end;
  1066. do {
  1067. start = stylesheetContent.indexOf("/*");
  1068. end = stylesheetContent.indexOf("*/", start + 2);
  1069. if (start != -1 && end != -1) {
  1070. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1071. }
  1072. } while (start != -1 && end != -1);
  1073. return stylesheetContent;
  1074. }
  1075. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1076. if (mediaQuery) {
  1077. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1078. } else {
  1079. return stylesheetContent;
  1080. }
  1081. }
  1082. }
  1083. // -----
  1084. // Stats
  1085. // -----
  1086. const STATS_DEFAULT_VALUES = {
  1087. discarded: {
  1088. "HTML bytes": 0,
  1089. "hidden elements": 0,
  1090. "HTML imports": 0,
  1091. scripts: 0,
  1092. objects: 0,
  1093. "audio sources": 0,
  1094. "video sources": 0,
  1095. frames: 0,
  1096. "CSS rules": 0,
  1097. canvas: 0,
  1098. stylesheets: 0,
  1099. resources: 0,
  1100. medias: 0
  1101. },
  1102. processed: {
  1103. "HTML bytes": 0,
  1104. "hidden elements": 0,
  1105. "HTML imports": 0,
  1106. scripts: 0,
  1107. objects: 0,
  1108. "audio sources": 0,
  1109. "video sources": 0,
  1110. frames: 0,
  1111. "CSS rules": 0,
  1112. canvas: 0,
  1113. stylesheets: 0,
  1114. resources: 0,
  1115. medias: 0
  1116. }
  1117. };
  1118. class Stats {
  1119. constructor(options) {
  1120. this.options = options;
  1121. if (options.displayStats) {
  1122. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1123. }
  1124. }
  1125. set(type, subType, value) {
  1126. if (this.options.displayStats) {
  1127. this.data[type][subType] = value;
  1128. }
  1129. }
  1130. add(type, subType, value) {
  1131. if (this.options.displayStats) {
  1132. this.data[type][subType] += value;
  1133. }
  1134. }
  1135. addAll(pageData) {
  1136. if (this.options.displayStats) {
  1137. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1138. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1139. }
  1140. }
  1141. }
  1142. return { getClass };
  1143. })();