single-file-core.js 51 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231
  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\"]").forEach(element => element.remove());
  352. if (this.options.compressHTML) {
  353. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  354. }
  355. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  356. if (this.options.removeScripts) {
  357. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  358. this.doc.querySelectorAll("[href]").forEach(element => {
  359. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  360. element.removeAttribute("href");
  361. }
  362. });
  363. this.doc.querySelectorAll("[src]").forEach(element => {
  364. if (element.src && element.src.match(/^\s*javascript:/)) {
  365. element.removeAttribute("src");
  366. }
  367. });
  368. }
  369. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  370. this.stats.set("processed", "audio sources", audioSourceElements.length);
  371. if (this.options.removeAudioSrc) {
  372. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  373. audioSourceElements.forEach(element => element.removeAttribute("src"));
  374. }
  375. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  376. this.stats.set("processed", "video sources", videoSourceElements.length);
  377. if (this.options.removeVideoSrc) {
  378. this.stats.set("discarded", "video sources", videoSourceElements.length);
  379. videoSourceElements.forEach(element => element.removeAttribute("src"));
  380. }
  381. }
  382. removeDefaultHeadTags() {
  383. this.doc.querySelectorAll("base").forEach(element => element.remove());
  384. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  385. this.doc.head.querySelector("meta[charset]").remove();
  386. }
  387. }
  388. removeUIElements() {
  389. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  390. }
  391. removeScripts() {
  392. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  393. this.stats.set("discarded", "scripts", scriptElements.length);
  394. this.stats.set("processed", "scripts", scriptElements.length);
  395. scriptElements.forEach(element => element.remove());
  396. }
  397. removeFrames() {
  398. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  399. this.stats.set("discarded", "frames", frameElements.length);
  400. this.stats.set("processed", "frames", frameElements.length);
  401. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  402. }
  403. removeImports() {
  404. const importElements = this.doc.querySelectorAll("link[rel=import]");
  405. this.stats.set("discarded", "HTML imports", importElements.length);
  406. this.stats.set("processed", "HTML imports", importElements.length);
  407. importElements.forEach(element => element.remove());
  408. }
  409. resetCharsetMeta() {
  410. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  411. const metaElement = this.doc.createElement("meta");
  412. metaElement.setAttribute("charset", "utf-8");
  413. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  414. }
  415. insertFaviconLink() {
  416. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  417. if (!faviconElement) {
  418. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  419. }
  420. if (!faviconElement) {
  421. faviconElement = this.doc.createElement("link");
  422. faviconElement.setAttribute("type", "image/x-icon");
  423. faviconElement.setAttribute("rel", "shortcut icon");
  424. faviconElement.setAttribute("href", "/favicon.ico");
  425. }
  426. this.doc.head.appendChild(faviconElement);
  427. }
  428. resolveHrefs() {
  429. this.doc.querySelectorAll("[href]").forEach(element => {
  430. if (element.href) {
  431. const href = element.href.baseVal ? element.href.baseVal : element.href;
  432. const normalizedHref = DomUtil.normalizeURL(href);
  433. if (!normalizedHref || normalizedHref != this.baseURI) {
  434. element.setAttribute("href", href);
  435. }
  436. }
  437. });
  438. }
  439. removeUnusedStyles() {
  440. const stats = DOM.minifyCSS(this.doc);
  441. this.stats.set("processed", "CSS rules", stats.processed);
  442. this.stats.set("discarded", "CSS rules", stats.discarded);
  443. }
  444. removeAlternativeFonts() {
  445. DOM.minifyFonts(this.doc);
  446. }
  447. removeSrcSet() {
  448. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  449. const tagName = element.tagName.toLowerCase();
  450. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  451. const imageData = this.options.imageData[Number(element.getAttribute(dataAttributeName))];
  452. element.removeAttribute(dataAttributeName);
  453. if (imageData) {
  454. if (tagName == "img") {
  455. if (imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  456. element.removeAttribute("srcset");
  457. element.removeAttribute("sizes");
  458. element.src = imageData.source.src;
  459. }
  460. }
  461. if (tagName == "picture") {
  462. const imageElement = element.querySelector("img");
  463. if (imageData.source && imageData.source.src && imageData.source.naturalWidth > 1 && imageData.source.naturalHeight > 1) {
  464. imageElement.removeAttribute("srcset");
  465. imageElement.removeAttribute("sizes");
  466. imageElement.src = imageData.source.src;
  467. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  468. } else {
  469. if (imageData.sources) {
  470. element.querySelectorAll("source").forEach(sourceElement => {
  471. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  472. sourceElement.remove();
  473. }
  474. });
  475. const sourceElements = element.querySelectorAll("source");
  476. if (sourceElements.length) {
  477. const lastSourceElement = sourceElements[sourceElements.length - 1];
  478. if (imageElement) {
  479. if (lastSourceElement.src) {
  480. imageElement.src = lastSourceElement.src;
  481. } else {
  482. imageElement.removeAttribute("src");
  483. }
  484. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  485. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  486. } else {
  487. imageElement.removeAttribute("srcset");
  488. }
  489. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  490. }
  491. }
  492. }
  493. }
  494. }
  495. }
  496. });
  497. }
  498. postRemoveAlternativeFonts() {
  499. DOM.minifyFonts(this.doc, true);
  500. }
  501. removeHiddenElements() {
  502. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  503. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  504. this.stats.set("processed", "hidden elements", hiddenElements.length);
  505. hiddenElements.forEach(element => element.remove());
  506. }
  507. compressHTML() {
  508. let size;
  509. if (this.options.displayStats) {
  510. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  511. }
  512. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  513. if (this.options.displayStats) {
  514. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  515. }
  516. }
  517. postCompressHTML() {
  518. let size;
  519. if (this.options.displayStats) {
  520. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  521. }
  522. DOM.postMinifyHTML(this.doc);
  523. if (this.options.displayStats) {
  524. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  525. }
  526. }
  527. removeAlternativeMedias() {
  528. const stats = DOM.minifyMedias(this.doc);
  529. this.stats.set("processed", "medias", stats.processed);
  530. this.stats.set("discarded", "medias", stats.discarded);
  531. }
  532. compressCSS() {
  533. this.doc.querySelectorAll("style").forEach(styleElement => {
  534. if (styleElement) {
  535. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  536. }
  537. });
  538. this.doc.querySelectorAll("[style]").forEach(element => {
  539. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  540. });
  541. }
  542. insertSingleFileComment() {
  543. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  544. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  545. }
  546. replaceCanvasElements() {
  547. if (this.options.canvasData) {
  548. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  549. const canvasData = this.options.canvasData[indexCanvasElement];
  550. if (canvasData) {
  551. const imgElement = this.doc.createElement("img");
  552. imgElement.setAttribute("src", canvasData.dataURI);
  553. Array.from(canvasElement.attributes).forEach(attribute => {
  554. if (attribute.value) {
  555. imgElement.setAttribute(attribute.name, attribute.value);
  556. }
  557. });
  558. if (!imgElement.width && canvasData.width) {
  559. imgElement.style.pixelWidth = canvasData.width;
  560. }
  561. if (!imgElement.height && canvasData.height) {
  562. imgElement.style.pixelHeight = canvasData.height;
  563. }
  564. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  565. this.stats.add("processed", "canvas", 1);
  566. }
  567. });
  568. }
  569. }
  570. replaceStyleContents() {
  571. if (this.options.stylesheetContents) {
  572. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  573. if (this.options.stylesheetContents[styleIndex]) {
  574. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  575. }
  576. });
  577. }
  578. }
  579. async pageResources() {
  580. const resourcePromises = [
  581. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", this.baseURI, this.batchRequest),
  582. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", this.baseURI, this.batchRequest),
  583. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image], embed[src*=\".svg\"]"), "src", this.baseURI, this.batchRequest),
  584. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", this.baseURI, this.batchRequest),
  585. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", this.baseURI, this.batchRequest),
  586. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", this.baseURI, this.batchRequest),
  587. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  588. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("[srcset]"), "srcset", this.baseURI, this.batchRequest)
  589. ];
  590. if (!this.options.removeAudioSrc) {
  591. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.batchRequest));
  592. }
  593. if (!this.options.removeVideoSrc) {
  594. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.batchRequest));
  595. }
  596. if (this.options.lazyLoadImages) {
  597. const imageSelectors = DOM.lazyLoaderImageSelectors();
  598. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], this.baseURI, this.batchRequest)));
  599. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], this.baseURI, this.batchRequest)));
  600. }
  601. await resourcePromises;
  602. }
  603. async inlineStylesheets(initialization) {
  604. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  605. if (!initialization) {
  606. this.stats.add("processed", "stylesheets", 1);
  607. }
  608. let stylesheetContent = styleElement.textContent;
  609. if (initialization) {
  610. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  611. } else {
  612. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  613. }
  614. styleElement.textContent = stylesheetContent;
  615. }));
  616. }
  617. async scripts() {
  618. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  619. if (scriptElement.src) {
  620. this.stats.add("processed", "scripts", 1);
  621. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  622. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  623. }
  624. scriptElement.removeAttribute("src");
  625. }));
  626. }
  627. async frames(initialization) {
  628. if (this.options.framesData) {
  629. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  630. await Promise.all(frameElements.map(async frameElement => {
  631. DomProcessorHelper.setFrameEmptySrc(frameElement);
  632. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  633. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  634. if (frameWindowId) {
  635. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  636. if (frameData) {
  637. if (initialization) {
  638. const options = Object.create(this.options);
  639. options.insertSingleFileComment = false;
  640. options.insertFaviconLink = false;
  641. options.doc = null;
  642. options.win = null;
  643. options.url = frameData.baseURI;
  644. options.windowId = frameWindowId;
  645. if (frameData.content) {
  646. options.content = frameData.content;
  647. options.canvasData = frameData.canvasData;
  648. options.stylesheetContents = frameData.stylesheetContents;
  649. options.currentSrcImages = frameData.currentSrcImages;
  650. frameData.processor = new PageProcessor(options);
  651. frameData.frameElement = frameElement;
  652. await frameData.processor.loadPage();
  653. await frameData.processor.initialize();
  654. frameData.maxResources = this.batchRequest.getMaxResources();
  655. }
  656. } else {
  657. if (frameData.processor) {
  658. this.stats.add("processed", "frames", 1);
  659. await frameData.processor.preparePageData();
  660. const pageData = await frameData.processor.getPageData();
  661. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  662. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  663. this.stats.addAll(pageData);
  664. } else {
  665. this.stats.add("discarded", "frames", 1);
  666. }
  667. }
  668. }
  669. }
  670. }));
  671. }
  672. }
  673. async htmlImports(initialization) {
  674. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  675. if (!this.relImportProcessors) {
  676. this.relImportProcessors = new Map();
  677. }
  678. await Promise.all(linkElements.map(async linkElement => {
  679. if (initialization) {
  680. const resourceURL = linkElement.href;
  681. const options = Object.create(this.options);
  682. options.insertSingleFileComment = false;
  683. options.insertFaviconLink = false;
  684. options.doc = null;
  685. options.win = null;
  686. options.url = resourceURL;
  687. if (resourceURL) {
  688. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  689. const processor = new PageProcessor(options);
  690. this.relImportProcessors.set(linkElement, processor);
  691. await processor.loadPage();
  692. return processor.initialize();
  693. }
  694. }
  695. } else {
  696. linkElement.setAttribute("href", EMPTY_DATA_URI);
  697. const processor = this.relImportProcessors.get(linkElement);
  698. if (processor) {
  699. this.stats.add("processed", "HTML imports", 1);
  700. this.relImportProcessors.delete(linkElement);
  701. const pageData = await processor.getPageData();
  702. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  703. this.stats.addAll(pageData);
  704. } else {
  705. this.stats.add("discarded", "HTML imports", 1);
  706. }
  707. }
  708. }));
  709. }
  710. async attributeStyles(initialization) {
  711. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  712. let stylesheetContent = element.getAttribute("style");
  713. if (initialization) {
  714. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  715. } else {
  716. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  717. }
  718. element.setAttribute("style", stylesheetContent);
  719. }));
  720. }
  721. async linkStylesheets() {
  722. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  723. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  724. const styleElement = this.doc.createElement("style");
  725. styleElement.textContent = stylesheetContent;
  726. linkElement.parentElement.replaceChild(styleElement, linkElement);
  727. }));
  728. }
  729. }
  730. // ---------
  731. // DomHelper
  732. // ---------
  733. const REGEXP_AMP = /&/g;
  734. const REGEXP_NBSP = /\u00a0/g;
  735. const REGEXP_START_TAG = /</g;
  736. const REGEXP_END_TAG = />/g;
  737. const REGEXP_URL_HASH = /(#.+?)$/;
  738. const PREFIX_DATA_URI_IMAGE = "data:image/";
  739. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  740. class DomProcessorHelper {
  741. static async getFilename(options, content) {
  742. let filename = options.filenameTemplate;
  743. const date = new Date();
  744. const url = new URL(options.url);
  745. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  746. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  747. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  748. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  749. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  750. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  751. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  752. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  753. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  754. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  755. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  756. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  757. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  758. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  759. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  760. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  761. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  762. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  763. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  764. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  765. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  766. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  767. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  768. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  769. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  770. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  771. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  772. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  773. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  774. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  775. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  776. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  777. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  778. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  779. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  780. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  781. if (!options.backgroundSave) {
  782. filename = filename.replace(/\//g, "_");
  783. }
  784. if (filename.length > 192) {
  785. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  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 = "Unnamed 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 async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1010. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1011. if (template.match(replaceRegExp)) {
  1012. let value = await valueGetter();
  1013. if (!dontReplaceSlash) {
  1014. value = value.replace(/\/+/g, "_");
  1015. }
  1016. return template.replace(replaceRegExp, value);
  1017. }
  1018. return template;
  1019. }
  1020. static getLastSegment(url) {
  1021. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1022. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1023. if (!lastSegment) {
  1024. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1025. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1026. }
  1027. if (!lastSegment) {
  1028. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1029. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1030. }
  1031. if (!lastSegment) {
  1032. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1033. }
  1034. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1035. return lastSegment;
  1036. }
  1037. static getRegExp(string) {
  1038. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1039. }
  1040. static getUrlFunctions(stylesheetContent) {
  1041. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1042. }
  1043. static getImportFunctions(stylesheetContent) {
  1044. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1045. }
  1046. static matchURL(stylesheetContent) {
  1047. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1048. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1049. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1050. return match && match[1];
  1051. }
  1052. static testValidPath(resourceURL) {
  1053. 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));
  1054. }
  1055. static matchImport(stylesheetContent) {
  1056. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1057. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1058. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1059. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1060. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1061. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1062. if (match) {
  1063. const [, resourceURL, media] = match;
  1064. return { resourceURL, media };
  1065. }
  1066. }
  1067. static removeCssComments(stylesheetContent) {
  1068. let start, end;
  1069. do {
  1070. start = stylesheetContent.indexOf("/*");
  1071. end = stylesheetContent.indexOf("*/", start + 2);
  1072. if (start != -1 && end != -1) {
  1073. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1074. }
  1075. } while (start != -1 && end != -1);
  1076. return stylesheetContent;
  1077. }
  1078. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1079. if (mediaQuery) {
  1080. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1081. } else {
  1082. return stylesheetContent;
  1083. }
  1084. }
  1085. }
  1086. // -----
  1087. // Stats
  1088. // -----
  1089. const STATS_DEFAULT_VALUES = {
  1090. discarded: {
  1091. "HTML bytes": 0,
  1092. "hidden elements": 0,
  1093. "HTML imports": 0,
  1094. scripts: 0,
  1095. objects: 0,
  1096. "audio sources": 0,
  1097. "video sources": 0,
  1098. frames: 0,
  1099. "CSS rules": 0,
  1100. canvas: 0,
  1101. stylesheets: 0,
  1102. resources: 0,
  1103. medias: 0
  1104. },
  1105. processed: {
  1106. "HTML bytes": 0,
  1107. "hidden elements": 0,
  1108. "HTML imports": 0,
  1109. scripts: 0,
  1110. objects: 0,
  1111. "audio sources": 0,
  1112. "video sources": 0,
  1113. frames: 0,
  1114. "CSS rules": 0,
  1115. canvas: 0,
  1116. stylesheets: 0,
  1117. resources: 0,
  1118. medias: 0
  1119. }
  1120. };
  1121. class Stats {
  1122. constructor(options) {
  1123. this.options = options;
  1124. if (options.displayStats) {
  1125. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1126. }
  1127. }
  1128. set(type, subType, value) {
  1129. if (this.options.displayStats) {
  1130. this.data[type][subType] = value;
  1131. }
  1132. }
  1133. add(type, subType, value) {
  1134. if (this.options.displayStats) {
  1135. this.data[type][subType] += value;
  1136. }
  1137. }
  1138. addAll(pageData) {
  1139. if (this.options.displayStats) {
  1140. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1141. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1142. }
  1143. }
  1144. }
  1145. return { getClass };
  1146. })();