single-file-core.js 59 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399
  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. /* global CSSRule */
  21. this.SingleFileCore = this.SingleFileCore || (() => {
  22. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  23. const SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = "data-single-file-selected-content-root";
  24. const DEBUG = false;
  25. let Download, DOM, URL, sessionId = 0;
  26. function getClass(...args) {
  27. [Download, DOM, URL] = args;
  28. return class {
  29. constructor(options) {
  30. this.options = options;
  31. options.sessionId = sessionId;
  32. sessionId++;
  33. this.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  34. this.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  35. }
  36. async initialize() {
  37. this.processor = new PageProcessor(this.options);
  38. await this.processor.loadPage();
  39. await this.processor.initialize();
  40. }
  41. async preparePageData() {
  42. await this.processor.preparePageData();
  43. }
  44. async getPageData() {
  45. return this.processor.getPageData();
  46. }
  47. };
  48. }
  49. // -------------
  50. // ProgressEvent
  51. // -------------
  52. const PAGE_LOADING = "page-loading";
  53. const PAGE_LOADED = "page-loaded";
  54. const RESOURCES_INITIALIZING = "resource-initializing";
  55. const RESOURCES_INITIALIZED = "resources-initialized";
  56. const RESOURCE_LOADED = "resource-loaded";
  57. const PAGE_ENDED = "page-ended";
  58. class ProgressEvent {
  59. constructor(type, details) {
  60. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  61. }
  62. }
  63. // -------------
  64. // PageProcessor
  65. // -------------
  66. const STAGES = [{
  67. sequential: [
  68. { action: "removeUIElements" },
  69. { action: "replaceStyleContents" },
  70. { option: "removeVideoSrc", action: "insertVideoPosters" },
  71. { option: "removeSrcSet", action: "removeSrcSet" },
  72. { option: "removeFrames", action: "removeFrames" },
  73. { option: "removeImports", action: "removeImports" },
  74. { option: "removeScripts", action: "removeScripts" },
  75. { action: "removeDiscardedResources" },
  76. { action: "resetCharsetMeta" },
  77. { action: "setInputValues" },
  78. { option: "insertFaviconLink", action: "insertFaviconLink" },
  79. { action: "resolveHrefs" },
  80. { action: "replaceCanvasElements" },
  81. { action: "insertFonts" },
  82. { option: "removeHiddenElements", action: "removeHiddenElements" }
  83. ],
  84. parallel: [
  85. { action: "resolveStylesheetURLs" },
  86. { action: "resolveLinkedStylesheetURLs" },
  87. { action: "resolveStyleAttributeURLs" },
  88. { option: "!removeFrames", action: "resolveFrameURLs" },
  89. { option: "!removeImports", action: "resolveHtmlImportURLs" }
  90. ]
  91. }, {
  92. sequential: [
  93. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  94. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  95. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  96. ],
  97. parallel: [
  98. { action: "processStylesheets" },
  99. { action: "processStyleAttributes" },
  100. { action: "pageResources" },
  101. { option: "!removeScripts", action: "processScripts" }
  102. ]
  103. }, {
  104. sequential: [
  105. { option: "lazyLoadImages", action: "lazyLoadImages" },
  106. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  107. { option: "compressCSS", action: "compressCSS" }
  108. ],
  109. parallel: [
  110. { option: "!removeFrames", action: "processFrames" },
  111. { option: "!removeImports", action: "processHtmlImports" },
  112. ]
  113. }, {
  114. sequential: [
  115. { option: "compressHTML", action: "compressHTML" },
  116. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  117. { action: "removeDefaultHeadTags" }
  118. ]
  119. }];
  120. class PageProcessor {
  121. constructor(options) {
  122. this.options = options;
  123. this.options.url = this.options.url || this.options.doc.location.href;
  124. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  125. this.batchRequest = new BatchRequest();
  126. this.processor = new DOMProcessor(options, this.batchRequest);
  127. if (this.options.doc) {
  128. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  129. this.options.canvasData = docData.canvasData;
  130. this.options.fontsData = docData.fontsData;
  131. this.options.stylesheetContents = docData.stylesheetContents;
  132. this.options.responsiveImageData = docData.responsiveImageData;
  133. this.options.imageData = docData.imageData;
  134. this.options.postersData = docData.postersData;
  135. }
  136. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  137. this.onprogress = options.onprogress || (() => { });
  138. }
  139. async loadPage() {
  140. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  141. await this.processor.loadPage(this.options.content);
  142. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  143. }
  144. async initialize() {
  145. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  146. await this.executeStage(0);
  147. this.pendingPromises = this.executeStage(1);
  148. if (this.options.doc) {
  149. DOM.postProcessDoc(this.options.doc, this.options);
  150. this.options.doc = null;
  151. this.options.win = null;
  152. }
  153. }
  154. async preparePageData() {
  155. if (!this.options.windowId) {
  156. this.processor.initialize(this.batchRequest);
  157. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  158. }
  159. await this.batchRequest.run(details => {
  160. details.pageURL = this.options.url;
  161. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  162. }, this.options);
  163. await this.pendingPromises;
  164. await this.executeStage(2);
  165. await this.executeStage(3);
  166. }
  167. async getPageData() {
  168. if (!this.options.windowId) {
  169. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  170. }
  171. return this.processor.getPageData();
  172. }
  173. async executeStage(step) {
  174. if (DEBUG) {
  175. log("**** STARTED STAGE", step, "****");
  176. }
  177. STAGES[step].sequential.forEach(task => {
  178. let startTime;
  179. if (DEBUG) {
  180. startTime = Date.now();
  181. log(" -- STARTED task =", task.action);
  182. }
  183. this.executeTask(task);
  184. if (DEBUG) {
  185. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  186. }
  187. });
  188. if (STAGES[step].parallel) {
  189. return await Promise.all(STAGES[step].parallel.map(task => {
  190. let startTime;
  191. if (DEBUG) {
  192. startTime = Date.now();
  193. log(" // STARTED task =", task.action);
  194. }
  195. const promise = this.executeTask(task);
  196. if (DEBUG) {
  197. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  198. }
  199. return promise;
  200. }));
  201. }
  202. if (DEBUG) {
  203. log("**** ENDED STAGE", step, "****");
  204. }
  205. }
  206. executeTask(task) {
  207. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  208. return this.processor[task.action]();
  209. }
  210. }
  211. }
  212. // --------
  213. // BatchRequest
  214. // --------
  215. class BatchRequest {
  216. constructor() {
  217. this.requests = new Map();
  218. }
  219. async addURL(resourceURL, asDataURI = true) {
  220. return new Promise((resolve, reject) => {
  221. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  222. const resourceRequests = this.requests.get(requestKey);
  223. if (resourceRequests) {
  224. resourceRequests.push({ resolve, reject });
  225. } else {
  226. this.requests.set(requestKey, [{ resolve, reject }]);
  227. }
  228. });
  229. }
  230. getMaxResources() {
  231. return Array.from(this.requests.keys()).length;
  232. }
  233. async run(onloadListener, options) {
  234. const resourceURLs = Array.from(this.requests.keys());
  235. let indexResource = 0;
  236. return Promise.all(resourceURLs.map(async requestKey => {
  237. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  238. const resourceRequests = this.requests.get(requestKey);
  239. try {
  240. const resourceContent = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  241. indexResource = indexResource + 1;
  242. onloadListener({ index: indexResource, url: resourceURL });
  243. resourceRequests.forEach(resourceRequest => resourceRequest.resolve({ content: resourceContent, indexResource, duplicate: Boolean(resourceRequests.length > 1) }));
  244. } catch (error) {
  245. indexResource = indexResource + 1;
  246. onloadListener({ index: indexResource, url: resourceURL });
  247. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  248. }
  249. this.requests.delete(requestKey);
  250. }));
  251. }
  252. }
  253. // ------------
  254. // DOMProcessor
  255. // ------------
  256. const EMPTY_DATA_URI = "data:base64,";
  257. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  258. class DOMProcessor {
  259. constructor(options, batchRequest) {
  260. this.options = options;
  261. this.stats = new Stats(options);
  262. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  263. this.batchRequest = batchRequest;
  264. }
  265. initialize() {
  266. this.maxResources = this.batchRequest.getMaxResources();
  267. if (!this.options.removeFrames) {
  268. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  269. }
  270. this.stats.set("processed", "resources", this.maxResources);
  271. }
  272. async loadPage(pageContent) {
  273. if (!pageContent || this.options.saveRawPage) {
  274. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  275. }
  276. this.doc = DOM.createDoc(pageContent, this.baseURI);
  277. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  278. }
  279. async getPageData() {
  280. DOM.postProcessDoc(this.doc, this.options);
  281. if (this.options.selected) {
  282. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  283. if (rootElement) {
  284. DomProcessorHelper.isolateElements(rootElement);
  285. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  286. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  287. }
  288. }
  289. const titleElement = this.doc.querySelector("title");
  290. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  291. this.options.info = {};
  292. const descriptionElement = this.doc.querySelector("meta[name=description]");
  293. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  294. this.options.info.lang = this.doc.documentElement.lang;
  295. const authorElement = this.doc.querySelector("meta[name=author]");
  296. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  297. const creatorElement = this.doc.querySelector("meta[name=creator]");
  298. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  299. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  300. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  301. const url = new URL(this.baseURI);
  302. let size;
  303. if (this.options.displayStats) {
  304. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  305. }
  306. const content = DOM.serialize(this.doc, this.options.compressHTML);
  307. if (this.options.displayStats) {
  308. const contentSize = DOM.getContentSize(content);
  309. this.stats.set("processed", "HTML bytes", contentSize);
  310. this.stats.add("discarded", "HTML bytes", size - contentSize);
  311. }
  312. const filename = await DomProcessorHelper.getFilename(this.options, content);
  313. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  314. return {
  315. stats: this.stats.data,
  316. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  317. filename,
  318. content
  319. };
  320. }
  321. setInputValues() {
  322. this.doc.querySelectorAll("input").forEach(input => {
  323. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  324. if (value) {
  325. input.setAttribute("value", value);
  326. }
  327. });
  328. this.doc.querySelectorAll("textarea").forEach(textarea => {
  329. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  330. if (value) {
  331. textarea.textContent = value;
  332. }
  333. });
  334. this.doc.querySelectorAll("select").forEach(select => {
  335. select.querySelectorAll("option").forEach(option => {
  336. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  337. if (selected) {
  338. option.setAttribute("selected", "");
  339. }
  340. });
  341. });
  342. }
  343. lazyLoadImages() {
  344. DOM.lazyLoad(this.doc);
  345. }
  346. removeDiscardedResources() {
  347. 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\"])");
  348. this.stats.set("discarded", "objects", objectElements.length);
  349. this.stats.set("processed", "objects", objectElements.length);
  350. objectElements.forEach(element => element.remove());
  351. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  352. replacedAttributeValue.forEach(element => {
  353. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  354. if (relValue.length) {
  355. element.setAttribute("rel", relValue);
  356. } else {
  357. element.remove();
  358. }
  359. });
  360. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  361. if (this.options.compressHTML) {
  362. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  363. }
  364. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  365. if (this.options.removeScripts) {
  366. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  367. this.doc.querySelectorAll("[href]").forEach(element => {
  368. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  369. element.removeAttribute("href");
  370. }
  371. });
  372. this.doc.querySelectorAll("[src]").forEach(element => {
  373. if (element.src && element.src.match(/^\s*javascript:/)) {
  374. element.removeAttribute("src");
  375. }
  376. });
  377. }
  378. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  379. this.stats.set("processed", "audio sources", audioSourceElements.length);
  380. if (this.options.removeAudioSrc) {
  381. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  382. audioSourceElements.forEach(element => element.removeAttribute("src"));
  383. }
  384. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  385. this.stats.set("processed", "video sources", videoSourceElements.length);
  386. if (this.options.removeVideoSrc) {
  387. this.stats.set("discarded", "video sources", videoSourceElements.length);
  388. videoSourceElements.forEach(element => element.removeAttribute("src"));
  389. }
  390. }
  391. removeDefaultHeadTags() {
  392. this.doc.querySelectorAll("base").forEach(element => element.remove());
  393. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  394. this.doc.head.querySelector("meta[charset]").remove();
  395. }
  396. }
  397. removeUIElements() {
  398. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  399. }
  400. removeScripts() {
  401. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  402. this.stats.set("discarded", "scripts", scriptElements.length);
  403. this.stats.set("processed", "scripts", scriptElements.length);
  404. scriptElements.forEach(element => element.remove());
  405. }
  406. removeFrames() {
  407. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  408. this.stats.set("discarded", "frames", frameElements.length);
  409. this.stats.set("processed", "frames", frameElements.length);
  410. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  411. }
  412. removeImports() {
  413. const importElements = this.doc.querySelectorAll("link[rel=import]");
  414. this.stats.set("discarded", "HTML imports", importElements.length);
  415. this.stats.set("processed", "HTML imports", importElements.length);
  416. importElements.forEach(element => element.remove());
  417. }
  418. resetCharsetMeta() {
  419. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  420. const charSetDeclaration = element.content.split(";")[1];
  421. if (charSetDeclaration) {
  422. const charSet = charSetDeclaration.split("=")[1];
  423. if (charSet) {
  424. this.charSet = charSet.trim().toLowerCase();
  425. }
  426. }
  427. element.remove();
  428. });
  429. const metaElement = this.doc.createElement("meta");
  430. metaElement.setAttribute("charset", "utf-8");
  431. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  432. }
  433. insertFaviconLink() {
  434. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  435. if (!faviconElement) {
  436. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  437. }
  438. if (!faviconElement) {
  439. faviconElement = this.doc.createElement("link");
  440. faviconElement.setAttribute("type", "image/x-icon");
  441. faviconElement.setAttribute("rel", "shortcut icon");
  442. faviconElement.setAttribute("href", "/favicon.ico");
  443. }
  444. this.doc.head.appendChild(faviconElement);
  445. }
  446. resolveHrefs() {
  447. this.doc.querySelectorAll("[href]").forEach(element => {
  448. if (element.href) {
  449. const href = element.href.baseVal ? element.href.baseVal : element.href;
  450. const normalizedHref = DomUtil.normalizeURL(href);
  451. if (!normalizedHref || normalizedHref != this.baseURI) {
  452. element.setAttribute("href", href);
  453. }
  454. }
  455. });
  456. }
  457. removeUnusedStyles() {
  458. if (!this.mediaAllInfo) {
  459. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  460. }
  461. const stats = DOM.minifyCSS(this.doc, this.mediaAllInfo);
  462. this.stats.set("processed", "CSS rules", stats.processed);
  463. this.stats.set("discarded", "CSS rules", stats.discarded);
  464. }
  465. removeAlternativeFonts() {
  466. DOM.minifyFonts(this.doc);
  467. }
  468. removeSrcSet() {
  469. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  470. const tagName = element.tagName.toLowerCase();
  471. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  472. const responsiveImageData = this.options.responsiveImageData[Number(element.getAttribute(dataAttributeName))];
  473. element.removeAttribute(dataAttributeName);
  474. if (responsiveImageData) {
  475. if (tagName == "img") {
  476. if (responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  477. element.removeAttribute("srcset");
  478. element.removeAttribute("sizes");
  479. element.src = responsiveImageData.source.src;
  480. }
  481. }
  482. if (tagName == "picture") {
  483. const imageElement = element.querySelector("img");
  484. if (responsiveImageData.source && responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  485. imageElement.removeAttribute("srcset");
  486. imageElement.removeAttribute("sizes");
  487. imageElement.src = responsiveImageData.source.src;
  488. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  489. } else {
  490. if (responsiveImageData.sources) {
  491. element.querySelectorAll("source").forEach(sourceElement => {
  492. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  493. sourceElement.remove();
  494. }
  495. });
  496. const sourceElements = element.querySelectorAll("source");
  497. if (sourceElements.length) {
  498. const lastSourceElement = sourceElements[sourceElements.length - 1];
  499. if (imageElement) {
  500. if (lastSourceElement.src) {
  501. imageElement.src = lastSourceElement.src;
  502. } else {
  503. imageElement.removeAttribute("src");
  504. }
  505. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  506. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  507. } else {
  508. imageElement.removeAttribute("srcset");
  509. }
  510. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  511. }
  512. }
  513. }
  514. }
  515. }
  516. }
  517. });
  518. }
  519. postRemoveAlternativeFonts() {
  520. DOM.minifyFonts(this.doc, true);
  521. }
  522. removeHiddenElements() {
  523. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  524. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  525. this.stats.set("processed", "hidden elements", hiddenElements.length);
  526. hiddenElements.forEach(element => element.remove());
  527. }
  528. compressHTML() {
  529. let size;
  530. if (this.options.displayStats) {
  531. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  532. }
  533. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  534. if (this.options.displayStats) {
  535. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  536. }
  537. }
  538. removeAlternativeMedias() {
  539. const stats = DOM.minifyMedias(this.doc);
  540. this.stats.set("processed", "medias", stats.processed);
  541. this.stats.set("discarded", "medias", stats.discarded);
  542. }
  543. compressCSS() {
  544. this.doc.querySelectorAll("style").forEach(styleElement => {
  545. if (styleElement) {
  546. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  547. }
  548. });
  549. this.doc.querySelectorAll("[style]").forEach(element => {
  550. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  551. });
  552. }
  553. insertSingleFileComment() {
  554. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  555. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  556. }
  557. replaceCanvasElements() {
  558. if (this.options.canvasData) {
  559. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  560. const canvasData = this.options.canvasData[indexCanvasElement];
  561. if (canvasData) {
  562. DomProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  563. this.stats.add("processed", "canvas", 1);
  564. }
  565. });
  566. }
  567. }
  568. insertFonts() {
  569. if (this.options.fontsData && this.options.fontsData.length) {
  570. let stylesheetContent = "";
  571. this.options.fontsData.forEach(fontStyles => {
  572. if (fontStyles["font-family"] && fontStyles.src) {
  573. stylesheetContent += "@font-face{";
  574. let stylesContent = "";
  575. Object.keys(fontStyles).forEach(fontStyle => {
  576. if (stylesContent) {
  577. stylesContent += ";";
  578. }
  579. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  580. });
  581. stylesheetContent += stylesContent + "}";
  582. }
  583. });
  584. if (stylesheetContent) {
  585. const styleElement = this.doc.createElement("style");
  586. styleElement.textContent = stylesheetContent;
  587. const existingStyleElement = this.doc.querySelector("style");
  588. if (existingStyleElement) {
  589. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  590. } else {
  591. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  592. }
  593. }
  594. }
  595. }
  596. replaceStyleContents() {
  597. if (this.options.stylesheetContents) {
  598. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  599. if (this.options.stylesheetContents[styleIndex]) {
  600. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  601. }
  602. });
  603. }
  604. }
  605. insertVideoPosters() {
  606. if (this.options.postersData) {
  607. this.doc.querySelectorAll("video[src], video > source[src]").forEach((videoElement, videoIndex) => {
  608. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  609. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  610. }
  611. });
  612. }
  613. }
  614. async pageResources() {
  615. const resourcePromises = [
  616. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest, this.options, false, true),
  617. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest, this.options),
  618. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("img[src], input[src][type=image]"), "src", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest, this.options, true),
  619. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest, this.options),
  620. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest, this.options),
  621. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest, this.options),
  622. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest, this.options),
  623. DomProcessorHelper.processXLinks(this.doc, this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  624. DomProcessorHelper.processSrcset(this.doc, this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)
  625. ];
  626. if (!this.options.removeAudioSrc) {
  627. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", PREFIX_DATA_URI_AUDIO, this.baseURI, this.batchRequest, this.options));
  628. }
  629. if (!this.options.removeVideoSrc) {
  630. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("video[src], video > source[src]"), "src", PREFIX_DATA_URI_VIDEO, this.baseURI, this.batchRequest, this.options));
  631. }
  632. if (this.options.lazyLoadImages) {
  633. const imageSelectors = DOM.lazyLoaderImageSelectors();
  634. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll(selector), imageSelectors.src[selector], PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest, this.options)));
  635. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc, this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)));
  636. }
  637. await resourcePromises;
  638. }
  639. async resolveStylesheetURLs() {
  640. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => styleElement.textContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled })));
  641. }
  642. async processStylesheets() {
  643. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => {
  644. this.stats.add("processed", "stylesheets", 1);
  645. styleElement.textContent = await DomProcessorHelper.processStylesheet(this.doc, styleElement.textContent, styleElement.sheet.cssRules, this.baseURI, this.options, this.batchRequest);
  646. }));
  647. }
  648. async processScripts() {
  649. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  650. if (scriptElement.src) {
  651. this.stats.add("processed", "scripts", 1);
  652. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  653. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  654. }
  655. scriptElement.removeAttribute("src");
  656. }));
  657. }
  658. async resolveFrameURLs() {
  659. if (this.options.framesData) {
  660. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  661. await Promise.all(frameElements.map(async frameElement => {
  662. DomProcessorHelper.setFrameEmptySrc(frameElement);
  663. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  664. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  665. if (frameWindowId) {
  666. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  667. if (frameData) {
  668. const options = Object.create(this.options);
  669. options.insertSingleFileComment = false;
  670. options.insertFaviconLink = false;
  671. options.doc = null;
  672. options.win = null;
  673. options.url = frameData.baseURI;
  674. options.windowId = frameWindowId;
  675. if (frameData.content) {
  676. options.content = frameData.content;
  677. options.canvasData = frameData.canvasData;
  678. options.stylesheetContents = frameData.stylesheetContents;
  679. options.currentSrcImages = frameData.currentSrcImages;
  680. options.fontsData = frameData.fontsData;
  681. options.imageData = frameData.imageData;
  682. options.responsiveImageData = frameData.responsiveImageData;
  683. frameData.processor = new PageProcessor(options);
  684. frameData.frameElement = frameElement;
  685. await frameData.processor.loadPage();
  686. await frameData.processor.initialize();
  687. frameData.maxResources = this.batchRequest.getMaxResources();
  688. }
  689. }
  690. }
  691. }));
  692. }
  693. }
  694. async processFrames() {
  695. if (this.options.framesData) {
  696. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  697. await Promise.all(frameElements.map(async frameElement => {
  698. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  699. if (frameWindowId) {
  700. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  701. if (frameData) {
  702. if (frameData.processor) {
  703. this.stats.add("processed", "frames", 1);
  704. await frameData.processor.preparePageData();
  705. const pageData = await frameData.processor.getPageData();
  706. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  707. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  708. this.stats.addAll(pageData);
  709. } else {
  710. this.stats.add("discarded", "frames", 1);
  711. }
  712. }
  713. }
  714. }));
  715. }
  716. }
  717. async resolveHtmlImportURLs() {
  718. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  719. if (!this.relImportProcessors) {
  720. this.relImportProcessors = new Map();
  721. }
  722. await Promise.all(linkElements.map(async linkElement => {
  723. const resourceURL = linkElement.href;
  724. linkElement.setAttribute("href", EMPTY_DATA_URI);
  725. const options = Object.create(this.options);
  726. options.insertSingleFileComment = false;
  727. options.insertFaviconLink = false;
  728. options.doc = null;
  729. options.win = null;
  730. options.url = resourceURL;
  731. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, this.baseURI)) {
  732. const processor = new PageProcessor(options);
  733. this.relImportProcessors.set(linkElement, processor);
  734. await processor.loadPage();
  735. return processor.initialize();
  736. }
  737. }));
  738. }
  739. async processHtmlImports() {
  740. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  741. await Promise.all(linkElements.map(async linkElement => {
  742. const processor = this.relImportProcessors.get(linkElement);
  743. if (processor) {
  744. this.stats.add("processed", "HTML imports", 1);
  745. this.relImportProcessors.delete(linkElement);
  746. const pageData = await processor.getPageData();
  747. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  748. this.stats.addAll(pageData);
  749. } else {
  750. this.stats.add("discarded", "HTML imports", 1);
  751. }
  752. }));
  753. }
  754. async resolveStyleAttributeURLs() {
  755. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => element.setAttribute("style", DomProcessorHelper.resolveStylesheetURLs(element.getAttribute("style"), this.baseURI))));
  756. }
  757. async processStyleAttributes() {
  758. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => element.setAttribute("style", await DomProcessorHelper.processStylesheet(this.doc, element.getAttribute("style"), [{ type: CSSRule.STYLE_RULE, cssText: element.getAttribute("style") }], this.baseURI, this.options, this.batchRequest))));
  759. }
  760. async resolveLinkedStylesheetURLs() {
  761. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  762. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charSet: this.charSet };
  763. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, options);
  764. const styleElement = this.doc.createElement("style");
  765. styleElement.textContent = stylesheetContent;
  766. linkElement.parentElement.replaceChild(styleElement, linkElement);
  767. }));
  768. }
  769. }
  770. // ---------
  771. // DomHelper
  772. // ---------
  773. const REGEXP_AMP = /&/g;
  774. const REGEXP_NBSP = /\u00a0/g;
  775. const REGEXP_START_TAG = /</g;
  776. const REGEXP_END_TAG = />/g;
  777. const REGEXP_URL_HASH = /(#.+?)$/;
  778. const PREFIX_DATA_URI_IMAGE = "data:image/";
  779. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  780. const PREFIX_DATA_URI_VIDEO = "data:video/";
  781. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  782. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  783. const PREFIX_DATA_URI_OCTET_STREAM = "data:application/octet-stream";
  784. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  785. class DomProcessorHelper {
  786. static async getFilename(options, content) {
  787. let filename = options.filenameTemplate;
  788. const date = new Date();
  789. const url = new URL(options.url);
  790. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  791. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  792. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  793. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  794. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  795. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  796. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  797. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  798. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  799. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  800. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  801. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  802. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  803. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  804. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  805. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  806. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  807. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  808. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  809. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  810. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  811. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  812. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  813. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  814. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  815. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  816. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  817. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  818. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  819. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  820. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  821. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  822. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  823. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  824. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  825. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  826. if (!options.backgroundSave) {
  827. filename = filename.replace(/\//g, "_");
  828. }
  829. if (filename.length > 192) {
  830. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  831. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  832. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  833. }
  834. if (!filename) {
  835. filename = "Unnamed page";
  836. }
  837. return filename;
  838. }
  839. static setBackgroundImage(element, url) {
  840. element.style.setProperty("background-blend-mode", "normal");
  841. element.style.setProperty("background-clip", "content-box");
  842. element.style.setProperty("background-position", "center");
  843. element.style.setProperty("background-color", "transparent", "important");
  844. element.style.setProperty("background-image", url, "important");
  845. element.style.setProperty("background-size", "contain", "important");
  846. element.style.setProperty("background-origin", "content-box", "important");
  847. element.style.setProperty("background-repeat", "no-repeat", "important");
  848. }
  849. static setFrameEmptySrc(frameElement) {
  850. if (frameElement.tagName == "OBJECT") {
  851. frameElement.setAttribute("data", "data:text/html,");
  852. } else {
  853. frameElement.setAttribute("srcdoc", "");
  854. frameElement.removeAttribute("src");
  855. }
  856. }
  857. static setFrameContent(frameElement, content) {
  858. if (frameElement.tagName == "OBJECT") {
  859. frameElement.setAttribute("data", "data:text/html," + content);
  860. } else {
  861. frameElement.setAttribute("srcdoc", content);
  862. frameElement.removeAttribute("src");
  863. }
  864. }
  865. static isolateElements(rootElement) {
  866. rootElement.querySelectorAll("*").forEach(element => {
  867. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  868. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  869. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  870. element.remove();
  871. }
  872. });
  873. isolateParentElements(rootElement.parentElement, rootElement);
  874. function isolateParentElements(parentElement, element) {
  875. if (parentElement) {
  876. Array.from(parentElement.childNodes).forEach(node => {
  877. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  878. node.remove();
  879. }
  880. });
  881. }
  882. element = element.parentElement;
  883. if (element && element.parentElement) {
  884. isolateParentElements(element.parentElement, element);
  885. }
  886. }
  887. }
  888. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  889. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  890. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  891. const imports = DomUtil.getImportFunctions(stylesheetContent);
  892. await Promise.all(imports.map(async cssImport => {
  893. const match = DomUtil.matchImport(cssImport);
  894. if (match) {
  895. let resourceURL = DomUtil.normalizeURL(match.resourceURL);
  896. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL)) {
  897. resourceURL = new URL(match.resourceURL, baseURI).href;
  898. if (DomUtil.testValidURL(resourceURL, baseURI)) {
  899. let importedStylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  900. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  901. if (stylesheetContent.includes(cssImport)) {
  902. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options);
  903. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  904. }
  905. }
  906. }
  907. }
  908. }));
  909. return stylesheetContent;
  910. }
  911. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  912. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  913. urlFunctions.map(urlFunction => {
  914. const originalResourceURL = DomUtil.matchURL(urlFunction);
  915. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  916. if (!DomUtil.testIgnoredPath(resourceURL)) {
  917. if (DomUtil.testValidPath(resourceURL, baseURI)) {
  918. const resolvedURL = new URL(resourceURL, baseURI).href;
  919. if (DomUtil.testValidURL(resolvedURL, baseURI) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  920. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  921. }
  922. }
  923. } else {
  924. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  925. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  926. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  927. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  928. }
  929. }
  930. }
  931. });
  932. return stylesheetContent;
  933. }
  934. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  935. resourceURL = DomUtil.normalizeURL(resourceURL);
  936. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  937. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charSet: options.charSet });
  938. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  939. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  940. return stylesheetContent;
  941. }
  942. }
  943. static async processStylesheet(doc, stylesheetContent, cssRules, baseURI, options, batchRequest) {
  944. let sheetContent = "", variablesInfo = { index: 0, cssText: "" };
  945. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  946. const resourceInfos = new Map();
  947. await Promise.all(urlFunctions.map(async urlFunction => {
  948. const originalResourceURL = DomUtil.matchURL(urlFunction);
  949. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  950. if (!DomUtil.testIgnoredPath(resourceURL)) {
  951. if (DomUtil.testValidURL(resourceURL, baseURI) && stylesheetContent.includes(urlFunction)) {
  952. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  953. const regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  954. if (duplicate && options.groupDuplicateImages) {
  955. resourceInfos.set(resourceURL, { regExpUrlFunction, dataURI: content, variableName: "var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")" });
  956. DomUtil.insertVariable(doc, indexResource, content, options);
  957. } else {
  958. resourceInfos.set(resourceURL, { regExpUrlFunction, dataURI: content });
  959. }
  960. }
  961. }
  962. }));
  963. const rulesContent = processRules(cssRules);
  964. if (variablesInfo.cssText) {
  965. sheetContent += ":root{" + variablesInfo.cssText + "}";
  966. }
  967. return sheetContent + rulesContent;
  968. function processRules(cssRules) {
  969. let rulesContent = "";
  970. Array.from(cssRules).forEach(cssRule => {
  971. if (cssRule.type == CSSRule.MEDIA_RULE) {
  972. const mediaRulesContent = processRules(cssRule.cssRules);
  973. rulesContent += "@media " + Array.from(cssRule.media).join(",") + "{" + mediaRulesContent + "}";
  974. } else if (cssRule.type == CSSRule.STYLE_RULE) {
  975. rulesContent += processURLFunctions(cssRule.cssText);
  976. } else {
  977. rulesContent += processURLFunctions(cssRule.cssText);
  978. }
  979. });
  980. return rulesContent;
  981. }
  982. function processURLFunctions(cssText) {
  983. const urlFunctions = DomUtil.getUrlFunctions(cssText);
  984. urlFunctions.forEach(urlFunction => {
  985. const originalResourceURL = DomUtil.matchURL(urlFunction);
  986. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  987. if (!DomUtil.testIgnoredPath(resourceURL)) {
  988. if (DomUtil.testValidURL(resourceURL, baseURI) && cssText.includes(urlFunction)) {
  989. const resourceInfo = resourceInfos.get(resourceURL);
  990. if (options.groupDuplicateImages && resourceInfo.variableName) {
  991. cssText = cssText.replace(resourceInfo.regExpUrlFunction, resourceInfo.variableName);
  992. } else {
  993. cssText = cssText.replace(resourceInfo.regExpUrlFunction, urlFunction.replace(originalResourceURL, resourceInfo.dataURI));
  994. }
  995. }
  996. }
  997. });
  998. return cssText;
  999. }
  1000. }
  1001. static async processAttribute(doc, resourceElements, attributeName, prefixDataURI, baseURI, batchRequest, options, processDuplicates, removeElementIfMissing) {
  1002. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1003. let resourceURL = resourceElement.getAttribute(attributeName);
  1004. resourceURL = DomUtil.normalizeURL(resourceURL);
  1005. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1006. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1007. if (DomUtil.testValidPath(resourceURL, baseURI)) {
  1008. resourceURL = new URL(resourceURL, baseURI).href;
  1009. if (DomUtil.testValidURL(resourceURL, baseURI)) {
  1010. try {
  1011. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1012. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1013. resourceElement.remove();
  1014. } else {
  1015. if (content.startsWith(prefixDataURI) || content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) || content.startsWith(PREFIX_DATA_URI_OCTET_STREAM)) {
  1016. if (processDuplicates && duplicate && options.groupDuplicateImages && DomUtil.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1017. DomUtil.insertVariable(doc, indexResource, content, options);
  1018. } else {
  1019. resourceElement.setAttribute(attributeName, content);
  1020. }
  1021. }
  1022. }
  1023. } catch (error) {
  1024. /* ignored */
  1025. }
  1026. }
  1027. }
  1028. }
  1029. }));
  1030. }
  1031. static async processXLinks(doc, resourceElements, baseURI, batchRequest) {
  1032. const attributeName = "xlink:href";
  1033. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1034. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1035. let resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1036. if (DomUtil.testValidPath(resourceURL) && !DomUtil.testIgnoredPath(resourceURL)) {
  1037. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1038. resourceURL = new URL(resourceURL, baseURI).href;
  1039. if (DomUtil.testValidURL(resourceURL, baseURI)) {
  1040. try {
  1041. const { content } = await batchRequest.addURL(resourceURL, false);
  1042. const DOMParser = DOM.getParser();
  1043. if (DOMParser) {
  1044. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  1045. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1046. if (hashMatch && hashMatch[0]) {
  1047. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1048. if (symbolElement) {
  1049. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1050. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1051. }
  1052. } else {
  1053. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1054. }
  1055. } else {
  1056. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1057. }
  1058. } catch (error) {
  1059. /* ignored */
  1060. }
  1061. }
  1062. }
  1063. }));
  1064. }
  1065. static async processSrcset(doc, resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  1066. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1067. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1068. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1069. let resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1070. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1071. if (DomUtil.testValidPath(resourceURL)) {
  1072. resourceURL = new URL(resourceURL, baseURI).href;
  1073. if (DomUtil.testValidURL(resourceURL, baseURI)) {
  1074. try {
  1075. const { content } = await batchRequest.addURL(resourceURL);
  1076. if (!content.startsWith(prefixDataURI) && !content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) && !content.startsWith(PREFIX_DATA_URI_OCTET_STREAM)) {
  1077. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1078. }
  1079. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1080. } catch (error) {
  1081. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1082. }
  1083. } else {
  1084. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1085. }
  1086. } else {
  1087. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1088. }
  1089. }
  1090. }));
  1091. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1092. }));
  1093. }
  1094. }
  1095. // -------
  1096. // DomUtil
  1097. // -------
  1098. const DATA_URI_PREFIX = "data:";
  1099. const BLOB_URI_PREFIX = "blob:";
  1100. const HTTP_URI_PREFIX = /^https?:\/\//;
  1101. const EMPTY_URL = /^https?:\/\/+.*$/;
  1102. const ABOUT_BLANK_URI = "about:blank";
  1103. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1104. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1105. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1106. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1107. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1108. 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;
  1109. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1110. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1111. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1112. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1113. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1114. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1115. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1116. class DomUtil {
  1117. static normalizeURL(url) {
  1118. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1119. return url;
  1120. } else {
  1121. return url.split("#")[0];
  1122. }
  1123. }
  1124. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1125. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1126. if (template.match(replaceRegExp)) {
  1127. let value = await valueGetter();
  1128. if (!dontReplaceSlash) {
  1129. value = value.replace(/\/+/g, "_");
  1130. }
  1131. return template.replace(replaceRegExp, value);
  1132. }
  1133. return template;
  1134. }
  1135. static getLastSegment(url) {
  1136. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1137. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1138. if (!lastSegment) {
  1139. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1140. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1141. }
  1142. if (!lastSegment) {
  1143. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1144. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1145. }
  1146. if (!lastSegment) {
  1147. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1148. }
  1149. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1150. return lastSegment;
  1151. }
  1152. static getRegExp(string) {
  1153. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1154. }
  1155. static getUrlFunctions(stylesheetContent) {
  1156. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1157. }
  1158. static getImportFunctions(stylesheetContent) {
  1159. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1160. }
  1161. static matchURL(stylesheetContent) {
  1162. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1163. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1164. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1165. return match && match[1];
  1166. }
  1167. static testIgnoredPath(resourceURL) {
  1168. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1169. }
  1170. static testValidPath(resourceURL, baseURI) {
  1171. return resourceURL && resourceURL != baseURI && !resourceURL.match(EMPTY_URL);
  1172. }
  1173. static testValidURL(resourceURL, baseURI) {
  1174. return DomUtil.testValidPath(resourceURL, baseURI) && resourceURL.match(HTTP_URI_PREFIX) && resourceURL.match(NOT_EMPTY_URL);
  1175. }
  1176. static matchImport(stylesheetContent) {
  1177. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1178. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1179. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1180. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1181. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1182. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1183. if (match) {
  1184. const [, resourceURL, media] = match;
  1185. return { resourceURL, media };
  1186. }
  1187. }
  1188. static removeCssComments(stylesheetContent) {
  1189. let start, end;
  1190. do {
  1191. start = stylesheetContent.indexOf("/*");
  1192. end = stylesheetContent.indexOf("*/", start + 2);
  1193. if (start != -1 && end != -1) {
  1194. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1195. }
  1196. } while (start != -1 && end != -1);
  1197. return stylesheetContent;
  1198. }
  1199. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1200. if (mediaQuery) {
  1201. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1202. } else {
  1203. return stylesheetContent;
  1204. }
  1205. }
  1206. static insertVariable(doc, indexResource, content, options) {
  1207. const sheetAttributeName = DOM.sheetAttributeName(options.sessionId);
  1208. let styleElement = doc.querySelector("style[" + sheetAttributeName + "]"), insertedVariables;
  1209. if (!styleElement) {
  1210. styleElement = doc.createElement("style");
  1211. if (doc.head.firstChild) {
  1212. doc.head.insertBefore(styleElement, doc.head.firstChild);
  1213. } else {
  1214. doc.head.appendChild(styleElement);
  1215. }
  1216. styleElement.setAttribute(sheetAttributeName, "[]");
  1217. insertedVariables = [];
  1218. } else {
  1219. insertedVariables = JSON.parse(styleElement.getAttribute(sheetAttributeName));
  1220. }
  1221. if (!insertedVariables.includes(indexResource)) {
  1222. insertedVariables.push(indexResource);
  1223. styleElement.textContent = styleElement.textContent + `:root{${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")}`;
  1224. styleElement.setAttribute(sheetAttributeName, JSON.stringify(insertedVariables));
  1225. }
  1226. }
  1227. static replaceImageSource(imgElement, variableName, options) {
  1228. const dataAttributeName = DOM.imagesAttributeName(options.sessionId);
  1229. if (imgElement.getAttribute(dataAttributeName) != null) {
  1230. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1231. imgElement.setAttribute("src", `data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="${imgData.pxWidth}" height="${imgData.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1232. DomProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")");
  1233. imgElement.removeAttribute(dataAttributeName);
  1234. return true;
  1235. }
  1236. }
  1237. }
  1238. function log(...args) {
  1239. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1240. }
  1241. // -----
  1242. // Stats
  1243. // -----
  1244. const STATS_DEFAULT_VALUES = {
  1245. discarded: {
  1246. "HTML bytes": 0,
  1247. "hidden elements": 0,
  1248. "HTML imports": 0,
  1249. scripts: 0,
  1250. objects: 0,
  1251. "audio sources": 0,
  1252. "video sources": 0,
  1253. frames: 0,
  1254. "CSS rules": 0,
  1255. canvas: 0,
  1256. stylesheets: 0,
  1257. resources: 0,
  1258. medias: 0
  1259. },
  1260. processed: {
  1261. "HTML bytes": 0,
  1262. "hidden elements": 0,
  1263. "HTML imports": 0,
  1264. scripts: 0,
  1265. objects: 0,
  1266. "audio sources": 0,
  1267. "video sources": 0,
  1268. frames: 0,
  1269. "CSS rules": 0,
  1270. canvas: 0,
  1271. stylesheets: 0,
  1272. resources: 0,
  1273. medias: 0
  1274. }
  1275. };
  1276. class Stats {
  1277. constructor(options) {
  1278. this.options = options;
  1279. if (options.displayStats) {
  1280. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1281. }
  1282. }
  1283. set(type, subType, value) {
  1284. if (this.options.displayStats) {
  1285. this.data[type][subType] = value;
  1286. }
  1287. }
  1288. add(type, subType, value) {
  1289. if (this.options.displayStats) {
  1290. this.data[type][subType] += value;
  1291. }
  1292. }
  1293. addAll(pageData) {
  1294. if (this.options.displayStats) {
  1295. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1296. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1297. }
  1298. }
  1299. }
  1300. return { getClass };
  1301. })();