single-file-core.js 62 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480
  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 SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. if (options.sessionId === undefined) {
  34. options.sessionId = sessionId;
  35. sessionId++;
  36. }
  37. }
  38. async initialize() {
  39. this.processor = new PageProcessor(this.options);
  40. await this.processor.loadPage();
  41. await this.processor.initialize();
  42. }
  43. async run() {
  44. await this.processor.run();
  45. }
  46. async getPageData() {
  47. return this.processor.getPageData();
  48. }
  49. }
  50. SingleFileClass.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  51. SingleFileClass.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  52. // -------------
  53. // ProgressEvent
  54. // -------------
  55. const PAGE_LOADING = "page-loading";
  56. const PAGE_LOADED = "page-loaded";
  57. const RESOURCES_INITIALIZING = "resource-initializing";
  58. const RESOURCES_INITIALIZED = "resources-initialized";
  59. const RESOURCE_LOADED = "resource-loaded";
  60. const PAGE_ENDED = "page-ended";
  61. class ProgressEvent {
  62. constructor(type, details) {
  63. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  64. }
  65. }
  66. // -------------
  67. // PageProcessor
  68. // -------------
  69. const RESOLVE_URLS_STAGE = 0;
  70. const REPLACE_DATA_STAGE = 1;
  71. const REPLACE_DOCS_STAGE = 2;
  72. const CLEANUP_STAGE = 3;
  73. const STAGES = [{
  74. sequential: [
  75. { action: "preProcessPage" },
  76. { action: "replaceStyleContents" },
  77. { option: "selected", action: "removeUnselectedElements" },
  78. { option: "removeVideoSrc", action: "insertVideoPosters" },
  79. { option: "removeFrames", action: "removeFrames" },
  80. { option: "removeImports", action: "removeImports" },
  81. { option: "removeScripts", action: "removeScripts" },
  82. { action: "removeDiscardedResources" },
  83. { action: "resetCharsetMeta" },
  84. { action: "setInputValues" },
  85. { option: "insertFaviconLink", action: "insertFaviconLink" },
  86. { action: "resolveHrefs" },
  87. { action: "replaceCanvasElements" },
  88. { action: "insertFonts" },
  89. { option: "removeHiddenElements", action: "removeHiddenElements" },
  90. { action: "resolveStyleAttributeURLs" }
  91. ],
  92. parallel: [
  93. { action: "resolveStylesheetURLs" },
  94. { action: "resolveLinkedStylesheetURLs" },
  95. { option: "!removeFrames", action: "resolveFrameURLs" },
  96. { option: "!removeImports", action: "resolveHtmlImportURLs" }
  97. ]
  98. }, {
  99. sequential: [
  100. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  101. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  102. { option: "removeUnusedStyles", action: "removeUnusedFonts" }
  103. ],
  104. parallel: [
  105. { action: "processStylesheets" },
  106. { action: "processStyleAttributes" },
  107. { action: "processPageResources" },
  108. { option: "!removeScripts", action: "processScripts" }
  109. ]
  110. }, {
  111. sequential: [
  112. { option: "removeAlternativeImages", action: "removeAlternativeImages" },
  113. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  114. { option: "compressCSS", action: "compressCSS" }
  115. ],
  116. parallel: [
  117. { option: "!removeFrames", action: "processFrames" },
  118. { option: "!removeImports", action: "processHtmlImports" },
  119. ]
  120. }, {
  121. sequential: [
  122. { option: "compressHTML", action: "compressHTML" },
  123. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  124. { action: "cleanup" }
  125. ]
  126. }];
  127. class PageProcessor {
  128. constructor(options) {
  129. this.options = options;
  130. this.options.url = this.options.url || this.options.doc.location.href;
  131. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  132. this.batchRequest = new BatchRequest();
  133. this.processor = new DOMProcessor(options, this.batchRequest);
  134. if (this.options.doc) {
  135. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  136. this.options.canvasData = docData.canvasData;
  137. this.options.fontsData = docData.fontsData;
  138. this.options.stylesheetContents = docData.stylesheetContents;
  139. this.options.imageData = docData.imageData;
  140. this.options.postersData = docData.postersData;
  141. this.options.usedFonts = docData.usedFonts;
  142. }
  143. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  144. this.onprogress = options.onprogress || (() => { });
  145. }
  146. async loadPage() {
  147. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  148. await this.processor.loadPage(this.options.content);
  149. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  150. }
  151. async initialize() {
  152. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  153. await this.executeStage(RESOLVE_URLS_STAGE);
  154. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  155. if (this.options.doc) {
  156. DOM.postProcessDoc(this.options.doc, this.options);
  157. this.options.doc = null;
  158. this.options.win = null;
  159. }
  160. }
  161. async run() {
  162. if (!this.options.windowId) {
  163. this.processor.initialize(this.batchRequest);
  164. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  165. }
  166. await this.batchRequest.run(details => {
  167. details.pageURL = this.options.url;
  168. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  169. }, this.options);
  170. await this.pendingPromises;
  171. await this.executeStage(REPLACE_DOCS_STAGE);
  172. await this.executeStage(CLEANUP_STAGE);
  173. }
  174. async getPageData() {
  175. if (!this.options.windowId) {
  176. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  177. }
  178. return this.processor.getPageData();
  179. }
  180. async executeStage(step) {
  181. if (DEBUG) {
  182. log("**** STARTED STAGE", step, "****");
  183. }
  184. const sequentialPromises = [];
  185. STAGES[step].sequential.forEach(task => {
  186. let startTime;
  187. if (DEBUG) {
  188. startTime = Date.now();
  189. log(" -- STARTED task =", task.action);
  190. }
  191. sequentialPromises.push(this.executeTask(task));
  192. if (DEBUG) {
  193. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  194. }
  195. });
  196. await Promise.all(sequentialPromises);
  197. if (STAGES[step].parallel) {
  198. return await Promise.all(STAGES[step].parallel.map(task => {
  199. let startTime;
  200. if (DEBUG) {
  201. startTime = Date.now();
  202. log(" // STARTED task =", task.action);
  203. }
  204. const promise = this.executeTask(task);
  205. if (DEBUG) {
  206. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  207. }
  208. return promise;
  209. }));
  210. }
  211. if (DEBUG) {
  212. log("**** ENDED STAGE", step, "****");
  213. }
  214. }
  215. executeTask(task) {
  216. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  217. return this.processor[task.action]();
  218. }
  219. }
  220. }
  221. // --------
  222. // BatchRequest
  223. // --------
  224. class BatchRequest {
  225. constructor() {
  226. this.requests = new Map();
  227. }
  228. async addURL(resourceURL, asDataURI = true) {
  229. return new Promise((resolve, reject) => {
  230. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  231. const resourceRequests = this.requests.get(requestKey);
  232. if (resourceRequests) {
  233. resourceRequests.push({ resolve, reject });
  234. } else {
  235. this.requests.set(requestKey, [{ resolve, reject }]);
  236. }
  237. });
  238. }
  239. getMaxResources() {
  240. return Array.from(this.requests.keys()).length;
  241. }
  242. async run(onloadListener, options) {
  243. const resourceURLs = Array.from(this.requests.keys());
  244. let indexResource = 0;
  245. return Promise.all(resourceURLs.map(async requestKey => {
  246. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  247. const resourceRequests = this.requests.get(requestKey);
  248. try {
  249. const resourceContent = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  250. indexResource = indexResource + 1;
  251. onloadListener({ index: indexResource, url: resourceURL });
  252. resourceRequests.forEach(resourceRequest => resourceRequest.resolve({ content: resourceContent, indexResource, duplicate: Boolean(resourceRequests.length > 1) }));
  253. } catch (error) {
  254. indexResource = indexResource + 1;
  255. onloadListener({ index: indexResource, url: resourceURL });
  256. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  257. }
  258. this.requests.delete(requestKey);
  259. }));
  260. }
  261. }
  262. // ------------
  263. // DOMProcessor
  264. // ------------
  265. const EMPTY_DATA_URI = "data:base64,";
  266. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  267. const SCRIPT_TAG_FOUND = /<script/gi;
  268. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  269. class DOMProcessor {
  270. constructor(options, batchRequest) {
  271. this.options = options;
  272. this.stats = new Stats(options);
  273. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  274. this.batchRequest = batchRequest;
  275. }
  276. initialize() {
  277. this.maxResources = this.batchRequest.getMaxResources();
  278. if (!this.options.removeFrames) {
  279. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  280. }
  281. this.stats.set("processed", "resources", this.maxResources);
  282. }
  283. async loadPage(pageContent) {
  284. if (!pageContent || this.options.saveRawPage) {
  285. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  286. }
  287. this.doc = DOM.createDoc(pageContent, this.baseURI);
  288. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  289. }
  290. async getPageData() {
  291. DOM.postProcessDoc(this.doc, this.options);
  292. const titleElement = this.doc.querySelector("title");
  293. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  294. this.options.info = {};
  295. const descriptionElement = this.doc.querySelector("meta[name=description]");
  296. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  297. this.options.info.lang = this.doc.documentElement.lang;
  298. const authorElement = this.doc.querySelector("meta[name=author]");
  299. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  300. const creatorElement = this.doc.querySelector("meta[name=creator]");
  301. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  302. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  303. this.options.info.publisher = publisherElement ? publisherElement.content.trim() : "";
  304. const url = new URL(this.baseURI);
  305. let size;
  306. if (this.options.displayStats) {
  307. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  308. }
  309. const content = DOM.serialize(this.doc, this.options.compressHTML);
  310. if (this.options.displayStats) {
  311. const contentSize = DOM.getContentSize(content);
  312. this.stats.set("processed", "HTML bytes", contentSize);
  313. this.stats.add("discarded", "HTML bytes", size - contentSize);
  314. }
  315. const filename = await DomProcessorHelper.getFilename(this.options, content);
  316. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  317. return {
  318. stats: this.stats.data,
  319. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  320. filename,
  321. content
  322. };
  323. }
  324. removeUnselectedElements() {
  325. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  326. if (rootElement) {
  327. DomProcessorHelper.isolateElements(rootElement);
  328. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  329. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  330. }
  331. }
  332. setInputValues() {
  333. this.doc.querySelectorAll("input").forEach(input => {
  334. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  335. input.setAttribute("value", value || "");
  336. });
  337. this.doc.querySelectorAll("textarea").forEach(textarea => {
  338. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  339. textarea.textContent = value || "";
  340. });
  341. this.doc.querySelectorAll("select").forEach(select => {
  342. select.querySelectorAll("option").forEach(option => {
  343. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  344. if (selected) {
  345. option.setAttribute("selected", "");
  346. }
  347. });
  348. });
  349. }
  350. removeDiscardedResources() {
  351. const objectElements = this.doc.querySelectorAll("applet, meta[http-equiv=refresh], object[data]:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed[src]:not([src*=\".svg\"])");
  352. this.stats.set("discarded", "objects", objectElements.length);
  353. this.stats.set("processed", "objects", objectElements.length);
  354. objectElements.forEach(element => element.remove());
  355. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  356. replacedAttributeValue.forEach(element => {
  357. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  358. if (relValue.length) {
  359. element.setAttribute("rel", relValue);
  360. } else {
  361. element.remove();
  362. }
  363. });
  364. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  365. if (this.options.compressHTML) {
  366. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  367. }
  368. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  369. if (this.options.removeScripts) {
  370. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  371. this.doc.querySelectorAll("[href]").forEach(element => {
  372. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  373. element.setAttribute("href", "");
  374. }
  375. });
  376. this.doc.querySelectorAll("[src]").forEach(element => {
  377. if (element.src && element.src.match(/^\s*javascript:/)) {
  378. element.removeAttribute("src");
  379. }
  380. });
  381. }
  382. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  383. this.stats.set("processed", "audio sources", audioSourceElements.length);
  384. if (this.options.removeAudioSrc) {
  385. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  386. audioSourceElements.forEach(element => element.removeAttribute("src"));
  387. }
  388. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  389. this.stats.set("processed", "video sources", videoSourceElements.length);
  390. if (this.options.removeVideoSrc) {
  391. this.stats.set("discarded", "video sources", videoSourceElements.length);
  392. videoSourceElements.forEach(element => element.removeAttribute("src"));
  393. }
  394. }
  395. cleanup() {
  396. const metaCharset = this.doc.head.querySelector("meta[charset]");
  397. if (metaCharset) {
  398. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  399. }
  400. this.doc.querySelectorAll("style[data-single-file-sheet]").forEach(element => element.removeAttribute("data-single-file-sheet"));
  401. this.doc.querySelectorAll("base").forEach(element => element.remove());
  402. if (this.doc.head.querySelectorAll("*").length == 1 && metaCharset && this.doc.body.childNodes.length == 0) {
  403. this.doc.head.querySelector("meta[charset]").remove();
  404. }
  405. }
  406. preProcessPage() {
  407. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  408. if (this.options.win) {
  409. this.doc.body.querySelectorAll(":not(svg) title, meta").forEach(element => element instanceof this.options.win.HTMLElement && this.doc.head.appendChild(element));
  410. }
  411. if (this.options.imageData) {
  412. const dataAttributeName = DOM.imagesAttributeName(this.options.sessionId);
  413. this.doc.querySelectorAll("img").forEach(imgElement => {
  414. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  415. if (imgData.currentSrc) {
  416. imgElement.setAttribute("src", imgData.currentSrc);
  417. }
  418. });
  419. if (this.options.lazyLoadImages) {
  420. this.doc.querySelectorAll("img[data-src]").forEach(imgElement => {
  421. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  422. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  423. imgData.src = imgElement.dataset.src;
  424. imgElement.setAttribute("src", imgElement.dataset.src);
  425. imgElement.removeAttribute("data-src");
  426. }
  427. });
  428. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  429. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  430. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  431. imgElement.removeAttribute("data-srcset");
  432. }
  433. });
  434. }
  435. }
  436. }
  437. removeScripts() {
  438. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  439. this.stats.set("discarded", "scripts", scriptElements.length);
  440. this.stats.set("processed", "scripts", scriptElements.length);
  441. scriptElements.forEach(element => element.remove());
  442. }
  443. removeFrames() {
  444. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  445. this.stats.set("discarded", "frames", frameElements.length);
  446. this.stats.set("processed", "frames", frameElements.length);
  447. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  448. }
  449. removeImports() {
  450. const importElements = this.doc.querySelectorAll("link[rel=import]");
  451. this.stats.set("discarded", "HTML imports", importElements.length);
  452. this.stats.set("processed", "HTML imports", importElements.length);
  453. importElements.forEach(element => element.remove());
  454. }
  455. resetCharsetMeta() {
  456. let charSet;
  457. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  458. const charSetDeclaration = element.content.split(";")[1];
  459. if (charSetDeclaration && !charSet) {
  460. charSet = charSetDeclaration.split("=")[1];
  461. if (charSet) {
  462. this.charSet = charSet.trim().toLowerCase();
  463. }
  464. }
  465. element.remove();
  466. });
  467. const metaElement = this.doc.createElement("meta");
  468. metaElement.setAttribute("charset", "utf-8");
  469. if (this.doc.head.firstChild) {
  470. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  471. } else {
  472. this.doc.head.appendChild(metaElement);
  473. }
  474. }
  475. insertFaviconLink() {
  476. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  477. if (!faviconElement) {
  478. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  479. }
  480. if (!faviconElement) {
  481. faviconElement = this.doc.createElement("link");
  482. faviconElement.setAttribute("type", "image/x-icon");
  483. faviconElement.setAttribute("rel", "shortcut icon");
  484. faviconElement.setAttribute("href", "/favicon.ico");
  485. }
  486. this.doc.head.appendChild(faviconElement);
  487. }
  488. resolveHrefs() {
  489. this.doc.querySelectorAll("[href]").forEach(element => {
  490. if (element.href) {
  491. const href = element.href.baseVal ? element.href.baseVal : element.href;
  492. const normalizedHref = DomUtil.normalizeURL(href);
  493. if (normalizedHref == href && href) {
  494. element.setAttribute("href", href);
  495. }
  496. }
  497. });
  498. }
  499. removeUnusedStyles() {
  500. if (!this.mediaAllInfo) {
  501. this.mediaAllInfo = DOM.getMediaAllInfo(this.doc);
  502. }
  503. const stats = DOM.minifyCSS(this.doc, this.mediaAllInfo);
  504. this.stats.set("processed", "CSS rules", stats.processed);
  505. this.stats.set("discarded", "CSS rules", stats.discarded);
  506. }
  507. removeUnusedFonts() {
  508. DOM.removeUnusedFonts(this.doc, this.options);
  509. }
  510. removeAlternativeFonts() {
  511. DOM.removeAlternativeFonts(this.doc);
  512. }
  513. removeAlternativeImages() {
  514. DOM.removeAlternativeImages(this.doc);
  515. }
  516. removeHiddenElements() {
  517. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  518. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  519. this.stats.set("processed", "hidden elements", hiddenElements.length);
  520. hiddenElements.forEach(element => element.remove());
  521. }
  522. compressHTML() {
  523. let size;
  524. if (this.options.displayStats) {
  525. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  526. }
  527. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  528. if (this.options.displayStats) {
  529. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  530. }
  531. }
  532. removeAlternativeMedias() {
  533. const stats = DOM.minifyMedias(this.doc);
  534. this.stats.set("processed", "medias", stats.processed);
  535. this.stats.set("discarded", "medias", stats.discarded);
  536. }
  537. compressCSS() {
  538. this.doc.querySelectorAll("style").forEach(styleElement => {
  539. if (styleElement) {
  540. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  541. }
  542. });
  543. this.doc.querySelectorAll("[style]").forEach(element => {
  544. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  545. });
  546. }
  547. insertSingleFileComment() {
  548. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.options.url + " \n saved date: " + new Date() + " \n");
  549. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  550. }
  551. replaceCanvasElements() {
  552. if (this.options.canvasData) {
  553. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  554. const canvasData = this.options.canvasData[indexCanvasElement];
  555. if (canvasData) {
  556. DomProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  557. this.stats.add("processed", "canvas", 1);
  558. }
  559. });
  560. }
  561. }
  562. insertFonts() {
  563. if (this.options.fontsData && this.options.fontsData.length) {
  564. let stylesheetContent = "";
  565. this.options.fontsData.forEach(fontStyles => {
  566. if (fontStyles["font-family"] && fontStyles.src) {
  567. stylesheetContent += "@font-face{";
  568. let stylesContent = "";
  569. Object.keys(fontStyles).forEach(fontStyle => {
  570. if (stylesContent) {
  571. stylesContent += ";";
  572. }
  573. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  574. });
  575. stylesheetContent += stylesContent + "}";
  576. }
  577. });
  578. if (stylesheetContent) {
  579. const styleElement = this.doc.createElement("style");
  580. styleElement.textContent = stylesheetContent;
  581. const existingStyleElement = this.doc.querySelector("style");
  582. if (existingStyleElement) {
  583. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  584. } else {
  585. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  586. }
  587. }
  588. }
  589. }
  590. replaceStyleContents() {
  591. if (this.options.stylesheetContents) {
  592. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  593. if (this.options.stylesheetContents[styleIndex]) {
  594. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  595. }
  596. });
  597. }
  598. }
  599. insertVideoPosters() {
  600. if (this.options.postersData) {
  601. this.doc.querySelectorAll("video[src], video > source[src]").forEach((element, videoIndex) => {
  602. let videoElement;
  603. if (element.tagName == "VIDEO") {
  604. videoElement = element;
  605. } else {
  606. videoElement = element.parentElement;
  607. }
  608. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  609. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  610. }
  611. });
  612. }
  613. }
  614. async processPageResources() {
  615. const resourcePromises = [
  616. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", "data:", this.baseURI, this.options, this.batchRequest, 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.options, this.batchRequest),
  618. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("img[src], input[src][type=image]"), "src", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest, true),
  619. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.options, this.batchRequest),
  620. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  621. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  622. DomProcessorHelper.processAttribute(this.doc, this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, this.batchRequest),
  623. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  624. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.options, 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.options, this.batchRequest));
  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.options, this.batchRequest));
  631. }
  632. await resourcePromises;
  633. if (this.options.removeAlternativeImages) {
  634. let shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  635. if (!shortcutIcon) {
  636. shortcutIcon = findShortcutIcon(Array.from(this.doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  637. if (shortcutIcon) {
  638. shortcutIcon.rel = "icon";
  639. }
  640. }
  641. if (shortcutIcon) {
  642. this.doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  643. if (linkElement != shortcutIcon) {
  644. linkElement.remove();
  645. }
  646. });
  647. }
  648. }
  649. function findShortcutIcon(shortcutIcons) {
  650. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  651. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  652. return shortcutIcons[0];
  653. }
  654. }
  655. async resolveStylesheetURLs() {
  656. await Promise.all(Array.from(this.doc.querySelectorAll("style"))
  657. .map(async styleElement => {
  658. const options = { url: this.options.url, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled };
  659. const textContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, options);
  660. styleElement.textContent = textContent;
  661. }));
  662. }
  663. async processStylesheets() {
  664. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async styleElement => {
  665. this.stats.add("processed", "stylesheets", 1);
  666. if (styleElement.sheet) {
  667. styleElement.textContent = await DomProcessorHelper.processStylesheet(this.doc, styleElement.textContent, styleElement.sheet.cssRules, this.baseURI, this.options, this.batchRequest);
  668. } else {
  669. styleElement.remove();
  670. }
  671. }));
  672. }
  673. async processScripts() {
  674. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  675. if (scriptElement.src) {
  676. this.stats.add("processed", "scripts", 1);
  677. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  678. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  679. }
  680. scriptElement.removeAttribute("src");
  681. }));
  682. }
  683. async resolveFrameURLs() {
  684. if (this.options.framesData) {
  685. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  686. await Promise.all(frameElements.map(async frameElement => {
  687. DomProcessorHelper.setFrameEmptySrc(frameElement);
  688. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  689. if (frameWindowId) {
  690. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  691. if (frameData) {
  692. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  693. }
  694. }
  695. }));
  696. }
  697. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  698. options.insertSingleFileComment = false;
  699. options.insertFaviconLink = false;
  700. options.doc = null;
  701. options.win = null;
  702. options.url = frameData.baseURI;
  703. options.windowId = frameWindowId;
  704. if (frameData.content) {
  705. options.content = frameData.content;
  706. options.canvasData = frameData.canvasData;
  707. options.stylesheetContents = frameData.stylesheetContents;
  708. options.fontsData = frameData.fontsData;
  709. options.imageData = frameData.imageData;
  710. options.usedFonts = frameData.usedFonts;
  711. frameData.processor = new PageProcessor(options);
  712. frameData.frameElement = frameElement;
  713. await frameData.processor.loadPage();
  714. await frameData.processor.initialize();
  715. frameData.maxResources = batchRequest.getMaxResources();
  716. }
  717. }
  718. }
  719. async processFrames() {
  720. if (this.options.framesData) {
  721. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  722. await Promise.all(frameElements.map(async frameElement => {
  723. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  724. if (frameWindowId) {
  725. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  726. if (frameData) {
  727. if (frameData.processor) {
  728. this.stats.add("processed", "frames", 1);
  729. await frameData.processor.run();
  730. const pageData = await frameData.processor.getPageData();
  731. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  732. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  733. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  734. } else {
  735. frameElement.setAttribute("sandbox", "");
  736. }
  737. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  738. this.stats.addAll(pageData);
  739. } else {
  740. this.stats.add("discarded", "frames", 1);
  741. }
  742. }
  743. }
  744. }));
  745. }
  746. }
  747. async resolveHtmlImportURLs() {
  748. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  749. if (!this.relImportProcessors) {
  750. this.relImportProcessors = new Map();
  751. }
  752. await Promise.all(linkElements.map(async linkElement => {
  753. const resourceURL = linkElement.href;
  754. linkElement.removeAttribute("href");
  755. const options = Object.create(this.options);
  756. options.insertSingleFileComment = false;
  757. options.insertFaviconLink = false;
  758. options.doc = null;
  759. options.win = null;
  760. options.url = resourceURL;
  761. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, this.baseURI, this.options.url)) {
  762. const processor = new PageProcessor(options);
  763. this.relImportProcessors.set(linkElement, processor);
  764. await processor.loadPage();
  765. return processor.initialize();
  766. }
  767. }));
  768. }
  769. async processHtmlImports() {
  770. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  771. await Promise.all(linkElements.map(async linkElement => {
  772. const processor = this.relImportProcessors.get(linkElement);
  773. if (processor) {
  774. this.stats.add("processed", "HTML imports", 1);
  775. this.relImportProcessors.delete(linkElement);
  776. const pageData = await processor.getPageData();
  777. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  778. this.stats.addAll(pageData);
  779. } else {
  780. this.stats.add("discarded", "HTML imports", 1);
  781. }
  782. }));
  783. }
  784. resolveStyleAttributeURLs() {
  785. Array.from(this.doc.querySelectorAll("[style]"))
  786. .map(element => element.setAttribute("style", DomProcessorHelper.resolveStylesheetURLs(element.getAttribute("style"), this.baseURI, this.options)));
  787. }
  788. async processStyleAttributes() {
  789. await Promise.all(Array.from(this.doc.querySelectorAll("[style]"))
  790. .map(async element => {
  791. const rules = [{ type: CSSRule.STYLE_RULE, cssText: element.getAttribute("style") }];
  792. const textContent = await DomProcessorHelper.processStylesheet(this.doc, element.getAttribute("style"), rules, this.baseURI, this.options, this.batchRequest);
  793. element.setAttribute("style", textContent);
  794. }));
  795. }
  796. async resolveLinkedStylesheetURLs() {
  797. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  798. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charSet: this.charSet };
  799. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, options);
  800. if (stylesheetContent) {
  801. const styleElement = this.doc.createElement("style");
  802. if (linkElement.media) {
  803. styleElement.media = linkElement.media;
  804. }
  805. styleElement.textContent = stylesheetContent;
  806. linkElement.parentElement.replaceChild(styleElement, linkElement);
  807. } else {
  808. linkElement.remove();
  809. }
  810. }));
  811. }
  812. }
  813. // ---------
  814. // DomHelper
  815. // ---------
  816. const REGEXP_AMP = /&/g;
  817. const REGEXP_NBSP = /\u00a0/g;
  818. const REGEXP_START_TAG = /</g;
  819. const REGEXP_END_TAG = />/g;
  820. const REGEXP_URL_HASH = /(#.+?)$/;
  821. const PREFIX_DATA_URI_IMAGE = "data:image/";
  822. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  823. const PREFIX_DATA_URI_VIDEO = "data:video/";
  824. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  825. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  826. const PREFIX_DATA_URI_OCTET_STREAM = /^data:(application|binary)\/octet-stream/;
  827. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  828. class DomProcessorHelper {
  829. static async getFilename(options, content) {
  830. let filename = options.filenameTemplate;
  831. const date = new Date();
  832. const url = new URL(options.url);
  833. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  834. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  835. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  836. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  837. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  838. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  839. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  840. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  841. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  842. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  843. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  844. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()).padStart(2, "0"));
  845. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"));
  846. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  847. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  848. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  849. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"));
  850. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"));
  851. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  852. filename = await DomUtil.evalTemplateVariable(filename, "hours-locale", () => String(date.getHours()).padStart(2, "0"));
  853. filename = await DomUtil.evalTemplateVariable(filename, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"));
  854. filename = await DomUtil.evalTemplateVariable(filename, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"));
  855. filename = await DomUtil.evalTemplateVariable(filename, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"));
  856. filename = await DomUtil.evalTemplateVariable(filename, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"));
  857. filename = await DomUtil.evalTemplateVariable(filename, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"));
  858. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  859. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  860. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  861. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  862. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  863. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  864. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  865. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  866. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  867. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  868. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  869. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  870. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  871. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  872. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  873. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  874. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  875. if (!options.backgroundSave) {
  876. filename = filename.replace(/\//g, "_");
  877. }
  878. if (filename.length > 192) {
  879. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  880. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  881. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  882. }
  883. if (!filename) {
  884. filename = "Unnamed page";
  885. }
  886. return filename;
  887. }
  888. static setBackgroundImage(element, url, style) {
  889. element.style.setProperty("background-blend-mode", "normal", "important");
  890. element.style.setProperty("background-clip", "content-box", "important");
  891. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  892. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  893. element.style.setProperty("background-image", url, "important");
  894. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  895. element.style.setProperty("background-origin", "content-box", "important");
  896. element.style.setProperty("background-repeat", "no-repeat", "important");
  897. }
  898. static setFrameEmptySrc(frameElement) {
  899. if (frameElement.tagName == "OBJECT") {
  900. frameElement.setAttribute("data", "data:text/html,");
  901. } else {
  902. frameElement.setAttribute("srcdoc", "");
  903. frameElement.removeAttribute("src");
  904. }
  905. }
  906. static setFrameContent(frameElement, content) {
  907. if (frameElement.tagName == "OBJECT") {
  908. frameElement.setAttribute("data", "data:text/html," + content);
  909. } else {
  910. frameElement.setAttribute("srcdoc", content);
  911. frameElement.removeAttribute("src");
  912. }
  913. }
  914. static isolateElements(rootElement) {
  915. rootElement.querySelectorAll("*:not(style)").forEach(element => {
  916. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  917. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  918. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  919. element.remove();
  920. }
  921. });
  922. isolateParentElements(rootElement.parentElement, rootElement);
  923. function isolateParentElements(parentElement, element) {
  924. let elementFound = false;
  925. Array.from(parentElement.childNodes).forEach(node => {
  926. elementFound = elementFound || (node == element);
  927. const tagName = node.tagName && node.tagName.toLowerCase();
  928. if (node != element && tagName != "svg" && tagName != "style" && tagName != "link") {
  929. if (elementFound) {
  930. node.remove();
  931. } else {
  932. node.hidden = true;
  933. if (node.childNodes && node.childNodes.length) {
  934. Array.from(node.childNodes).forEach(node => node.remove());
  935. }
  936. }
  937. }
  938. });
  939. element = element.parentElement;
  940. if (element && element.parentElement && element.parentElement.tagName != "HTML") {
  941. isolateParentElements(element.parentElement, element);
  942. }
  943. }
  944. }
  945. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  946. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, options);
  947. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  948. const imports = DomUtil.getImportFunctions(stylesheetContent);
  949. await Promise.all(imports.map(async cssImport => {
  950. const match = DomUtil.matchImport(cssImport);
  951. if (match) {
  952. let resourceURL = DomUtil.normalizeURL(match.resourceURL);
  953. if (!DomUtil.testIgnoredPath(resourceURL) && DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  954. resourceURL = new URL(match.resourceURL, baseURI).href;
  955. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  956. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  957. let importedStylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  958. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  959. if (stylesheetContent.includes(cssImport)) {
  960. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options);
  961. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  962. }
  963. }
  964. }
  965. }
  966. }));
  967. return stylesheetContent;
  968. }
  969. static resolveStylesheetURLs(stylesheetContent, baseURI, options) {
  970. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  971. urlFunctions.map(urlFunction => {
  972. const originalResourceURL = DomUtil.matchURL(urlFunction);
  973. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  974. if (!DomUtil.testIgnoredPath(resourceURL)) {
  975. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  976. const resolvedURL = new URL(resourceURL, baseURI).href;
  977. if (DomUtil.testValidURL(resolvedURL, baseURI, options.url) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  978. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  979. }
  980. }
  981. } else {
  982. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  983. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  984. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  985. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  986. }
  987. }
  988. }
  989. });
  990. return stylesheetContent;
  991. }
  992. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options) {
  993. resourceURL = DomUtil.normalizeURL(resourceURL);
  994. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  995. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charSet: options.charSet };
  996. let stylesheetContent = await Download.getContent(resourceURL, downloadOptions);
  997. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  998. return stylesheetContent;
  999. }
  1000. }
  1001. static async processStylesheet(doc, stylesheetContent, cssRules, baseURI, options, batchRequest) {
  1002. const urlFunctions = getURLFunctions(cssRules);
  1003. const styleResources = Promise.all(urlFunctions.style.map(urlFunction => processURLFunction(urlFunction)));
  1004. const otherResources = Promise.all(urlFunctions.other.map(urlFunction => processURLFunction(urlFunction, true)));
  1005. await Promise.all([styleResources, otherResources]);
  1006. return stylesheetContent;
  1007. async function processURLFunction(urlFunction, processFont) {
  1008. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1009. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1010. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1011. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1012. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1013. if (processFont) {
  1014. const validFont = await DOM.validFont(content);
  1015. if (!validFont) {
  1016. content = EMPTY_DATA_URI;
  1017. }
  1018. }
  1019. let regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1020. if (!stylesheetContent.match(regExpUrlFunction)) {
  1021. urlFunction = "url(" + originalResourceURL + ")";
  1022. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1023. }
  1024. if (!stylesheetContent.match(regExpUrlFunction)) {
  1025. urlFunction = "url('" + originalResourceURL + "')";
  1026. regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  1027. }
  1028. if (stylesheetContent.match(regExpUrlFunction)) {
  1029. if (duplicate && options.groupDuplicateImages && !processFont) {
  1030. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")");
  1031. DomUtil.insertVariable(doc, indexResource, content, options);
  1032. } else {
  1033. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, content));
  1034. }
  1035. }
  1036. }
  1037. }
  1038. }
  1039. function getURLFunctions(cssRules, urlFunctions = { style: [], other: [] }) {
  1040. Array.from(cssRules).forEach(cssRule => {
  1041. if (cssRule.type == CSSRule.MEDIA_RULE) {
  1042. getURLFunctions(cssRule.cssRules, urlFunctions);
  1043. } else if (cssRule.type == CSSRule.STYLE_RULE) {
  1044. getURLFunctionsRule(cssRule.cssText, urlFunctions.style);
  1045. } else {
  1046. getURLFunctionsRule(cssRule.cssText, urlFunctions.other);
  1047. }
  1048. });
  1049. return { style: Array.from(urlFunctions.style), other: Array.from(urlFunctions.other) };
  1050. }
  1051. function getURLFunctionsRule(cssText, urlFunctions) {
  1052. const urlFunctionsRule = DomUtil.getUrlFunctions(cssText);
  1053. urlFunctionsRule.forEach(urlFunction => {
  1054. const originalResourceURL = DomUtil.matchURL(urlFunction);
  1055. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1056. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1057. if (DomUtil.testValidURL(resourceURL, baseURI, options.url) && cssText.includes(urlFunction)) {
  1058. urlFunctions.push(urlFunction);
  1059. }
  1060. }
  1061. });
  1062. return urlFunctions;
  1063. }
  1064. }
  1065. static async processAttribute(doc, resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest, processDuplicates, removeElementIfMissing) {
  1066. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1067. let resourceURL = resourceElement.getAttribute(attributeName);
  1068. resourceURL = DomUtil.normalizeURL(resourceURL);
  1069. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1070. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1071. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1072. resourceURL = new URL(resourceURL, baseURI).href;
  1073. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1074. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL);
  1075. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1076. resourceElement.remove();
  1077. } else {
  1078. if (content.startsWith(prefixDataURI) || content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) || content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1079. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1080. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1081. if (DomUtil.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1082. DomUtil.insertVariable(doc, indexResource, content, options);
  1083. } else {
  1084. resourceElement.setAttribute(attributeName, content);
  1085. }
  1086. } else {
  1087. resourceElement.setAttribute(attributeName, content);
  1088. }
  1089. }
  1090. }
  1091. }
  1092. }
  1093. }
  1094. }));
  1095. }
  1096. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1097. const attributeName = "xlink:href";
  1098. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1099. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1100. let resourceURL = DomUtil.normalizeURL(originalResourceURL);
  1101. if (DomUtil.testValidPath(resourceURL, baseURI, options.url) && !DomUtil.testIgnoredPath(resourceURL)) {
  1102. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1103. resourceURL = new URL(resourceURL, baseURI).href;
  1104. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1105. try {
  1106. const { content } = await batchRequest.addURL(resourceURL, false);
  1107. const DOMParser = DOM.getParser();
  1108. if (DOMParser) {
  1109. let svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  1110. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1111. if (hashMatch && hashMatch[0]) {
  1112. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1113. if (symbolElement) {
  1114. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1115. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1116. }
  1117. } else {
  1118. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1119. }
  1120. } else {
  1121. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1122. }
  1123. } catch (error) {
  1124. /* ignored */
  1125. }
  1126. }
  1127. } else if (resourceURL == options.url) {
  1128. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1129. }
  1130. }));
  1131. }
  1132. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, options, batchRequest) {
  1133. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1134. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1135. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1136. let resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1137. if (!DomUtil.testIgnoredPath(resourceURL)) {
  1138. if (DomUtil.testValidPath(resourceURL, baseURI, options.url)) {
  1139. resourceURL = new URL(resourceURL, baseURI).href;
  1140. if (DomUtil.testValidURL(resourceURL, baseURI, options.url)) {
  1141. const { content } = await batchRequest.addURL(resourceURL);
  1142. if (!content.startsWith(prefixDataURI) && !content.startsWith(PREFIX_DATA_URI_NO_MIMETYPE) && !content.match(PREFIX_DATA_URI_OCTET_STREAM)) {
  1143. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1144. }
  1145. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1146. } else {
  1147. return "";
  1148. }
  1149. } else {
  1150. return "";
  1151. }
  1152. }
  1153. }));
  1154. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1155. }));
  1156. }
  1157. }
  1158. // -------
  1159. // DomUtil
  1160. // -------
  1161. const DATA_URI_PREFIX = "data:";
  1162. const BLOB_URI_PREFIX = "blob:";
  1163. const HTTP_URI_PREFIX = /^https?:\/\//;
  1164. const FILE_URI_PREFIX = /^file:\/\//;
  1165. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1166. const ABOUT_BLANK_URI = "about:blank";
  1167. const NOT_EMPTY_URL = /^(https?|file):\/\/.+/;
  1168. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1169. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1170. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1171. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1172. 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;
  1173. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1174. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1175. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1176. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1177. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1178. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1179. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1180. class DomUtil {
  1181. static normalizeURL(url) {
  1182. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1183. return url;
  1184. } else {
  1185. return url.split("#")[0];
  1186. }
  1187. }
  1188. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1189. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1190. if (template.match(replaceRegExp)) {
  1191. let value = await valueGetter();
  1192. if (!dontReplaceSlash) {
  1193. value = value.replace(/\/+/g, "_");
  1194. }
  1195. return template.replace(replaceRegExp, value);
  1196. }
  1197. return template;
  1198. }
  1199. static getLastSegment(url) {
  1200. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1201. if (!lastSegment) {
  1202. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1203. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1204. }
  1205. if (!lastSegment) {
  1206. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1207. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1208. }
  1209. if (!lastSegment) {
  1210. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1211. }
  1212. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1213. return lastSegment;
  1214. }
  1215. static getRegExp(string) {
  1216. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1217. }
  1218. static getUrlFunctions(stylesheetContent) {
  1219. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1220. }
  1221. static getImportFunctions(stylesheetContent) {
  1222. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1223. }
  1224. static matchURL(stylesheetContent) {
  1225. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1226. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1227. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1228. return match && match[1];
  1229. }
  1230. static testIgnoredPath(resourceURL) {
  1231. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1232. }
  1233. static testValidPath(resourceURL, baseURI, docURL) {
  1234. return resourceURL && resourceURL != baseURI && resourceURL != docURL && !resourceURL.match(EMPTY_URL);
  1235. }
  1236. static testValidURL(resourceURL, baseURI, docURL) {
  1237. return DomUtil.testValidPath(resourceURL, baseURI, docURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1238. }
  1239. static matchImport(stylesheetContent) {
  1240. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1241. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1242. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1243. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1244. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1245. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1246. if (match) {
  1247. const [, resourceURL, media] = match;
  1248. return { resourceURL, media };
  1249. }
  1250. }
  1251. static removeCssComments(stylesheetContent) {
  1252. let start, end;
  1253. do {
  1254. start = stylesheetContent.indexOf("/*");
  1255. end = stylesheetContent.indexOf("*/", start + 2);
  1256. if (start != -1 && end != -1) {
  1257. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1258. }
  1259. } while (start != -1 && end != -1);
  1260. return stylesheetContent;
  1261. }
  1262. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1263. if (mediaQuery) {
  1264. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1265. } else {
  1266. return stylesheetContent;
  1267. }
  1268. }
  1269. static insertVariable(doc, indexResource, content, options) {
  1270. const sheetAttributeName = DOM.sheetAttributeName(options.sessionId);
  1271. let styleElement = doc.querySelector("style[" + sheetAttributeName + "]"), insertedVariables;
  1272. if (!styleElement) {
  1273. styleElement = doc.createElement("style");
  1274. if (doc.head.firstChild) {
  1275. doc.head.insertBefore(styleElement, doc.head.firstChild);
  1276. } else {
  1277. doc.head.appendChild(styleElement);
  1278. }
  1279. styleElement.setAttribute(sheetAttributeName, "[]");
  1280. insertedVariables = [];
  1281. } else {
  1282. insertedVariables = JSON.parse(styleElement.getAttribute(sheetAttributeName));
  1283. }
  1284. if (!insertedVariables.includes(indexResource)) {
  1285. insertedVariables.push(indexResource);
  1286. styleElement.textContent = styleElement.textContent + `:root{${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")}`;
  1287. styleElement.setAttribute(sheetAttributeName, JSON.stringify(insertedVariables));
  1288. }
  1289. }
  1290. static replaceImageSource(imgElement, variableName, options) {
  1291. const dataAttributeName = DOM.imagesAttributeName(options.sessionId);
  1292. if (imgElement.getAttribute(dataAttributeName) != null) {
  1293. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1294. if (imgData.replaceable) {
  1295. imgElement.setAttribute("src", `data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="${imgData.size.pxWidth}" height="${imgData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1296. const backgroundStyle = {};
  1297. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1298. if (backgroundSize) {
  1299. backgroundStyle["background-size"] = imgData.objectFit;
  1300. }
  1301. if (imgData.objectPosition) {
  1302. backgroundStyle["background-position"] = imgData.objectPosition;
  1303. }
  1304. if (imgData.backgroundColor) {
  1305. backgroundStyle["background-color"] = imgData.backgroundColor;
  1306. }
  1307. DomProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1308. imgElement.removeAttribute(dataAttributeName);
  1309. return true;
  1310. }
  1311. }
  1312. }
  1313. }
  1314. function log(...args) {
  1315. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1316. }
  1317. // -----
  1318. // Stats
  1319. // -----
  1320. const STATS_DEFAULT_VALUES = {
  1321. discarded: {
  1322. "HTML bytes": 0,
  1323. "hidden elements": 0,
  1324. "HTML imports": 0,
  1325. scripts: 0,
  1326. objects: 0,
  1327. "audio sources": 0,
  1328. "video sources": 0,
  1329. frames: 0,
  1330. "CSS rules": 0,
  1331. canvas: 0,
  1332. stylesheets: 0,
  1333. resources: 0,
  1334. medias: 0
  1335. },
  1336. processed: {
  1337. "HTML bytes": 0,
  1338. "hidden elements": 0,
  1339. "HTML imports": 0,
  1340. scripts: 0,
  1341. objects: 0,
  1342. "audio sources": 0,
  1343. "video sources": 0,
  1344. frames: 0,
  1345. "CSS rules": 0,
  1346. canvas: 0,
  1347. stylesheets: 0,
  1348. resources: 0,
  1349. medias: 0
  1350. }
  1351. };
  1352. class Stats {
  1353. constructor(options) {
  1354. this.options = options;
  1355. if (options.displayStats) {
  1356. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1357. }
  1358. }
  1359. set(type, subType, value) {
  1360. if (this.options.displayStats) {
  1361. this.data[type][subType] = value;
  1362. }
  1363. }
  1364. add(type, subType, value) {
  1365. if (this.options.displayStats) {
  1366. this.data[type][subType] += value;
  1367. }
  1368. }
  1369. addAll(pageData) {
  1370. if (this.options.displayStats) {
  1371. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1372. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1373. }
  1374. }
  1375. }
  1376. return { getClass };
  1377. })();