single-file-core.js 74 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784
  1. /*
  2. * Copyright 2010-2019 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file 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 GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.SingleFileCore = this.SingleFileCore || (() => {
  24. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  25. const DEBUG = false;
  26. let docUtil, cssTree, sessionId = 0;
  27. function getClass(...args) {
  28. [docUtil, cssTree] = args;
  29. return SingleFileClass;
  30. }
  31. class SingleFileClass {
  32. constructor(options) {
  33. this.options = options;
  34. options.insertFaviconLink = true;
  35. if (options.sessionId === undefined) {
  36. options.sessionId = sessionId;
  37. sessionId++;
  38. }
  39. }
  40. async run() {
  41. this.runner = new Runner(this.options, true);
  42. await this.runner.loadPage();
  43. await this.runner.initialize();
  44. await this.runner.run();
  45. }
  46. async getPageData() {
  47. return this.runner.getPageData();
  48. }
  49. }
  50. SingleFileClass.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  51. // -------------
  52. // ProgressEvent
  53. // -------------
  54. const PAGE_LOADING = "page-loading";
  55. const PAGE_LOADED = "page-loaded";
  56. const RESOURCES_INITIALIZING = "resource-initializing";
  57. const RESOURCES_INITIALIZED = "resources-initialized";
  58. const RESOURCE_LOADED = "resource-loaded";
  59. const PAGE_ENDED = "page-ended";
  60. const STAGE_STARTED = "stage-started";
  61. const STAGE_ENDED = "stage-ended";
  62. const STAGE_TASK_STARTED = "stage-task-started";
  63. const STAGE_TASK_ENDED = "stage-task-ended";
  64. class ProgressEvent {
  65. constructor(type, detail) {
  66. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  67. }
  68. }
  69. // ------
  70. // Runner
  71. // ------
  72. const RESOLVE_URLS_STAGE = 0;
  73. const REPLACE_DATA_STAGE = 1;
  74. const REPLACE_DOCS_STAGE = 2;
  75. const POST_PROCESS_STAGE = 3;
  76. const STAGES = [{
  77. sequential: [
  78. { action: "preProcessPage" },
  79. { action: "insertShadowRootContents" },
  80. { action: "replaceStyleContents" },
  81. { option: "selected", action: "removeUnselectedElements" },
  82. { option: "removeVideoSrc", action: "insertVideoPosters" },
  83. { option: "removeFrames", action: "removeFrames" },
  84. { option: "removeImports", action: "removeImports" },
  85. { option: "removeScripts", action: "removeScripts" },
  86. { option: "removeVideoSrc", action: "removeVideoSources" },
  87. { option: "removeAudioSrc", action: "removeAudioSources" },
  88. { action: "removeDiscardedResources" },
  89. { action: "resetCharsetMeta" },
  90. { action: "setInputValues" },
  91. { option: "insertFaviconLink", action: "insertFaviconLink" },
  92. { action: "replaceCanvasElements" },
  93. { action: "insertFonts" },
  94. { option: "removeHiddenElements", action: "removeHiddenElements" },
  95. { action: "resolveHrefs" },
  96. { action: "resolveStyleAttributeURLs" }
  97. ],
  98. parallel: [
  99. { action: "resolveStylesheetURLs" },
  100. { option: "!removeFrames", action: "resolveFrameURLs" },
  101. { option: "!removeImports", action: "resolveHtmlImportURLs" }
  102. ]
  103. }, {
  104. sequential: [
  105. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  106. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  107. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  108. ],
  109. parallel: [
  110. { action: "processStylesheets" },
  111. { action: "processStyleAttributes" },
  112. { action: "processPageResources" },
  113. { option: "!removeScripts", action: "processScripts" }
  114. ]
  115. }, {
  116. sequential: [
  117. { option: "removeAlternativeImages", action: "removeAlternativeImages" },
  118. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" }
  119. ],
  120. parallel: [
  121. { option: "!removeFrames", action: "processFrames" },
  122. { option: "!removeImports", action: "processHtmlImports" },
  123. ]
  124. }, {
  125. sequential: [
  126. { action: "replaceStylesheets" },
  127. { action: "replaceStyleAttributes" },
  128. { action: "insertVariables" },
  129. { option: "compressHTML", action: "compressHTML" }
  130. ]
  131. }];
  132. class Runner {
  133. constructor(options, root) {
  134. this.root = root;
  135. this.options = options;
  136. this.options.url = this.options.url || (this.options.doc && this.options.doc.location.href);
  137. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  138. this.batchRequest = new BatchRequest();
  139. this.processor = new Processor(options, this.batchRequest);
  140. if (this.options.doc) {
  141. const docData = docUtil.preProcessDoc(this.options.doc, this.options.win, this.options);
  142. this.options.canvasData = docData.canvasData;
  143. this.options.fontsData = docData.fontsData;
  144. this.options.stylesheetContents = docData.stylesheetContents;
  145. this.options.imageData = docData.imageData;
  146. this.options.postersData = docData.postersData;
  147. this.options.usedFonts = docData.usedFonts;
  148. this.options.shadowRootContents = docData.shadowRootContents;
  149. this.options.referrer = docData.referrer;
  150. }
  151. if (this.options.saveRawPage) {
  152. this.options.removeFrames = true;
  153. }
  154. this.options.content = this.options.content || (this.options.doc ? docUtil.serialize(this.options.doc, false) : null);
  155. this.onprogress = options.onprogress || (() => { });
  156. }
  157. async loadPage() {
  158. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  159. await this.processor.loadPage(this.options.content);
  160. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  161. }
  162. async initialize() {
  163. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  164. await this.executeStage(RESOLVE_URLS_STAGE);
  165. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  166. if (this.options.doc) {
  167. docUtil.postProcessDoc(this.options.doc, this.options);
  168. }
  169. this.options.doc = null;
  170. this.options.win = null;
  171. }
  172. async run() {
  173. if (this.root) {
  174. this.processor.initialize(this.batchRequest);
  175. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  176. }
  177. await this.batchRequest.run(detail => {
  178. detail.pageURL = this.options.url;
  179. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  180. }, this.options);
  181. await this.pendingPromises;
  182. await this.executeStage(REPLACE_DOCS_STAGE);
  183. await this.executeStage(POST_PROCESS_STAGE);
  184. await this.processor.finalize();
  185. }
  186. getDocument() {
  187. return this.processor.doc;
  188. }
  189. async getPageData() {
  190. if (this.root) {
  191. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  192. }
  193. return this.processor.getPageData();
  194. }
  195. async executeStage(step) {
  196. if (DEBUG) {
  197. log("**** STARTED STAGE", step, "****");
  198. }
  199. const frame = !this.root;
  200. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  201. STAGES[step].sequential.forEach(task => {
  202. let startTime;
  203. if (DEBUG) {
  204. startTime = Date.now();
  205. log(" -- STARTED task =", task.action);
  206. }
  207. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  208. this.executeTask(task);
  209. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  210. if (DEBUG) {
  211. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  212. }
  213. });
  214. let parallelTasksPromise;
  215. if (STAGES[step].parallel) {
  216. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  217. let startTime;
  218. if (DEBUG) {
  219. startTime = Date.now();
  220. log(" // STARTED task =", task.action);
  221. }
  222. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  223. await this.executeTask(task);
  224. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  225. if (DEBUG) {
  226. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  227. }
  228. }));
  229. } else {
  230. parallelTasksPromise = Promise.resolve();
  231. }
  232. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  233. if (DEBUG) {
  234. log("**** ENDED STAGE", step, "****");
  235. }
  236. return parallelTasksPromise;
  237. }
  238. executeTask(task) {
  239. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  240. return this.processor[task.action]();
  241. }
  242. }
  243. }
  244. // ------------
  245. // BatchRequest
  246. // ------------
  247. class BatchRequest {
  248. constructor() {
  249. this.requests = new Map();
  250. this.duplicates = new Map();
  251. }
  252. async addURL(resourceURL, asDataURI, groupDuplicates) {
  253. return new Promise((resolve, reject) => {
  254. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  255. let resourceRequests = this.requests.get(requestKey);
  256. if (!resourceRequests) {
  257. resourceRequests = [];
  258. this.requests.set(requestKey, resourceRequests);
  259. }
  260. const callbacks = { resolve, reject };
  261. resourceRequests.push(callbacks);
  262. if (groupDuplicates) {
  263. let duplicateRequests = this.duplicates.get(requestKey);
  264. if (!duplicateRequests) {
  265. duplicateRequests = [];
  266. this.duplicates.set(requestKey, duplicateRequests);
  267. }
  268. duplicateRequests.push(callbacks);
  269. }
  270. });
  271. }
  272. getMaxResources() {
  273. return this.requests.size;
  274. }
  275. async run(onloadListener, options) {
  276. const resourceURLs = Array.from(this.requests.keys());
  277. let indexResource = 0;
  278. return Promise.all(resourceURLs.map(async requestKey => {
  279. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  280. const resourceRequests = this.requests.get(requestKey);
  281. this.requests.delete(requestKey);
  282. try {
  283. const content = await docUtil.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  284. indexResource = indexResource + 1;
  285. onloadListener({ index: indexResource, url: resourceURL });
  286. resourceRequests.forEach(callbacks => {
  287. const duplicateCallbacks = this.duplicates.get(requestKey);
  288. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  289. callbacks.resolve({ content: content.data, indexResource, duplicate });
  290. });
  291. } catch (error) {
  292. indexResource = indexResource + 1;
  293. onloadListener({ index: indexResource, url: resourceURL });
  294. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  295. }
  296. }));
  297. }
  298. }
  299. // ---------
  300. // Processor
  301. // ---------
  302. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  303. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  304. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  305. const SCRIPT_TAG_FOUND = /<script/gi;
  306. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  307. class Processor {
  308. constructor(options, batchRequest) {
  309. this.options = options;
  310. this.stats = new Stats(options);
  311. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  312. this.batchRequest = batchRequest;
  313. this.stylesheets = new Map();
  314. this.styles = new Map();
  315. this.cssVariables = new Map();
  316. }
  317. initialize() {
  318. this.options.saveDate = new Date();
  319. this.options.saveUrl = this.options.url;
  320. if (this.options.enableMaff) {
  321. this.maffMetaDataPromise = this.batchRequest.addURL(docUtil.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  322. }
  323. this.maxResources = this.batchRequest.getMaxResources();
  324. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.framesData) {
  325. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  326. }
  327. this.stats.set("processed", "resources", this.maxResources);
  328. }
  329. async loadPage(pageContent, charset) {
  330. let content;
  331. if (!pageContent || this.options.saveRawPage) {
  332. content = await docUtil.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charset });
  333. pageContent = content.data;
  334. }
  335. this.doc = docUtil.parseDocContent(pageContent, this.baseURI);
  336. if (this.options.saveRawPage) {
  337. let charset;
  338. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  339. const charsetDeclaration = element.content.split(";")[1];
  340. if (charsetDeclaration && !charset) {
  341. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  342. }
  343. });
  344. if (charset && charset != content.charset) {
  345. return this.loadPage(pageContent, charset);
  346. }
  347. }
  348. this.workStyleElement = this.doc.createElement("style");
  349. this.doc.body.appendChild(this.workStyleElement);
  350. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  351. }
  352. async finalize() {
  353. if (this.workStyleElement.parentNode) {
  354. this.workStyleElement.remove();
  355. }
  356. const metaCharset = this.doc.head.querySelector("meta[charset]");
  357. if (metaCharset) {
  358. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  359. }
  360. this.doc.querySelectorAll("base").forEach(element => element.remove());
  361. if (this.doc.head.querySelectorAll("*").length == 1 && metaCharset && this.doc.body.childNodes.length == 0) {
  362. this.doc.head.querySelector("meta[charset]").remove();
  363. }
  364. if (this.maffMetaDataPromise) {
  365. const maffMetaData = await this.maffMetaDataPromise;
  366. if (maffMetaData && maffMetaData.content) {
  367. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  368. const maffDoc = docUtil.parseXMLContent(maffMetaData.content);
  369. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  370. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  371. if (originalURLElement) {
  372. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  373. }
  374. if (archiveTimeElement) {
  375. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  376. if (value) {
  377. const date = new Date(value);
  378. if (!isNaN(date.getTime())) {
  379. this.options.saveDate = new Date(value);
  380. }
  381. }
  382. }
  383. }
  384. }
  385. const titleElement = this.doc.querySelector("title");
  386. const descriptionElement = this.doc.querySelector("meta[name=description]");
  387. const authorElement = this.doc.querySelector("meta[name=author]");
  388. const creatorElement = this.doc.querySelector("meta[name=creator]");
  389. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  390. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  391. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  392. this.options.info = {
  393. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  394. lang: this.doc.documentElement.lang,
  395. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  396. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  397. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : ""
  398. };
  399. }
  400. async getPageData() {
  401. docUtil.postProcessDoc(this.doc, this.options);
  402. const url = docUtil.parseURL(this.baseURI);
  403. if (this.options.insertSingleFileComment) {
  404. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  405. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  406. " \n url: " + this.options.saveUrl +
  407. " \n saved date: " + this.options.saveDate +
  408. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  409. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  410. }
  411. let size;
  412. if (this.options.displayStats) {
  413. size = docUtil.getContentSize(this.doc.documentElement.outerHTML);
  414. }
  415. const content = docUtil.serialize(this.doc, this.options.compressHTML);
  416. if (this.options.displayStats) {
  417. const contentSize = docUtil.getContentSize(content);
  418. this.stats.set("processed", "HTML bytes", contentSize);
  419. this.stats.add("discarded", "HTML bytes", size - contentSize);
  420. }
  421. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  422. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  423. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  424. if (!this.options.backgroundSave) {
  425. filename = filename.replace(/\//g, "_");
  426. }
  427. if (docUtil.getContentSize(filename) > 192) {
  428. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  429. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  430. filename = await docUtil.truncateContent(filename, 192 - extension.length);
  431. filename = filename + "…" + extension;
  432. }
  433. if (!filename) {
  434. filename = "Unnamed page";
  435. }
  436. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  437. return {
  438. stats: this.stats.data,
  439. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  440. filename,
  441. content
  442. };
  443. }
  444. preProcessPage() {
  445. if (this.options.win) {
  446. this.doc.body.querySelectorAll(":not(svg) title, meta, link[href][rel*=\"icon\"]").forEach(element => element instanceof this.options.win.HTMLElement && this.doc.head.appendChild(element));
  447. }
  448. if (this.options.imageData && !this.options.saveRawPage) {
  449. const dataAttributeName = docUtil.IMAGE_ATTRIBUTE_NAME;
  450. this.doc.querySelectorAll("img").forEach(imgElement => {
  451. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  452. if (this.options.removeHiddenElements && imgData.size && !imgData.size.pxWidth && !imgData.size.pxHeight) {
  453. imgElement.setAttribute("src", EMPTY_IMAGE);
  454. } else if (imgData.currentSrc) {
  455. imgElement.setAttribute("src", imgData.currentSrc);
  456. }
  457. });
  458. if (this.options.loadDeferredImages) {
  459. this.doc.querySelectorAll("img[data-src]").forEach(imgElement => {
  460. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  461. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  462. imgData.src = imgElement.dataset.src;
  463. imgElement.setAttribute("src", imgElement.dataset.src);
  464. imgElement.removeAttribute("data-src");
  465. }
  466. });
  467. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  468. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  469. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  470. imgElement.removeAttribute("data-srcset");
  471. }
  472. });
  473. }
  474. }
  475. }
  476. insertShadowRootContents() {
  477. if (this.options.shadowRootContents) {
  478. this.doc.querySelectorAll("[" + docUtil.SHADOW_ROOT_ATTRIBUTE_NAME + "]").forEach((element, elementIndex) => {
  479. const elementInfo = this.options.shadowRootContents[elementIndex];
  480. if (elementInfo) {
  481. const frameElement = this.doc.createElement("iframe");
  482. frameElement.setAttribute("style", "all:initial!important;border:0!important;width:100%!important;height:" + elementInfo.height + "px!important");
  483. const windowId = "shadow-" + this.options.framesData.length;
  484. frameElement.setAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME, windowId);
  485. this.options.framesData.push({ windowId, content: elementInfo.content, baseURI: this.baseURI, shadowRootContent: true });
  486. element.appendChild(frameElement);
  487. }
  488. });
  489. }
  490. }
  491. replaceStyleContents() {
  492. if (this.options.stylesheetContents) {
  493. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  494. if (this.options.stylesheetContents[styleIndex]) {
  495. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  496. }
  497. });
  498. }
  499. }
  500. removeUnselectedElements() {
  501. removeUnmarkedElements(this.doc.body);
  502. this.doc.body.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  503. function removeUnmarkedElements(element) {
  504. let selectedElementFound = false;
  505. Array.from(element.childNodes).forEach(node => {
  506. if (node.nodeType == 1) {
  507. const isSelectedElement = node.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  508. selectedElementFound = selectedElementFound || isSelectedElement;
  509. if (isSelectedElement) {
  510. node.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  511. removeUnmarkedElements(node);
  512. } else if (selectedElementFound) {
  513. removeNode(node);
  514. } else {
  515. node.style.setProperty("display", "none", "important");
  516. Array.from(node.childNodes).forEach(removeNode);
  517. }
  518. }
  519. });
  520. }
  521. function removeNode(node) {
  522. if (node.nodeType == 1) {
  523. const tagName = node.tagName && node.tagName.toLowerCase();
  524. if (tagName != "svg" && tagName != "style" && tagName != "link") {
  525. node.remove();
  526. }
  527. }
  528. }
  529. }
  530. insertVideoPosters() {
  531. if (this.options.postersData) {
  532. this.doc.querySelectorAll("video[src], video > source[src]").forEach((element, videoIndex) => {
  533. let videoElement;
  534. if (element.tagName == "VIDEO") {
  535. videoElement = element;
  536. } else {
  537. videoElement = element.parentElement;
  538. }
  539. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  540. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  541. }
  542. });
  543. }
  544. }
  545. removeFrames() {
  546. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  547. this.stats.set("discarded", "frames", frameElements.length);
  548. this.stats.set("processed", "frames", frameElements.length);
  549. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  550. }
  551. removeImports() {
  552. const importElements = this.doc.querySelectorAll("link[rel=import]");
  553. this.stats.set("discarded", "HTML imports", importElements.length);
  554. this.stats.set("processed", "HTML imports", importElements.length);
  555. importElements.forEach(element => element.remove());
  556. }
  557. removeScripts() {
  558. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  559. this.doc.querySelectorAll("[href]").forEach(element => {
  560. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  561. element.setAttribute("href", "");
  562. }
  563. });
  564. this.doc.querySelectorAll("[src]").forEach(element => {
  565. if (element.src && element.src.match(/^\s*javascript:/)) {
  566. element.removeAttribute("src");
  567. }
  568. });
  569. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  570. this.stats.set("discarded", "scripts", scriptElements.length);
  571. this.stats.set("processed", "scripts", scriptElements.length);
  572. scriptElements.forEach(element => element.remove());
  573. }
  574. removeVideoSources() {
  575. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  576. this.stats.set("discarded", "video sources", videoSourceElements.length);
  577. this.stats.set("processed", "video sources", videoSourceElements.length);
  578. videoSourceElements.forEach(element => {
  579. if (element.tagName == "SOURCE") {
  580. element.remove();
  581. } else {
  582. videoSourceElements.forEach(element => element.removeAttribute("src"));
  583. }
  584. });
  585. }
  586. removeAudioSources() {
  587. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  588. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  589. this.stats.set("processed", "audio sources", audioSourceElements.length);
  590. audioSourceElements.forEach(element => {
  591. if (element.tagName == "SOURCE") {
  592. element.remove();
  593. } else {
  594. audioSourceElements.forEach(element => element.removeAttribute("src"));
  595. }
  596. });
  597. }
  598. removeDiscardedResources() {
  599. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  600. const objectElements = this.doc.querySelectorAll("applet, meta[http-equiv=refresh], meta[disabled-http-equiv], object[data]:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed[src]:not([src*=\".svg\"]):not([src*=\".pdf\"])");
  601. this.stats.set("discarded", "objects", objectElements.length);
  602. this.stats.set("processed", "objects", objectElements.length);
  603. objectElements.forEach(element => element.remove());
  604. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  605. replacedAttributeValue.forEach(element => {
  606. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  607. if (relValue.length) {
  608. element.setAttribute("rel", relValue);
  609. } else {
  610. element.remove();
  611. }
  612. });
  613. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title]").forEach(element => element.remove());
  614. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  615. if (this.options.compressHTML) {
  616. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  617. }
  618. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  619. }
  620. resetCharsetMeta() {
  621. let charset;
  622. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  623. const charsetDeclaration = element.content.split(";")[1];
  624. if (charsetDeclaration && !charset) {
  625. charset = charsetDeclaration.split("=")[1];
  626. if (charset) {
  627. this.charset = charset.trim().toLowerCase();
  628. }
  629. }
  630. element.remove();
  631. });
  632. const metaElement = this.doc.createElement("meta");
  633. metaElement.setAttribute("charset", "utf-8");
  634. if (this.doc.head.firstChild) {
  635. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  636. } else {
  637. this.doc.head.appendChild(metaElement);
  638. }
  639. }
  640. setInputValues() {
  641. this.doc.querySelectorAll("input").forEach(input => {
  642. const value = input.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  643. input.setAttribute("value", value || "");
  644. });
  645. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  646. const value = input.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  647. if (value == "true") {
  648. input.setAttribute("checked", "");
  649. }
  650. });
  651. this.doc.querySelectorAll("textarea").forEach(textarea => {
  652. const value = textarea.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  653. textarea.textContent = value || "";
  654. });
  655. this.doc.querySelectorAll("select").forEach(select => {
  656. select.querySelectorAll("option").forEach(option => {
  657. const selected = option.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  658. if (selected) {
  659. option.setAttribute("selected", "");
  660. }
  661. });
  662. });
  663. }
  664. insertFaviconLink() {
  665. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  666. if (!faviconElement) {
  667. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  668. }
  669. if (!faviconElement) {
  670. faviconElement = this.doc.createElement("link");
  671. faviconElement.setAttribute("type", "image/x-icon");
  672. faviconElement.setAttribute("rel", "shortcut icon");
  673. faviconElement.setAttribute("href", "/favicon.ico");
  674. }
  675. this.doc.head.appendChild(faviconElement);
  676. }
  677. replaceCanvasElements() {
  678. if (this.options.canvasData) {
  679. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  680. const canvasData = this.options.canvasData[indexCanvasElement];
  681. if (canvasData) {
  682. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  683. this.stats.add("processed", "canvas", 1);
  684. }
  685. });
  686. }
  687. }
  688. insertFonts() {
  689. if (this.options.fontsData && this.options.fontsData.length) {
  690. let stylesheetContent = "";
  691. this.options.fontsData.forEach(fontStyles => {
  692. if (fontStyles["font-family"] && fontStyles.src) {
  693. stylesheetContent += "@font-face{";
  694. let stylesContent = "";
  695. Object.keys(fontStyles).forEach(fontStyle => {
  696. if (stylesContent) {
  697. stylesContent += ";";
  698. }
  699. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  700. });
  701. stylesheetContent += stylesContent + "}";
  702. }
  703. });
  704. if (stylesheetContent) {
  705. const styleElement = this.doc.createElement("style");
  706. styleElement.textContent = stylesheetContent;
  707. const existingStyleElement = this.doc.querySelector("style");
  708. if (existingStyleElement) {
  709. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  710. } else {
  711. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  712. }
  713. }
  714. }
  715. }
  716. removeHiddenElements() {
  717. const hiddenElements = this.doc.querySelectorAll("[" + docUtil.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  718. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  719. this.stats.set("processed", "hidden elements", hiddenElements.length);
  720. hiddenElements.forEach(element => element.remove());
  721. }
  722. resolveHrefs() {
  723. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  724. const href = element.getAttribute("href").trim();
  725. if (!Util.testIgnoredPath(href)) {
  726. let resolvedURL;
  727. try {
  728. resolvedURL = docUtil.resolveURL(href, this.options.baseURI || this.options.url);
  729. } catch (error) {
  730. // ignored
  731. }
  732. if (resolvedURL) {
  733. if (resolvedURL.startsWith(this.options.url + "#")) {
  734. resolvedURL = resolvedURL.substring(this.options.url.length);
  735. }
  736. try {
  737. element.setAttribute("href", resolvedURL);
  738. } catch (error) {
  739. // ignored
  740. }
  741. }
  742. }
  743. });
  744. }
  745. resolveStyleAttributeURLs() {
  746. this.doc.querySelectorAll("[style]").forEach(element => {
  747. let styleContent = element.getAttribute("style");
  748. if (this.options.compressCSS) {
  749. styleContent = docUtil.compressCSS(styleContent);
  750. }
  751. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  752. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  753. this.styles.set(element, declarationList);
  754. });
  755. }
  756. async resolveStylesheetURLs() {
  757. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  758. .map(async element => {
  759. let mediaText;
  760. if (element.media) {
  761. mediaText = element.media.toLowerCase();
  762. }
  763. this.stylesheets.set(element, { mediaText });
  764. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, url: this.options.url, charset: this.charset, compressCSS: this.options.compressCSS };
  765. let stylesheetContent;
  766. if (element.tagName.toLowerCase() == "link") {
  767. if (element.charset) {
  768. options.charset = element.charset;
  769. }
  770. stylesheetContent = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, this.baseURI, options, this.workStyleElement);
  771. } else {
  772. stylesheetContent = await ProcessorHelper.resolveImportURLs(element.textContent, this.baseURI, options, this.workStyleElement);
  773. }
  774. let stylesheet;
  775. try {
  776. stylesheet = cssTree.parse(stylesheetContent);
  777. } catch (error) {
  778. // ignored
  779. }
  780. if (stylesheet && stylesheet.children) {
  781. if (this.options.compressCSS) {
  782. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  783. }
  784. this.stylesheets.get(element).stylesheet = stylesheet;
  785. } else {
  786. this.stylesheets.delete(element);
  787. }
  788. }));
  789. }
  790. async resolveFrameURLs() {
  791. if (!this.options.saveRawPage && this.options.framesData) {
  792. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  793. await Promise.all(frameElements.map(async frameElement => {
  794. if (frameElement.tagName == "OBJECT") {
  795. frameElement.setAttribute("data", "data:text/html,");
  796. } else {
  797. frameElement.removeAttribute("src");
  798. frameElement.removeAttribute("srcdoc");
  799. }
  800. const frameWindowId = frameElement.getAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  801. if (frameWindowId) {
  802. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  803. if (frameData) {
  804. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  805. }
  806. }
  807. }));
  808. }
  809. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  810. options.insertSingleFileComment = false;
  811. options.insertFaviconLink = false;
  812. options.doc = null;
  813. options.win = null;
  814. options.url = frameData.baseURI;
  815. options.windowId = frameWindowId;
  816. if (frameData.content) {
  817. options.content = frameData.content;
  818. options.postersData = frameData.postersData;
  819. options.canvasData = frameData.canvasData;
  820. options.stylesheetContents = frameData.stylesheetContents;
  821. options.fontsData = frameData.fontsData;
  822. options.imageData = frameData.imageData;
  823. options.usedFonts = frameData.usedFonts;
  824. options.shadowRootContents = frameData.shadowRootContents;
  825. frameData.processor = new Runner(options);
  826. frameData.frameElement = frameElement;
  827. await frameData.processor.loadPage();
  828. await frameData.processor.initialize();
  829. frameData.maxResources = batchRequest.getMaxResources();
  830. }
  831. }
  832. }
  833. async resolveHtmlImportURLs() {
  834. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  835. if (!this.relImportProcessors) {
  836. this.relImportProcessors = new Map();
  837. }
  838. await Promise.all(linkElements.map(async linkElement => {
  839. const resourceURL = linkElement.href;
  840. linkElement.removeAttribute("href");
  841. const options = Object.create(this.options);
  842. options.insertSingleFileComment = false;
  843. options.insertFaviconLink = false;
  844. options.doc = null;
  845. options.win = null;
  846. options.url = resourceURL;
  847. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  848. const processor = new Runner(options);
  849. this.relImportProcessors.set(linkElement, processor);
  850. await processor.loadPage();
  851. return processor.initialize();
  852. }
  853. }));
  854. }
  855. removeUnusedStyles() {
  856. if (!this.mediaAllInfo) {
  857. this.mediaAllInfo = docUtil.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  858. }
  859. const stats = docUtil.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  860. this.stats.set("processed", "CSS rules", stats.processed);
  861. this.stats.set("discarded", "CSS rules", stats.discarded);
  862. }
  863. removeUnusedFonts() {
  864. docUtil.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  865. }
  866. removeAlternativeMedias() {
  867. const stats = docUtil.minifyMedias(this.stylesheets);
  868. this.stats.set("processed", "medias", stats.processed);
  869. this.stats.set("discarded", "medias", stats.discarded);
  870. }
  871. async processStylesheets() {
  872. await Promise.all(Array.from(this.stylesheets).map(([, stylesheetInfo]) =>
  873. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  874. ));
  875. }
  876. async processStyleAttributes() {
  877. return Promise.all(Array.from(this.styles).map(([, declarationList]) =>
  878. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  879. ));
  880. }
  881. async processPageResources() {
  882. const processAttributeArgs = [
  883. ["link[href][rel*=\"icon\"]", "href", false, true],
  884. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  885. ["img[src], input[src][type=image]", "src", true],
  886. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  887. ["video[poster]", "poster"],
  888. ["*[background]", "background"],
  889. ["image", "xlink:href"]
  890. ];
  891. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  892. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  893. );
  894. resourcePromises = resourcePromises.concat([
  895. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  896. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  897. ]);
  898. if (!this.options.removeAudioSrc) {
  899. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  900. }
  901. if (!this.options.removeVideoSrc) {
  902. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  903. }
  904. await Promise.all(resourcePromises);
  905. ProcessorHelper.processShortcutIcons(this.doc);
  906. }
  907. async processScripts() {
  908. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  909. let resourceURL;
  910. const scriptSrc = scriptElement.getAttribute("src");
  911. scriptElement.removeAttribute("src");
  912. scriptElement.textContent = "";
  913. try {
  914. resourceURL = docUtil.resolveURL(scriptSrc, this.baseURI);
  915. } catch (error) {
  916. // ignored
  917. }
  918. if (Util.testValidURL(resourceURL)) {
  919. this.stats.add("processed", "scripts", 1);
  920. const content = await docUtil.getContent(resourceURL, { asDataURI: true, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  921. scriptElement.setAttribute("src", content.data);
  922. }
  923. }));
  924. }
  925. removeAlternativeImages() {
  926. docUtil.removeAlternativeImages(this.doc);
  927. }
  928. removeAlternativeFonts() {
  929. docUtil.removeAlternativeFonts(this.doc, this.stylesheets);
  930. }
  931. async processFrames() {
  932. if (this.options.framesData) {
  933. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  934. await Promise.all(frameElements.map(async frameElement => {
  935. const frameWindowId = frameElement.getAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  936. if (frameWindowId) {
  937. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  938. if (frameData) {
  939. this.options.framesData = this.options.framesData.filter(frame => frame.windowId != frameWindowId);
  940. if (frameData.processor) {
  941. this.stats.add("processed", "frames", 1);
  942. await frameData.processor.run();
  943. if (frameData.shadowRootContent) {
  944. const frameDoc = frameData.processor.getDocument();
  945. frameDoc.documentElement.setAttribute("style", "overflow:hidden");
  946. }
  947. const pageData = await frameData.processor.getPageData();
  948. frameElement.removeAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  949. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  950. frameElement.setAttribute("sandbox", "allow-popups allow-scripts allow-same-origin");
  951. } else {
  952. frameElement.setAttribute("sandbox", "allow-popups");
  953. }
  954. if (frameElement.tagName == "OBJECT") {
  955. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  956. } else {
  957. if (frameElement.tagName == "FRAME") {
  958. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/#/g, "%23"));
  959. } else {
  960. frameElement.setAttribute("srcdoc", pageData.content);
  961. frameElement.removeAttribute("src");
  962. }
  963. }
  964. this.stats.addAll(pageData);
  965. } else {
  966. this.stats.add("discarded", "frames", 1);
  967. }
  968. }
  969. }
  970. }));
  971. }
  972. }
  973. async processHtmlImports() {
  974. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  975. await Promise.all(linkElements.map(async linkElement => {
  976. const processor = this.relImportProcessors.get(linkElement);
  977. if (processor) {
  978. this.stats.add("processed", "HTML imports", 1);
  979. this.relImportProcessors.delete(linkElement);
  980. const pageData = await processor.getPageData();
  981. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  982. this.stats.addAll(pageData);
  983. } else {
  984. this.stats.add("discarded", "HTML imports", 1);
  985. }
  986. }));
  987. }
  988. replaceStylesheets() {
  989. this.doc.querySelectorAll("style").forEach(styleElement => {
  990. const stylesheetInfo = this.stylesheets.get(styleElement);
  991. if (stylesheetInfo) {
  992. this.stylesheets.delete(styleElement);
  993. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  994. styleElement.textContent = stylesheetContent;
  995. if (stylesheetInfo.mediaText) {
  996. styleElement.media = stylesheetInfo.mediaText;
  997. }
  998. } else {
  999. styleElement.remove();
  1000. }
  1001. });
  1002. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1003. const stylesheetInfo = this.stylesheets.get(linkElement);
  1004. if (stylesheetInfo) {
  1005. this.stylesheets.delete(linkElement);
  1006. const styleElement = this.doc.createElement("style");
  1007. if (stylesheetInfo.mediaText) {
  1008. styleElement.media = stylesheetInfo.mediaText;
  1009. }
  1010. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1011. styleElement.textContent = stylesheetContent;
  1012. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1013. } else {
  1014. linkElement.remove();
  1015. }
  1016. });
  1017. }
  1018. replaceStyleAttributes() {
  1019. this.doc.querySelectorAll("[style]").forEach(async element => {
  1020. const declarations = this.styles.get(element);
  1021. if (declarations) {
  1022. this.styles.delete(element);
  1023. let styleContent = cssTree.generate(declarations);
  1024. element.setAttribute("style", styleContent);
  1025. } else {
  1026. element.setAttribute("style", "");
  1027. }
  1028. });
  1029. }
  1030. insertVariables() {
  1031. if (this.cssVariables.size) {
  1032. const styleElement = this.doc.createElement("style");
  1033. const firstStyleElement = this.doc.head.querySelector("style");
  1034. if (firstStyleElement) {
  1035. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1036. } else {
  1037. this.doc.head.appendChild(styleElement);
  1038. }
  1039. let stylesheetContent = "";
  1040. this.cssVariables.forEach((content, indexResource) => {
  1041. this.cssVariables.delete(indexResource);
  1042. if (stylesheetContent) {
  1043. stylesheetContent += ";";
  1044. }
  1045. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1046. });
  1047. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1048. }
  1049. }
  1050. compressHTML() {
  1051. let size;
  1052. if (this.options.displayStats) {
  1053. size = docUtil.getContentSize(this.doc.documentElement.outerHTML);
  1054. }
  1055. docUtil.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: docUtil.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1056. if (this.options.displayStats) {
  1057. this.stats.add("discarded", "HTML bytes", size - docUtil.getContentSize(this.doc.documentElement.outerHTML));
  1058. }
  1059. }
  1060. }
  1061. // ---------------
  1062. // ProcessorHelper
  1063. // ---------------
  1064. const DATA_URI_PREFIX = "data:";
  1065. const ABOUT_BLANK_URI = "about:blank";
  1066. const EMPTY_DATA_URI = "data:base64,";
  1067. const REGEXP_AMP = /&/g;
  1068. const REGEXP_NBSP = /\u00a0/g;
  1069. const REGEXP_START_TAG = /</g;
  1070. const REGEXP_END_TAG = />/g;
  1071. const REGEXP_URL_HASH = /(#.+?)$/;
  1072. const PREFIX_DATA_URI_VND = "data:application/vnd.";
  1073. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1074. class ProcessorHelper {
  1075. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1076. const date = options.saveDate;
  1077. const url = docUtil.parseURL(options.saveUrl);
  1078. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash);
  1079. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash);
  1080. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash);
  1081. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash);
  1082. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash);
  1083. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash);
  1084. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash);
  1085. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash);
  1086. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash);
  1087. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash);
  1088. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash);
  1089. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash);
  1090. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1091. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash);
  1092. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash);
  1093. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash);
  1094. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash);
  1095. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1096. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash);
  1097. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash);
  1098. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash);
  1099. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash);
  1100. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash);
  1101. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash);
  1102. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash);
  1103. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash);
  1104. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash);
  1105. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash);
  1106. template = await Util.evalTemplateVariable(template, "url-href", () => decodeURI(url.href) || "No href", dontReplaceSlash);
  1107. template = await Util.evalTemplateVariable(template, "url-referrer", () => decodeURI(options.referrer) || "No referrer", dontReplaceSlash);
  1108. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash);
  1109. template = await Util.evalTemplateVariable(template, "url-pathname", () => decodeURI(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash);
  1110. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash);
  1111. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash);
  1112. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash);
  1113. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1114. for (const [name, value] of params) {
  1115. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash);
  1116. }
  1117. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1118. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash);
  1119. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash);
  1120. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash);
  1121. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decodeURI(Util.getLastSegment(url)) || "No last segment", dontReplaceSlash);
  1122. if (content) {
  1123. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => docUtil.digest("SHA-256", content), dontReplaceSlash);
  1124. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => docUtil.digest("SHA-384", content), dontReplaceSlash);
  1125. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => docUtil.digest("SHA-512", content), dontReplaceSlash);
  1126. }
  1127. return template.trim();
  1128. }
  1129. static setBackgroundImage(element, url, style) {
  1130. element.style.setProperty("background-blend-mode", "normal", "important");
  1131. element.style.setProperty("background-clip", "content-box", "important");
  1132. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1133. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1134. element.style.setProperty("background-image", url, "important");
  1135. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1136. element.style.setProperty("background-origin", "content-box", "important");
  1137. element.style.setProperty("background-repeat", "no-repeat", "important");
  1138. }
  1139. static processShortcutIcons(doc) {
  1140. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1141. if (!shortcutIcon) {
  1142. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1143. if (shortcutIcon) {
  1144. shortcutIcon.rel = "icon";
  1145. }
  1146. }
  1147. if (shortcutIcon) {
  1148. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1149. if (linkElement != shortcutIcon) {
  1150. linkElement.remove();
  1151. }
  1152. });
  1153. }
  1154. }
  1155. static removeSingleLineCssComments(stylesheet) {
  1156. const removedRules = [];
  1157. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1158. const ruleData = cssRule.data;
  1159. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1160. removedRules.push(cssRule);
  1161. }
  1162. }
  1163. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1164. }
  1165. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet) {
  1166. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1167. const imports = Util.getImportFunctions(stylesheetContent);
  1168. await Promise.all(imports.map(async cssImport => {
  1169. const match = Util.matchImport(cssImport);
  1170. if (match) {
  1171. let resourceURL = Util.normalizeURL(match.resourceURL);
  1172. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1173. try {
  1174. resourceURL = docUtil.resolveURL(match.resourceURL, baseURI);
  1175. } catch (error) {
  1176. // ignored
  1177. }
  1178. if (Util.testValidURL(resourceURL)) {
  1179. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  1180. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1181. resourceURL = content.resourceURL;
  1182. let importedStylesheetContent = Util.removeCssComments(content.data);
  1183. if (options.compressCSS) {
  1184. importedStylesheetContent = docUtil.compressCSS(importedStylesheetContent);
  1185. }
  1186. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1187. if (stylesheetContent.includes(cssImport)) {
  1188. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet);
  1189. workStylesheet.textContent = importedStylesheetContent;
  1190. if (workStylesheet.sheet.cssRules.length) {
  1191. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), importedStylesheetContent);
  1192. } else {
  1193. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), "");
  1194. }
  1195. }
  1196. }
  1197. }
  1198. }
  1199. }));
  1200. return stylesheetContent;
  1201. }
  1202. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1203. const urlFunctions = Util.getUrlFunctions(stylesheetContent);
  1204. urlFunctions.map(urlFunction => {
  1205. const originalResourceURL = Util.matchURL(urlFunction);
  1206. const resourceURL = Util.normalizeURL(originalResourceURL);
  1207. if (!Util.testIgnoredPath(resourceURL)) {
  1208. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1209. let resolvedURL;
  1210. try {
  1211. resolvedURL = docUtil.resolveURL(resourceURL, baseURI);
  1212. } catch (error) {
  1213. // ignored
  1214. }
  1215. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1216. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  1217. }
  1218. } else {
  1219. let newUrlFunction;
  1220. if (originalResourceURL) {
  1221. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1222. } else {
  1223. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1224. }
  1225. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1226. }
  1227. } else {
  1228. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  1229. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  1230. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  1231. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  1232. }
  1233. }
  1234. }
  1235. });
  1236. return stylesheetContent;
  1237. }
  1238. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1239. resourceURL = Util.normalizeURL(resourceURL);
  1240. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1241. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charset: options.charset };
  1242. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1243. resourceURL = content.resourceURL;
  1244. let stylesheetContent = Util.removeCssComments(content.data);
  1245. if (options.compressCSS) {
  1246. stylesheetContent = docUtil.compressCSS(stylesheetContent);
  1247. }
  1248. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1249. return stylesheetContent;
  1250. }
  1251. }
  1252. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1253. const promises = [];
  1254. const removedRules = [];
  1255. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1256. const ruleData = cssRule.data;
  1257. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1258. removedRules.push(cssRule);
  1259. } else if (ruleData.block && ruleData.block.children) {
  1260. if (ruleData.type == "Rule") {
  1261. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1262. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1263. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1264. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1265. promises.push(processFontFaceRule(ruleData));
  1266. }
  1267. }
  1268. }
  1269. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1270. await Promise.all(promises);
  1271. async function processFontFaceRule(ruleData) {
  1272. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1273. if (declaration.type == "Declaration" && declaration.value.children) {
  1274. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1275. await Promise.all(urlFunctions.map(async urlFunction => {
  1276. const originalResourceURL = Util.matchURL(urlFunction);
  1277. const resourceURL = Util.normalizeURL(originalResourceURL);
  1278. if (!Util.testIgnoredPath(resourceURL)) {
  1279. if (Util.testValidURL(resourceURL)) {
  1280. let { content } = await batchRequest.addURL(resourceURL, true);
  1281. if (content != EMPTY_DATA_URI && !content.startsWith(PREFIX_DATA_URI_VND) && !content.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1282. const validResource = await docUtil.validFont(urlFunction);
  1283. if (!validResource) {
  1284. content = EMPTY_DATA_URI;
  1285. }
  1286. }
  1287. replaceURLs(declaration, originalResourceURL, content);
  1288. }
  1289. } else if (resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(PREFIX_DATA_URI_VND) && !resourceURL.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1290. const validResource = await docUtil.validFont(urlFunction);
  1291. if (!validResource) {
  1292. replaceURLs(declaration, originalResourceURL, EMPTY_DATA_URI);
  1293. }
  1294. }
  1295. }));
  1296. }
  1297. }));
  1298. function replaceURLs(declaration, oldURL, newURL) {
  1299. declaration.value.children.forEach(token => {
  1300. if (token.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1301. token.value.value = newURL;
  1302. }
  1303. });
  1304. }
  1305. }
  1306. }
  1307. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1308. await Promise.all(declarations.map(async declaration => {
  1309. if (declaration.type == "Declaration" && declaration.value.children) {
  1310. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1311. await Promise.all(urlFunctions.map(async urlFunction => {
  1312. const originalResourceURL = Util.matchURL(urlFunction);
  1313. const resourceURL = Util.normalizeURL(originalResourceURL);
  1314. if (!Util.testIgnoredPath(resourceURL)) {
  1315. if (Util.testValidURL(resourceURL)) {
  1316. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1317. let variableDefined;
  1318. const tokens = [];
  1319. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1320. if (duplicate && options.groupDuplicateImages && rootFunction) {
  1321. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1322. tokens.push({ parent, token, value });
  1323. variableDefined = true;
  1324. } else {
  1325. token.data.value.value = content;
  1326. }
  1327. });
  1328. if (variableDefined) {
  1329. cssVariables.set(indexResource, content);
  1330. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1331. }
  1332. }
  1333. }
  1334. }));
  1335. }
  1336. }));
  1337. function findURLToken(url, children, callback, depth = 0) {
  1338. for (let token = children.head; token; token = token.next) {
  1339. if (token.data.children) {
  1340. findURLToken(url, token.data.children, callback, depth + 1);
  1341. }
  1342. if (token.data.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1343. callback(token, children, depth == 0);
  1344. }
  1345. }
  1346. }
  1347. }
  1348. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1349. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1350. let resourceURL = resourceElement.getAttribute(attributeName);
  1351. resourceURL = Util.normalizeURL(resourceURL);
  1352. if (!Util.testIgnoredPath(resourceURL)) {
  1353. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1354. if (Util.testValidPath(resourceURL)) {
  1355. try {
  1356. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1357. } catch (error) {
  1358. // ignored
  1359. }
  1360. if (Util.testValidURL(resourceURL)) {
  1361. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1362. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1363. resourceElement.remove();
  1364. } else {
  1365. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1366. if (!forbiddenPrefixFound) {
  1367. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1368. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1369. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1370. cssVariables.set(indexResource, content);
  1371. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1372. styles.set(resourceElement, declarationList);
  1373. } else {
  1374. resourceElement.setAttribute(attributeName, content);
  1375. }
  1376. } else {
  1377. resourceElement.setAttribute(attributeName, content);
  1378. }
  1379. }
  1380. }
  1381. }
  1382. }
  1383. }
  1384. }));
  1385. }
  1386. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1387. const attributeName = "xlink:href";
  1388. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1389. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1390. let resourceURL = Util.normalizeURL(originalResourceURL);
  1391. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1392. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1393. try {
  1394. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1395. } catch (error) {
  1396. // ignored
  1397. }
  1398. if (Util.testValidURL(resourceURL)) {
  1399. try {
  1400. const { content } = await batchRequest.addURL(resourceURL);
  1401. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1402. if (hashMatch && hashMatch[0]) {
  1403. const svgDoc = docUtil.parseSVGContent(content);
  1404. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1405. if (symbolElement) {
  1406. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1407. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1408. }
  1409. } else {
  1410. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1411. }
  1412. } catch (error) {
  1413. /* ignored */
  1414. }
  1415. }
  1416. } else if (resourceURL == options.url) {
  1417. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1418. }
  1419. }));
  1420. }
  1421. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1422. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1423. const srcset = docUtil.parseSrcset(resourceElement.getAttribute(attributeName));
  1424. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1425. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1426. if (!Util.testIgnoredPath(resourceURL)) {
  1427. if (Util.testValidPath(resourceURL)) {
  1428. try {
  1429. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1430. } catch (error) {
  1431. // ignored
  1432. }
  1433. if (Util.testValidURL(resourceURL)) {
  1434. const { content } = await batchRequest.addURL(resourceURL, true);
  1435. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1436. if (forbiddenPrefixFound) {
  1437. return "";
  1438. }
  1439. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1440. } else {
  1441. return "";
  1442. }
  1443. } else {
  1444. return "";
  1445. }
  1446. } else {
  1447. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1448. }
  1449. }));
  1450. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1451. }));
  1452. }
  1453. static replaceImageSource(imgElement, variableName, options) {
  1454. const dataAttributeName = docUtil.IMAGE_ATTRIBUTE_NAME;
  1455. if (imgElement.getAttribute(dataAttributeName) != null) {
  1456. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1457. if (imgData.replaceable) {
  1458. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imgData.size.pxWidth}" height="${imgData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1459. const backgroundStyle = {};
  1460. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1461. if (backgroundSize) {
  1462. backgroundStyle["background-size"] = imgData.objectFit;
  1463. }
  1464. if (imgData.objectPosition) {
  1465. backgroundStyle["background-position"] = imgData.objectPosition;
  1466. }
  1467. if (imgData.backgroundColor) {
  1468. backgroundStyle["background-color"] = imgData.backgroundColor;
  1469. }
  1470. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1471. imgElement.removeAttribute(dataAttributeName);
  1472. return true;
  1473. }
  1474. }
  1475. }
  1476. }
  1477. // ----
  1478. // Util
  1479. // ----
  1480. const BLOB_URI_PREFIX = "blob:";
  1481. const HTTP_URI_PREFIX = /^https?:\/\//;
  1482. const FILE_URI_PREFIX = /^file:\/\//;
  1483. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1484. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1485. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1486. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1487. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1488. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1489. 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;
  1490. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1491. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1492. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1493. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1494. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1495. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1496. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1497. class Util {
  1498. static normalizeURL(url) {
  1499. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1500. return url;
  1501. } else {
  1502. return url.split("#")[0];
  1503. }
  1504. }
  1505. static getCSSValue(value) {
  1506. let result = "";
  1507. try {
  1508. result = cssTree.generate(value);
  1509. } catch (error) {
  1510. // ignored
  1511. }
  1512. return result;
  1513. }
  1514. static getOnEventAttributeNames(doc) {
  1515. const element = doc.createElement("div");
  1516. const attributeNames = [];
  1517. for (const propertyName in element) {
  1518. if (propertyName.startsWith("on")) {
  1519. attributeNames.push(propertyName);
  1520. }
  1521. }
  1522. return attributeNames;
  1523. }
  1524. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1525. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1526. if (template && template.match(replaceRegExp)) {
  1527. let value = await valueGetter();
  1528. if (!dontReplaceSlash) {
  1529. value = value.replace(/\/+/g, "_");
  1530. }
  1531. return template.replace(replaceRegExp, value);
  1532. }
  1533. return template;
  1534. }
  1535. static getLastSegment(url) {
  1536. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1537. if (!lastSegment) {
  1538. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1539. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1540. }
  1541. if (!lastSegment) {
  1542. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1543. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1544. }
  1545. if (!lastSegment) {
  1546. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1547. }
  1548. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1549. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1550. lastSegment = lastSegmentMatch[1];
  1551. }
  1552. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1553. return lastSegment;
  1554. }
  1555. static getRegExp(string) {
  1556. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1557. }
  1558. static getUrlFunctions(stylesheetContent) {
  1559. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1560. }
  1561. static getImportFunctions(stylesheetContent) {
  1562. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1563. }
  1564. static findShortcutIcon(shortcutIcons) {
  1565. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1566. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1567. return shortcutIcons[0];
  1568. }
  1569. static matchURL(stylesheetContent) {
  1570. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1571. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1572. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1573. return match && match[1];
  1574. }
  1575. static testIgnoredPath(resourceURL) {
  1576. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1577. }
  1578. static testValidPath(resourceURL) {
  1579. return resourceURL && !resourceURL.match(EMPTY_URL);
  1580. }
  1581. static testValidURL(resourceURL) {
  1582. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1583. }
  1584. static matchImport(stylesheetContent) {
  1585. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1586. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1587. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1588. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1589. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1590. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1591. if (match) {
  1592. const [, resourceURL, media] = match;
  1593. return { resourceURL, media };
  1594. }
  1595. }
  1596. static removeCssComments(stylesheetContent) {
  1597. let start, end;
  1598. do {
  1599. start = stylesheetContent.indexOf("/*");
  1600. end = stylesheetContent.indexOf("*/", start + 2);
  1601. if (start != -1 && end != -1) {
  1602. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1603. }
  1604. } while (start != -1 && end != -1);
  1605. return stylesheetContent;
  1606. }
  1607. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1608. if (mediaQuery) {
  1609. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1610. } else {
  1611. return stylesheetContent;
  1612. }
  1613. }
  1614. }
  1615. function log(...args) {
  1616. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1617. }
  1618. // -----
  1619. // Stats
  1620. // -----
  1621. const STATS_DEFAULT_VALUES = {
  1622. discarded: {
  1623. "HTML bytes": 0,
  1624. "hidden elements": 0,
  1625. "HTML imports": 0,
  1626. scripts: 0,
  1627. objects: 0,
  1628. "audio sources": 0,
  1629. "video sources": 0,
  1630. frames: 0,
  1631. "CSS rules": 0,
  1632. canvas: 0,
  1633. stylesheets: 0,
  1634. resources: 0,
  1635. medias: 0
  1636. },
  1637. processed: {
  1638. "HTML bytes": 0,
  1639. "hidden elements": 0,
  1640. "HTML imports": 0,
  1641. scripts: 0,
  1642. objects: 0,
  1643. "audio sources": 0,
  1644. "video sources": 0,
  1645. frames: 0,
  1646. "CSS rules": 0,
  1647. canvas: 0,
  1648. stylesheets: 0,
  1649. resources: 0,
  1650. medias: 0
  1651. }
  1652. };
  1653. class Stats {
  1654. constructor(options) {
  1655. this.options = options;
  1656. if (options.displayStats) {
  1657. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1658. }
  1659. }
  1660. set(type, subType, value) {
  1661. if (this.options.displayStats) {
  1662. this.data[type][subType] = value;
  1663. }
  1664. }
  1665. add(type, subType, value) {
  1666. if (this.options.displayStats) {
  1667. this.data[type][subType] += value;
  1668. }
  1669. }
  1670. addAll(pageData) {
  1671. if (this.options.displayStats) {
  1672. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1673. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1674. }
  1675. }
  1676. }
  1677. return { getClass };
  1678. })();