single-file-core.js 74 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788
  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(Util.removeCssComments(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. frameDoc.querySelectorAll("a[href]").forEach(element => {
  947. element.target = "_top";
  948. element.rel = "noopener";
  949. });
  950. }
  951. const pageData = await frameData.processor.getPageData();
  952. frameElement.removeAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  953. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND) || frameData.shadowRootContent) {
  954. frameElement.setAttribute("sandbox", "allow-popups allow-scripts allow-same-origin");
  955. } else {
  956. frameElement.setAttribute("sandbox", "allow-popups");
  957. }
  958. if (frameElement.tagName == "OBJECT") {
  959. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  960. } else {
  961. if (frameElement.tagName == "FRAME") {
  962. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/#/g, "%23"));
  963. } else {
  964. frameElement.setAttribute("srcdoc", pageData.content);
  965. frameElement.removeAttribute("src");
  966. }
  967. }
  968. this.stats.addAll(pageData);
  969. } else {
  970. this.stats.add("discarded", "frames", 1);
  971. }
  972. }
  973. }
  974. }));
  975. }
  976. }
  977. async processHtmlImports() {
  978. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  979. await Promise.all(linkElements.map(async linkElement => {
  980. const processor = this.relImportProcessors.get(linkElement);
  981. if (processor) {
  982. this.stats.add("processed", "HTML imports", 1);
  983. this.relImportProcessors.delete(linkElement);
  984. const pageData = await processor.getPageData();
  985. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  986. this.stats.addAll(pageData);
  987. } else {
  988. this.stats.add("discarded", "HTML imports", 1);
  989. }
  990. }));
  991. }
  992. replaceStylesheets() {
  993. this.doc.querySelectorAll("style").forEach(styleElement => {
  994. const stylesheetInfo = this.stylesheets.get(styleElement);
  995. if (stylesheetInfo) {
  996. this.stylesheets.delete(styleElement);
  997. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  998. styleElement.textContent = stylesheetContent;
  999. if (stylesheetInfo.mediaText) {
  1000. styleElement.media = stylesheetInfo.mediaText;
  1001. }
  1002. } else {
  1003. styleElement.remove();
  1004. }
  1005. });
  1006. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1007. const stylesheetInfo = this.stylesheets.get(linkElement);
  1008. if (stylesheetInfo) {
  1009. this.stylesheets.delete(linkElement);
  1010. const styleElement = this.doc.createElement("style");
  1011. if (stylesheetInfo.mediaText) {
  1012. styleElement.media = stylesheetInfo.mediaText;
  1013. }
  1014. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1015. styleElement.textContent = stylesheetContent;
  1016. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1017. } else {
  1018. linkElement.remove();
  1019. }
  1020. });
  1021. }
  1022. replaceStyleAttributes() {
  1023. this.doc.querySelectorAll("[style]").forEach(async element => {
  1024. const declarations = this.styles.get(element);
  1025. if (declarations) {
  1026. this.styles.delete(element);
  1027. let styleContent = cssTree.generate(declarations);
  1028. element.setAttribute("style", styleContent);
  1029. } else {
  1030. element.setAttribute("style", "");
  1031. }
  1032. });
  1033. }
  1034. insertVariables() {
  1035. if (this.cssVariables.size) {
  1036. const styleElement = this.doc.createElement("style");
  1037. const firstStyleElement = this.doc.head.querySelector("style");
  1038. if (firstStyleElement) {
  1039. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1040. } else {
  1041. this.doc.head.appendChild(styleElement);
  1042. }
  1043. let stylesheetContent = "";
  1044. this.cssVariables.forEach((content, indexResource) => {
  1045. this.cssVariables.delete(indexResource);
  1046. if (stylesheetContent) {
  1047. stylesheetContent += ";";
  1048. }
  1049. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1050. });
  1051. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1052. }
  1053. }
  1054. compressHTML() {
  1055. let size;
  1056. if (this.options.displayStats) {
  1057. size = docUtil.getContentSize(this.doc.documentElement.outerHTML);
  1058. }
  1059. docUtil.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: docUtil.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1060. if (this.options.displayStats) {
  1061. this.stats.add("discarded", "HTML bytes", size - docUtil.getContentSize(this.doc.documentElement.outerHTML));
  1062. }
  1063. }
  1064. }
  1065. // ---------------
  1066. // ProcessorHelper
  1067. // ---------------
  1068. const DATA_URI_PREFIX = "data:";
  1069. const ABOUT_BLANK_URI = "about:blank";
  1070. const EMPTY_DATA_URI = "data:base64,";
  1071. const REGEXP_AMP = /&/g;
  1072. const REGEXP_NBSP = /\u00a0/g;
  1073. const REGEXP_START_TAG = /</g;
  1074. const REGEXP_END_TAG = />/g;
  1075. const REGEXP_URL_HASH = /(#.+?)$/;
  1076. const PREFIX_DATA_URI_VND = "data:application/vnd.";
  1077. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1078. class ProcessorHelper {
  1079. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1080. const date = options.saveDate;
  1081. const url = docUtil.parseURL(options.saveUrl);
  1082. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash);
  1083. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash);
  1084. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash);
  1085. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash);
  1086. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash);
  1087. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash);
  1088. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash);
  1089. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash);
  1090. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash);
  1091. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash);
  1092. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash);
  1093. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash);
  1094. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1095. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash);
  1096. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash);
  1097. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash);
  1098. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash);
  1099. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1100. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash);
  1101. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash);
  1102. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash);
  1103. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash);
  1104. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash);
  1105. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash);
  1106. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash);
  1107. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash);
  1108. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash);
  1109. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash);
  1110. template = await Util.evalTemplateVariable(template, "url-href", () => decodeURI(url.href) || "No href", dontReplaceSlash);
  1111. template = await Util.evalTemplateVariable(template, "url-referrer", () => decodeURI(options.referrer) || "No referrer", dontReplaceSlash);
  1112. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash);
  1113. template = await Util.evalTemplateVariable(template, "url-pathname", () => decodeURI(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash);
  1114. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash);
  1115. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash);
  1116. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash);
  1117. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1118. for (const [name, value] of params) {
  1119. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash);
  1120. }
  1121. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1122. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash);
  1123. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash);
  1124. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash);
  1125. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decodeURI(Util.getLastSegment(url)) || "No last segment", dontReplaceSlash);
  1126. if (content) {
  1127. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => docUtil.digest("SHA-256", content), dontReplaceSlash);
  1128. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => docUtil.digest("SHA-384", content), dontReplaceSlash);
  1129. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => docUtil.digest("SHA-512", content), dontReplaceSlash);
  1130. }
  1131. return template.trim();
  1132. }
  1133. static setBackgroundImage(element, url, style) {
  1134. element.style.setProperty("background-blend-mode", "normal", "important");
  1135. element.style.setProperty("background-clip", "content-box", "important");
  1136. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1137. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1138. element.style.setProperty("background-image", url, "important");
  1139. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1140. element.style.setProperty("background-origin", "content-box", "important");
  1141. element.style.setProperty("background-repeat", "no-repeat", "important");
  1142. }
  1143. static processShortcutIcons(doc) {
  1144. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1145. if (!shortcutIcon) {
  1146. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1147. if (shortcutIcon) {
  1148. shortcutIcon.rel = "icon";
  1149. }
  1150. }
  1151. if (shortcutIcon) {
  1152. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1153. if (linkElement != shortcutIcon) {
  1154. linkElement.remove();
  1155. }
  1156. });
  1157. }
  1158. }
  1159. static removeSingleLineCssComments(stylesheet) {
  1160. const removedRules = [];
  1161. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1162. const ruleData = cssRule.data;
  1163. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1164. removedRules.push(cssRule);
  1165. }
  1166. }
  1167. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1168. }
  1169. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet) {
  1170. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1171. const imports = Util.getImportFunctions(stylesheetContent);
  1172. await Promise.all(imports.map(async cssImport => {
  1173. const match = Util.matchImport(cssImport);
  1174. if (match) {
  1175. let resourceURL = Util.normalizeURL(match.resourceURL);
  1176. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1177. try {
  1178. resourceURL = docUtil.resolveURL(match.resourceURL, baseURI);
  1179. } catch (error) {
  1180. // ignored
  1181. }
  1182. if (Util.testValidURL(resourceURL)) {
  1183. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  1184. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1185. resourceURL = content.resourceURL;
  1186. let importedStylesheetContent = Util.removeCssComments(content.data);
  1187. if (options.compressCSS) {
  1188. importedStylesheetContent = docUtil.compressCSS(importedStylesheetContent);
  1189. }
  1190. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1191. if (stylesheetContent.includes(cssImport)) {
  1192. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet);
  1193. workStylesheet.textContent = importedStylesheetContent;
  1194. if (workStylesheet.sheet.cssRules.length) {
  1195. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), importedStylesheetContent);
  1196. } else {
  1197. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), "");
  1198. }
  1199. }
  1200. }
  1201. }
  1202. }
  1203. }));
  1204. return stylesheetContent;
  1205. }
  1206. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1207. const urlFunctions = Util.getUrlFunctions(stylesheetContent);
  1208. urlFunctions.map(urlFunction => {
  1209. const originalResourceURL = Util.matchURL(urlFunction);
  1210. const resourceURL = Util.normalizeURL(originalResourceURL);
  1211. if (!Util.testIgnoredPath(resourceURL)) {
  1212. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1213. let resolvedURL;
  1214. try {
  1215. resolvedURL = docUtil.resolveURL(resourceURL, baseURI);
  1216. } catch (error) {
  1217. // ignored
  1218. }
  1219. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1220. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  1221. }
  1222. } else {
  1223. let newUrlFunction;
  1224. if (originalResourceURL) {
  1225. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1226. } else {
  1227. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1228. }
  1229. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1230. }
  1231. } else {
  1232. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  1233. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  1234. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  1235. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  1236. }
  1237. }
  1238. }
  1239. });
  1240. return stylesheetContent;
  1241. }
  1242. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1243. resourceURL = Util.normalizeURL(resourceURL);
  1244. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1245. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charset: options.charset };
  1246. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1247. resourceURL = content.resourceURL;
  1248. let stylesheetContent = Util.removeCssComments(content.data);
  1249. if (options.compressCSS) {
  1250. stylesheetContent = docUtil.compressCSS(stylesheetContent);
  1251. }
  1252. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1253. return stylesheetContent;
  1254. }
  1255. }
  1256. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1257. const promises = [];
  1258. const removedRules = [];
  1259. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1260. const ruleData = cssRule.data;
  1261. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1262. removedRules.push(cssRule);
  1263. } else if (ruleData.block && ruleData.block.children) {
  1264. if (ruleData.type == "Rule") {
  1265. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1266. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1267. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1268. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1269. promises.push(processFontFaceRule(ruleData));
  1270. }
  1271. }
  1272. }
  1273. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1274. await Promise.all(promises);
  1275. async function processFontFaceRule(ruleData) {
  1276. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1277. if (declaration.type == "Declaration" && declaration.value.children) {
  1278. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1279. await Promise.all(urlFunctions.map(async urlFunction => {
  1280. const originalResourceURL = Util.matchURL(urlFunction);
  1281. const resourceURL = Util.normalizeURL(originalResourceURL);
  1282. if (!Util.testIgnoredPath(resourceURL)) {
  1283. if (Util.testValidURL(resourceURL)) {
  1284. let { content } = await batchRequest.addURL(resourceURL, true);
  1285. if (content != EMPTY_DATA_URI && !content.startsWith(PREFIX_DATA_URI_VND) && !content.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1286. const validResource = await docUtil.validFont(urlFunction);
  1287. if (!validResource) {
  1288. content = EMPTY_DATA_URI;
  1289. }
  1290. }
  1291. replaceURLs(declaration, originalResourceURL, content);
  1292. }
  1293. } else if (resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(PREFIX_DATA_URI_VND) && !resourceURL.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1294. const validResource = await docUtil.validFont(urlFunction);
  1295. if (!validResource) {
  1296. replaceURLs(declaration, originalResourceURL, EMPTY_DATA_URI);
  1297. }
  1298. }
  1299. }));
  1300. }
  1301. }));
  1302. function replaceURLs(declaration, oldURL, newURL) {
  1303. declaration.value.children.forEach(token => {
  1304. if (token.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1305. token.value.value = newURL;
  1306. }
  1307. });
  1308. }
  1309. }
  1310. }
  1311. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1312. await Promise.all(declarations.map(async declaration => {
  1313. if (declaration.type == "Declaration" && declaration.value.children) {
  1314. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1315. await Promise.all(urlFunctions.map(async urlFunction => {
  1316. const originalResourceURL = Util.matchURL(urlFunction);
  1317. const resourceURL = Util.normalizeURL(originalResourceURL);
  1318. if (!Util.testIgnoredPath(resourceURL)) {
  1319. if (Util.testValidURL(resourceURL)) {
  1320. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1321. let variableDefined;
  1322. const tokens = [];
  1323. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1324. if (duplicate && options.groupDuplicateImages && rootFunction) {
  1325. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1326. tokens.push({ parent, token, value });
  1327. variableDefined = true;
  1328. } else {
  1329. token.data.value.value = content;
  1330. }
  1331. });
  1332. if (variableDefined) {
  1333. cssVariables.set(indexResource, content);
  1334. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1335. }
  1336. }
  1337. }
  1338. }));
  1339. }
  1340. }));
  1341. function findURLToken(url, children, callback, depth = 0) {
  1342. for (let token = children.head; token; token = token.next) {
  1343. if (token.data.children) {
  1344. findURLToken(url, token.data.children, callback, depth + 1);
  1345. }
  1346. if (token.data.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1347. callback(token, children, depth == 0);
  1348. }
  1349. }
  1350. }
  1351. }
  1352. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1353. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1354. let resourceURL = resourceElement.getAttribute(attributeName);
  1355. resourceURL = Util.normalizeURL(resourceURL);
  1356. if (!Util.testIgnoredPath(resourceURL)) {
  1357. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1358. if (Util.testValidPath(resourceURL)) {
  1359. try {
  1360. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1361. } catch (error) {
  1362. // ignored
  1363. }
  1364. if (Util.testValidURL(resourceURL)) {
  1365. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1366. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1367. resourceElement.remove();
  1368. } else {
  1369. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1370. if (!forbiddenPrefixFound) {
  1371. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1372. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1373. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1374. cssVariables.set(indexResource, content);
  1375. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1376. styles.set(resourceElement, declarationList);
  1377. } else {
  1378. resourceElement.setAttribute(attributeName, content);
  1379. }
  1380. } else {
  1381. resourceElement.setAttribute(attributeName, content);
  1382. }
  1383. }
  1384. }
  1385. }
  1386. }
  1387. }
  1388. }));
  1389. }
  1390. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1391. const attributeName = "xlink:href";
  1392. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1393. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1394. let resourceURL = Util.normalizeURL(originalResourceURL);
  1395. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1396. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1397. try {
  1398. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1399. } catch (error) {
  1400. // ignored
  1401. }
  1402. if (Util.testValidURL(resourceURL)) {
  1403. try {
  1404. const { content } = await batchRequest.addURL(resourceURL);
  1405. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1406. if (hashMatch && hashMatch[0]) {
  1407. const svgDoc = docUtil.parseSVGContent(content);
  1408. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1409. if (symbolElement) {
  1410. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1411. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1412. }
  1413. } else {
  1414. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1415. }
  1416. } catch (error) {
  1417. /* ignored */
  1418. }
  1419. }
  1420. } else if (resourceURL == options.url) {
  1421. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1422. }
  1423. }));
  1424. }
  1425. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1426. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1427. const srcset = docUtil.parseSrcset(resourceElement.getAttribute(attributeName));
  1428. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1429. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1430. if (!Util.testIgnoredPath(resourceURL)) {
  1431. if (Util.testValidPath(resourceURL)) {
  1432. try {
  1433. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1434. } catch (error) {
  1435. // ignored
  1436. }
  1437. if (Util.testValidURL(resourceURL)) {
  1438. const { content } = await batchRequest.addURL(resourceURL, true);
  1439. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1440. if (forbiddenPrefixFound) {
  1441. return "";
  1442. }
  1443. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1444. } else {
  1445. return "";
  1446. }
  1447. } else {
  1448. return "";
  1449. }
  1450. } else {
  1451. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1452. }
  1453. }));
  1454. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1455. }));
  1456. }
  1457. static replaceImageSource(imgElement, variableName, options) {
  1458. const dataAttributeName = docUtil.IMAGE_ATTRIBUTE_NAME;
  1459. if (imgElement.getAttribute(dataAttributeName) != null) {
  1460. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1461. if (imgData.replaceable) {
  1462. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imgData.size.pxWidth}" height="${imgData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1463. const backgroundStyle = {};
  1464. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1465. if (backgroundSize) {
  1466. backgroundStyle["background-size"] = imgData.objectFit;
  1467. }
  1468. if (imgData.objectPosition) {
  1469. backgroundStyle["background-position"] = imgData.objectPosition;
  1470. }
  1471. if (imgData.backgroundColor) {
  1472. backgroundStyle["background-color"] = imgData.backgroundColor;
  1473. }
  1474. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1475. imgElement.removeAttribute(dataAttributeName);
  1476. return true;
  1477. }
  1478. }
  1479. }
  1480. }
  1481. // ----
  1482. // Util
  1483. // ----
  1484. const BLOB_URI_PREFIX = "blob:";
  1485. const HTTP_URI_PREFIX = /^https?:\/\//;
  1486. const FILE_URI_PREFIX = /^file:\/\//;
  1487. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1488. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1489. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1490. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1491. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1492. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1493. 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;
  1494. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1495. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1496. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1497. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1498. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1499. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1500. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1501. class Util {
  1502. static normalizeURL(url) {
  1503. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1504. return url;
  1505. } else {
  1506. return url.split("#")[0];
  1507. }
  1508. }
  1509. static getCSSValue(value) {
  1510. let result = "";
  1511. try {
  1512. result = cssTree.generate(value);
  1513. } catch (error) {
  1514. // ignored
  1515. }
  1516. return result;
  1517. }
  1518. static getOnEventAttributeNames(doc) {
  1519. const element = doc.createElement("div");
  1520. const attributeNames = [];
  1521. for (const propertyName in element) {
  1522. if (propertyName.startsWith("on")) {
  1523. attributeNames.push(propertyName);
  1524. }
  1525. }
  1526. return attributeNames;
  1527. }
  1528. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1529. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1530. if (template && template.match(replaceRegExp)) {
  1531. let value = await valueGetter();
  1532. if (!dontReplaceSlash) {
  1533. value = value.replace(/\/+/g, "_");
  1534. }
  1535. return template.replace(replaceRegExp, value);
  1536. }
  1537. return template;
  1538. }
  1539. static getLastSegment(url) {
  1540. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1541. if (!lastSegment) {
  1542. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1543. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1544. }
  1545. if (!lastSegment) {
  1546. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1547. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1548. }
  1549. if (!lastSegment) {
  1550. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1551. }
  1552. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1553. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1554. lastSegment = lastSegmentMatch[1];
  1555. }
  1556. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1557. return lastSegment;
  1558. }
  1559. static getRegExp(string) {
  1560. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1561. }
  1562. static getUrlFunctions(stylesheetContent) {
  1563. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1564. }
  1565. static getImportFunctions(stylesheetContent) {
  1566. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1567. }
  1568. static findShortcutIcon(shortcutIcons) {
  1569. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1570. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1571. return shortcutIcons[0];
  1572. }
  1573. static matchURL(stylesheetContent) {
  1574. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1575. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1576. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1577. return match && match[1];
  1578. }
  1579. static testIgnoredPath(resourceURL) {
  1580. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1581. }
  1582. static testValidPath(resourceURL) {
  1583. return resourceURL && !resourceURL.match(EMPTY_URL);
  1584. }
  1585. static testValidURL(resourceURL) {
  1586. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1587. }
  1588. static matchImport(stylesheetContent) {
  1589. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1590. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1591. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1592. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1593. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1594. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1595. if (match) {
  1596. const [, resourceURL, media] = match;
  1597. return { resourceURL, media };
  1598. }
  1599. }
  1600. static removeCssComments(stylesheetContent) {
  1601. let start, end;
  1602. do {
  1603. start = stylesheetContent.indexOf("/*");
  1604. end = stylesheetContent.indexOf("*/", start + 2);
  1605. if (start != -1 && end != -1) {
  1606. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1607. }
  1608. } while (start != -1 && end != -1);
  1609. return stylesheetContent;
  1610. }
  1611. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1612. if (mediaQuery) {
  1613. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1614. } else {
  1615. return stylesheetContent;
  1616. }
  1617. }
  1618. }
  1619. function log(...args) {
  1620. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1621. }
  1622. // -----
  1623. // Stats
  1624. // -----
  1625. const STATS_DEFAULT_VALUES = {
  1626. discarded: {
  1627. "HTML bytes": 0,
  1628. "hidden elements": 0,
  1629. "HTML imports": 0,
  1630. scripts: 0,
  1631. objects: 0,
  1632. "audio sources": 0,
  1633. "video sources": 0,
  1634. frames: 0,
  1635. "CSS rules": 0,
  1636. canvas: 0,
  1637. stylesheets: 0,
  1638. resources: 0,
  1639. medias: 0
  1640. },
  1641. processed: {
  1642. "HTML bytes": 0,
  1643. "hidden elements": 0,
  1644. "HTML imports": 0,
  1645. scripts: 0,
  1646. objects: 0,
  1647. "audio sources": 0,
  1648. "video sources": 0,
  1649. frames: 0,
  1650. "CSS rules": 0,
  1651. canvas: 0,
  1652. stylesheets: 0,
  1653. resources: 0,
  1654. medias: 0
  1655. }
  1656. };
  1657. class Stats {
  1658. constructor(options) {
  1659. this.options = options;
  1660. if (options.displayStats) {
  1661. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1662. }
  1663. }
  1664. set(type, subType, value) {
  1665. if (this.options.displayStats) {
  1666. this.data[type][subType] = value;
  1667. }
  1668. }
  1669. add(type, subType, value) {
  1670. if (this.options.displayStats) {
  1671. this.data[type][subType] += value;
  1672. }
  1673. }
  1674. addAll(pageData) {
  1675. if (this.options.displayStats) {
  1676. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1677. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1678. }
  1679. }
  1680. }
  1681. return { getClass };
  1682. })();