single-file-core.js 74 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776
  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. async getPageData() {
  187. if (this.root) {
  188. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  189. }
  190. return this.processor.getPageData();
  191. }
  192. async executeStage(step) {
  193. if (DEBUG) {
  194. log("**** STARTED STAGE", step, "****");
  195. }
  196. const frame = !this.root;
  197. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  198. STAGES[step].sequential.forEach(task => {
  199. let startTime;
  200. if (DEBUG) {
  201. startTime = Date.now();
  202. log(" -- STARTED task =", task.action);
  203. }
  204. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  205. this.executeTask(task);
  206. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  207. if (DEBUG) {
  208. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  209. }
  210. });
  211. let parallelTasksPromise;
  212. if (STAGES[step].parallel) {
  213. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  214. let startTime;
  215. if (DEBUG) {
  216. startTime = Date.now();
  217. log(" // STARTED task =", task.action);
  218. }
  219. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  220. await this.executeTask(task);
  221. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  222. if (DEBUG) {
  223. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  224. }
  225. }));
  226. } else {
  227. parallelTasksPromise = Promise.resolve();
  228. }
  229. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  230. if (DEBUG) {
  231. log("**** ENDED STAGE", step, "****");
  232. }
  233. return parallelTasksPromise;
  234. }
  235. executeTask(task) {
  236. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  237. return this.processor[task.action]();
  238. }
  239. }
  240. }
  241. // ------------
  242. // BatchRequest
  243. // ------------
  244. class BatchRequest {
  245. constructor() {
  246. this.requests = new Map();
  247. this.duplicates = new Map();
  248. }
  249. async addURL(resourceURL, asDataURI, groupDuplicates) {
  250. return new Promise((resolve, reject) => {
  251. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  252. let resourceRequests = this.requests.get(requestKey);
  253. if (!resourceRequests) {
  254. resourceRequests = [];
  255. this.requests.set(requestKey, resourceRequests);
  256. }
  257. const callbacks = { resolve, reject };
  258. resourceRequests.push(callbacks);
  259. if (groupDuplicates) {
  260. let duplicateRequests = this.duplicates.get(requestKey);
  261. if (!duplicateRequests) {
  262. duplicateRequests = [];
  263. this.duplicates.set(requestKey, duplicateRequests);
  264. }
  265. duplicateRequests.push(callbacks);
  266. }
  267. });
  268. }
  269. getMaxResources() {
  270. return this.requests.size;
  271. }
  272. async run(onloadListener, options) {
  273. const resourceURLs = Array.from(this.requests.keys());
  274. let indexResource = 0;
  275. return Promise.all(resourceURLs.map(async requestKey => {
  276. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  277. const resourceRequests = this.requests.get(requestKey);
  278. this.requests.delete(requestKey);
  279. try {
  280. const content = await docUtil.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  281. indexResource = indexResource + 1;
  282. onloadListener({ index: indexResource, url: resourceURL });
  283. resourceRequests.forEach(callbacks => {
  284. const duplicateCallbacks = this.duplicates.get(requestKey);
  285. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  286. callbacks.resolve({ content: content.data, indexResource, duplicate });
  287. });
  288. } catch (error) {
  289. indexResource = indexResource + 1;
  290. onloadListener({ index: indexResource, url: resourceURL });
  291. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  292. }
  293. }));
  294. }
  295. }
  296. // ---------
  297. // Processor
  298. // ---------
  299. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  300. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  301. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  302. const SCRIPT_TAG_FOUND = /<script/gi;
  303. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  304. class Processor {
  305. constructor(options, batchRequest) {
  306. this.options = options;
  307. this.stats = new Stats(options);
  308. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  309. this.batchRequest = batchRequest;
  310. this.stylesheets = new Map();
  311. this.styles = new Map();
  312. this.cssVariables = new Map();
  313. }
  314. initialize() {
  315. this.options.saveDate = new Date();
  316. this.options.saveUrl = this.options.url;
  317. if (this.options.enableMaff) {
  318. this.maffMetaDataPromise = this.batchRequest.addURL(docUtil.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  319. }
  320. this.maxResources = this.batchRequest.getMaxResources();
  321. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.framesData) {
  322. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  323. }
  324. this.stats.set("processed", "resources", this.maxResources);
  325. }
  326. async loadPage(pageContent, charset) {
  327. let content;
  328. if (!pageContent || this.options.saveRawPage) {
  329. content = await docUtil.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charset });
  330. pageContent = content.data;
  331. }
  332. this.doc = docUtil.parseDocContent(pageContent, this.baseURI);
  333. if (this.options.saveRawPage) {
  334. let charset;
  335. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  336. const charsetDeclaration = element.content.split(";")[1];
  337. if (charsetDeclaration && !charset) {
  338. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  339. }
  340. });
  341. if (charset && charset != content.charset) {
  342. return this.loadPage(pageContent, charset);
  343. }
  344. }
  345. this.workStyleElement = this.doc.createElement("style");
  346. this.doc.body.appendChild(this.workStyleElement);
  347. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  348. }
  349. async finalize() {
  350. if (this.workStyleElement.parentNode) {
  351. this.workStyleElement.remove();
  352. }
  353. const metaCharset = this.doc.head.querySelector("meta[charset]");
  354. if (metaCharset) {
  355. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  356. }
  357. this.doc.querySelectorAll("base").forEach(element => element.remove());
  358. if (this.doc.head.querySelectorAll("*").length == 1 && metaCharset && this.doc.body.childNodes.length == 0) {
  359. this.doc.head.querySelector("meta[charset]").remove();
  360. }
  361. if (this.maffMetaDataPromise) {
  362. const maffMetaData = await this.maffMetaDataPromise;
  363. if (maffMetaData && maffMetaData.content) {
  364. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  365. const maffDoc = docUtil.parseXMLContent(maffMetaData.content);
  366. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  367. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  368. if (originalURLElement) {
  369. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  370. }
  371. if (archiveTimeElement) {
  372. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  373. if (value) {
  374. const date = new Date(value);
  375. if (!isNaN(date.getTime())) {
  376. this.options.saveDate = new Date(value);
  377. }
  378. }
  379. }
  380. }
  381. }
  382. const titleElement = this.doc.querySelector("title");
  383. const descriptionElement = this.doc.querySelector("meta[name=description]");
  384. const authorElement = this.doc.querySelector("meta[name=author]");
  385. const creatorElement = this.doc.querySelector("meta[name=creator]");
  386. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  387. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  388. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  389. this.options.info = {
  390. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  391. lang: this.doc.documentElement.lang,
  392. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  393. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  394. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : ""
  395. };
  396. }
  397. async getPageData() {
  398. docUtil.postProcessDoc(this.doc, this.options);
  399. const url = docUtil.parseURL(this.baseURI);
  400. if (this.options.insertSingleFileComment) {
  401. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  402. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  403. " \n url: " + this.options.saveUrl +
  404. " \n saved date: " + this.options.saveDate +
  405. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  406. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  407. }
  408. let size;
  409. if (this.options.displayStats) {
  410. size = docUtil.getContentSize(this.doc.documentElement.outerHTML);
  411. }
  412. const content = docUtil.serialize(this.doc, this.options.compressHTML);
  413. if (this.options.displayStats) {
  414. const contentSize = docUtil.getContentSize(content);
  415. this.stats.set("processed", "HTML bytes", contentSize);
  416. this.stats.add("discarded", "HTML bytes", size - contentSize);
  417. }
  418. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  419. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  420. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  421. if (!this.options.backgroundSave) {
  422. filename = filename.replace(/\//g, "_");
  423. }
  424. if (docUtil.getContentSize(filename) > 192) {
  425. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  426. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  427. filename = await docUtil.truncateContent(filename, 192 - extension.length);
  428. filename = filename + "…" + extension;
  429. }
  430. if (!filename) {
  431. filename = "Unnamed page";
  432. }
  433. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  434. return {
  435. stats: this.stats.data,
  436. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  437. filename,
  438. content
  439. };
  440. }
  441. preProcessPage() {
  442. if (this.options.win) {
  443. 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));
  444. }
  445. if (this.options.imageData && !this.options.saveRawPage) {
  446. const dataAttributeName = docUtil.IMAGE_ATTRIBUTE_NAME;
  447. this.doc.querySelectorAll("img").forEach(imgElement => {
  448. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  449. if (this.options.removeHiddenElements && imgData.size && !imgData.size.pxWidth && !imgData.size.pxHeight) {
  450. imgElement.setAttribute("src", EMPTY_IMAGE);
  451. } else if (imgData.currentSrc) {
  452. imgElement.setAttribute("src", imgData.currentSrc);
  453. }
  454. });
  455. if (this.options.loadDeferredImages) {
  456. this.doc.querySelectorAll("img[data-src]").forEach(imgElement => {
  457. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  458. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  459. imgData.src = imgElement.dataset.src;
  460. imgElement.setAttribute("src", imgElement.dataset.src);
  461. imgElement.removeAttribute("data-src");
  462. }
  463. });
  464. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  465. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  466. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  467. imgElement.removeAttribute("data-srcset");
  468. }
  469. });
  470. }
  471. }
  472. }
  473. insertShadowRootContents() {
  474. if (this.options.shadowRootContents) {
  475. this.doc.querySelectorAll("[" + docUtil.SHADOW_ROOT_ATTRIBUTE_NAME + "]").forEach((element, elementIndex) => {
  476. const elementInfo = this.options.shadowRootContents[elementIndex];
  477. if (elementInfo) {
  478. const frameElement = this.doc.createElement("iframe");
  479. frameElement.setAttribute("style", "all:initial!important;border:0!important;width:100%!important;height:" + elementInfo.height + "px!important");
  480. const windowId = "shadow-" + this.options.framesData.length;
  481. frameElement.setAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME, windowId);
  482. this.options.framesData.push({ windowId, content: elementInfo.content, baseURI: this.baseURI });
  483. element.appendChild(frameElement);
  484. }
  485. });
  486. }
  487. }
  488. replaceStyleContents() {
  489. if (this.options.stylesheetContents) {
  490. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  491. if (this.options.stylesheetContents[styleIndex]) {
  492. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  493. }
  494. });
  495. }
  496. }
  497. removeUnselectedElements() {
  498. removeUnmarkedElements(this.doc.body);
  499. this.doc.body.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  500. function removeUnmarkedElements(element) {
  501. let selectedElementFound = false;
  502. Array.from(element.childNodes).forEach(node => {
  503. if (node.nodeType == 1) {
  504. const isSelectedElement = node.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  505. selectedElementFound = selectedElementFound || isSelectedElement;
  506. if (isSelectedElement) {
  507. node.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  508. removeUnmarkedElements(node);
  509. } else if (selectedElementFound) {
  510. removeNode(node);
  511. } else {
  512. node.style.setProperty("display", "none", "important");
  513. Array.from(node.childNodes).forEach(removeNode);
  514. }
  515. }
  516. });
  517. }
  518. function removeNode(node) {
  519. if (node.nodeType == 1) {
  520. const tagName = node.tagName && node.tagName.toLowerCase();
  521. if (tagName != "svg" && tagName != "style" && tagName != "link") {
  522. node.remove();
  523. }
  524. }
  525. }
  526. }
  527. insertVideoPosters() {
  528. if (this.options.postersData) {
  529. this.doc.querySelectorAll("video[src], video > source[src]").forEach((element, videoIndex) => {
  530. let videoElement;
  531. if (element.tagName == "VIDEO") {
  532. videoElement = element;
  533. } else {
  534. videoElement = element.parentElement;
  535. }
  536. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  537. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  538. }
  539. });
  540. }
  541. }
  542. removeFrames() {
  543. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  544. this.stats.set("discarded", "frames", frameElements.length);
  545. this.stats.set("processed", "frames", frameElements.length);
  546. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  547. }
  548. removeImports() {
  549. const importElements = this.doc.querySelectorAll("link[rel=import]");
  550. this.stats.set("discarded", "HTML imports", importElements.length);
  551. this.stats.set("processed", "HTML imports", importElements.length);
  552. importElements.forEach(element => element.remove());
  553. }
  554. removeScripts() {
  555. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  556. this.doc.querySelectorAll("[href]").forEach(element => {
  557. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  558. element.setAttribute("href", "");
  559. }
  560. });
  561. this.doc.querySelectorAll("[src]").forEach(element => {
  562. if (element.src && element.src.match(/^\s*javascript:/)) {
  563. element.removeAttribute("src");
  564. }
  565. });
  566. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  567. this.stats.set("discarded", "scripts", scriptElements.length);
  568. this.stats.set("processed", "scripts", scriptElements.length);
  569. scriptElements.forEach(element => element.remove());
  570. }
  571. removeVideoSources() {
  572. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  573. this.stats.set("discarded", "video sources", videoSourceElements.length);
  574. this.stats.set("processed", "video sources", videoSourceElements.length);
  575. videoSourceElements.forEach(element => {
  576. if (element.tagName == "SOURCE") {
  577. element.remove();
  578. } else {
  579. videoSourceElements.forEach(element => element.removeAttribute("src"));
  580. }
  581. });
  582. }
  583. removeAudioSources() {
  584. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  585. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  586. this.stats.set("processed", "audio sources", audioSourceElements.length);
  587. audioSourceElements.forEach(element => {
  588. if (element.tagName == "SOURCE") {
  589. element.remove();
  590. } else {
  591. audioSourceElements.forEach(element => element.removeAttribute("src"));
  592. }
  593. });
  594. }
  595. removeDiscardedResources() {
  596. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  597. 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\"])");
  598. this.stats.set("discarded", "objects", objectElements.length);
  599. this.stats.set("processed", "objects", objectElements.length);
  600. objectElements.forEach(element => element.remove());
  601. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  602. replacedAttributeValue.forEach(element => {
  603. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  604. if (relValue.length) {
  605. element.setAttribute("rel", relValue);
  606. } else {
  607. element.remove();
  608. }
  609. });
  610. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title]").forEach(element => element.remove());
  611. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  612. if (this.options.compressHTML) {
  613. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  614. }
  615. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  616. }
  617. resetCharsetMeta() {
  618. let charset;
  619. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  620. const charsetDeclaration = element.content.split(";")[1];
  621. if (charsetDeclaration && !charset) {
  622. charset = charsetDeclaration.split("=")[1];
  623. if (charset) {
  624. this.charset = charset.trim().toLowerCase();
  625. }
  626. }
  627. element.remove();
  628. });
  629. const metaElement = this.doc.createElement("meta");
  630. metaElement.setAttribute("charset", "utf-8");
  631. if (this.doc.head.firstChild) {
  632. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  633. } else {
  634. this.doc.head.appendChild(metaElement);
  635. }
  636. }
  637. setInputValues() {
  638. this.doc.querySelectorAll("input").forEach(input => {
  639. const value = input.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  640. input.setAttribute("value", value || "");
  641. });
  642. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  643. const value = input.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  644. if (value == "true") {
  645. input.setAttribute("checked", "");
  646. }
  647. });
  648. this.doc.querySelectorAll("textarea").forEach(textarea => {
  649. const value = textarea.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  650. textarea.textContent = value || "";
  651. });
  652. this.doc.querySelectorAll("select").forEach(select => {
  653. select.querySelectorAll("option").forEach(option => {
  654. const selected = option.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  655. if (selected) {
  656. option.setAttribute("selected", "");
  657. }
  658. });
  659. });
  660. }
  661. insertFaviconLink() {
  662. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  663. if (!faviconElement) {
  664. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  665. }
  666. if (!faviconElement) {
  667. faviconElement = this.doc.createElement("link");
  668. faviconElement.setAttribute("type", "image/x-icon");
  669. faviconElement.setAttribute("rel", "shortcut icon");
  670. faviconElement.setAttribute("href", "/favicon.ico");
  671. }
  672. this.doc.head.appendChild(faviconElement);
  673. }
  674. replaceCanvasElements() {
  675. if (this.options.canvasData) {
  676. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  677. const canvasData = this.options.canvasData[indexCanvasElement];
  678. if (canvasData) {
  679. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  680. this.stats.add("processed", "canvas", 1);
  681. }
  682. });
  683. }
  684. }
  685. insertFonts() {
  686. if (this.options.fontsData && this.options.fontsData.length) {
  687. let stylesheetContent = "";
  688. this.options.fontsData.forEach(fontStyles => {
  689. if (fontStyles["font-family"] && fontStyles.src) {
  690. stylesheetContent += "@font-face{";
  691. let stylesContent = "";
  692. Object.keys(fontStyles).forEach(fontStyle => {
  693. if (stylesContent) {
  694. stylesContent += ";";
  695. }
  696. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  697. });
  698. stylesheetContent += stylesContent + "}";
  699. }
  700. });
  701. if (stylesheetContent) {
  702. const styleElement = this.doc.createElement("style");
  703. styleElement.textContent = stylesheetContent;
  704. const existingStyleElement = this.doc.querySelector("style");
  705. if (existingStyleElement) {
  706. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  707. } else {
  708. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  709. }
  710. }
  711. }
  712. }
  713. removeHiddenElements() {
  714. const hiddenElements = this.doc.querySelectorAll("[" + docUtil.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  715. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  716. this.stats.set("processed", "hidden elements", hiddenElements.length);
  717. hiddenElements.forEach(element => element.remove());
  718. }
  719. resolveHrefs() {
  720. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  721. const href = element.getAttribute("href").trim();
  722. if (!Util.testIgnoredPath(href)) {
  723. let resolvedURL;
  724. try {
  725. resolvedURL = docUtil.resolveURL(href, this.options.baseURI || this.options.url);
  726. } catch (error) {
  727. // ignored
  728. }
  729. if (resolvedURL) {
  730. if (resolvedURL.startsWith(this.options.url + "#")) {
  731. resolvedURL = resolvedURL.substring(this.options.url.length);
  732. }
  733. try {
  734. element.setAttribute("href", resolvedURL);
  735. } catch (error) {
  736. // ignored
  737. }
  738. }
  739. }
  740. });
  741. }
  742. resolveStyleAttributeURLs() {
  743. this.doc.querySelectorAll("[style]").forEach(element => {
  744. let styleContent = element.getAttribute("style");
  745. if (this.options.compressCSS) {
  746. styleContent = docUtil.compressCSS(styleContent);
  747. }
  748. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  749. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  750. this.styles.set(element, declarationList);
  751. });
  752. }
  753. async resolveStylesheetURLs() {
  754. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  755. .map(async element => {
  756. let mediaText;
  757. if (element.media) {
  758. mediaText = element.media.toLowerCase();
  759. }
  760. this.stylesheets.set(element, { mediaText });
  761. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, url: this.options.url, charset: this.charset, compressCSS: this.options.compressCSS };
  762. let stylesheetContent;
  763. if (element.tagName.toLowerCase() == "link") {
  764. if (element.charset) {
  765. options.charset = element.charset;
  766. }
  767. stylesheetContent = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, this.baseURI, options, this.workStyleElement);
  768. } else {
  769. stylesheetContent = await ProcessorHelper.resolveImportURLs(element.textContent, this.baseURI, options, this.workStyleElement);
  770. }
  771. let stylesheet;
  772. try {
  773. stylesheet = cssTree.parse(stylesheetContent);
  774. } catch (error) {
  775. // ignored
  776. }
  777. if (stylesheet && stylesheet.children) {
  778. if (this.options.compressCSS) {
  779. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  780. }
  781. this.stylesheets.get(element).stylesheet = stylesheet;
  782. } else {
  783. this.stylesheets.delete(element);
  784. }
  785. }));
  786. }
  787. async resolveFrameURLs() {
  788. if (!this.options.saveRawPage && this.options.framesData) {
  789. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  790. await Promise.all(frameElements.map(async frameElement => {
  791. if (frameElement.tagName == "OBJECT") {
  792. frameElement.setAttribute("data", "data:text/html,");
  793. } else {
  794. frameElement.removeAttribute("src");
  795. frameElement.removeAttribute("srcdoc");
  796. }
  797. const frameWindowId = frameElement.getAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  798. if (frameWindowId) {
  799. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  800. if (frameData) {
  801. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  802. }
  803. }
  804. }));
  805. }
  806. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  807. options.insertSingleFileComment = false;
  808. options.insertFaviconLink = false;
  809. options.doc = null;
  810. options.win = null;
  811. options.url = frameData.baseURI;
  812. options.windowId = frameWindowId;
  813. if (frameData.content) {
  814. options.content = frameData.content;
  815. options.postersData = frameData.postersData;
  816. options.canvasData = frameData.canvasData;
  817. options.stylesheetContents = frameData.stylesheetContents;
  818. options.fontsData = frameData.fontsData;
  819. options.imageData = frameData.imageData;
  820. options.usedFonts = frameData.usedFonts;
  821. options.shadowRootContents = frameData.shadowRootContents;
  822. frameData.processor = new Runner(options);
  823. frameData.frameElement = frameElement;
  824. await frameData.processor.loadPage();
  825. await frameData.processor.initialize();
  826. frameData.maxResources = batchRequest.getMaxResources();
  827. }
  828. }
  829. }
  830. async resolveHtmlImportURLs() {
  831. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  832. if (!this.relImportProcessors) {
  833. this.relImportProcessors = new Map();
  834. }
  835. await Promise.all(linkElements.map(async linkElement => {
  836. const resourceURL = linkElement.href;
  837. linkElement.removeAttribute("href");
  838. const options = Object.create(this.options);
  839. options.insertSingleFileComment = false;
  840. options.insertFaviconLink = false;
  841. options.doc = null;
  842. options.win = null;
  843. options.url = resourceURL;
  844. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  845. const processor = new Runner(options);
  846. this.relImportProcessors.set(linkElement, processor);
  847. await processor.loadPage();
  848. return processor.initialize();
  849. }
  850. }));
  851. }
  852. removeUnusedStyles() {
  853. if (!this.mediaAllInfo) {
  854. this.mediaAllInfo = docUtil.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  855. }
  856. const stats = docUtil.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  857. this.stats.set("processed", "CSS rules", stats.processed);
  858. this.stats.set("discarded", "CSS rules", stats.discarded);
  859. }
  860. removeUnusedFonts() {
  861. docUtil.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  862. }
  863. removeAlternativeMedias() {
  864. const stats = docUtil.minifyMedias(this.stylesheets);
  865. this.stats.set("processed", "medias", stats.processed);
  866. this.stats.set("discarded", "medias", stats.discarded);
  867. }
  868. async processStylesheets() {
  869. await Promise.all(Array.from(this.stylesheets).map(([, stylesheetInfo]) =>
  870. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  871. ));
  872. }
  873. async processStyleAttributes() {
  874. return Promise.all(Array.from(this.styles).map(([, declarationList]) =>
  875. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  876. ));
  877. }
  878. async processPageResources() {
  879. const processAttributeArgs = [
  880. ["link[href][rel*=\"icon\"]", "href", false, true],
  881. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  882. ["img[src], input[src][type=image]", "src", true],
  883. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  884. ["video[poster]", "poster"],
  885. ["*[background]", "background"],
  886. ["image", "xlink:href"]
  887. ];
  888. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  889. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  890. );
  891. resourcePromises = resourcePromises.concat([
  892. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  893. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  894. ]);
  895. if (!this.options.removeAudioSrc) {
  896. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  897. }
  898. if (!this.options.removeVideoSrc) {
  899. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  900. }
  901. await Promise.all(resourcePromises);
  902. ProcessorHelper.processShortcutIcons(this.doc);
  903. }
  904. async processScripts() {
  905. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  906. let resourceURL;
  907. const scriptSrc = scriptElement.getAttribute("src");
  908. scriptElement.removeAttribute("src");
  909. scriptElement.textContent = "";
  910. try {
  911. resourceURL = docUtil.resolveURL(scriptSrc, this.baseURI);
  912. } catch (error) {
  913. // ignored
  914. }
  915. if (Util.testValidURL(resourceURL)) {
  916. this.stats.add("processed", "scripts", 1);
  917. const content = await docUtil.getContent(resourceURL, { asDataURI: true, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  918. scriptElement.setAttribute("src", content.data);
  919. }
  920. }));
  921. }
  922. removeAlternativeImages() {
  923. docUtil.removeAlternativeImages(this.doc);
  924. }
  925. removeAlternativeFonts() {
  926. docUtil.removeAlternativeFonts(this.doc, this.stylesheets);
  927. }
  928. async processFrames() {
  929. if (this.options.framesData) {
  930. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  931. await Promise.all(frameElements.map(async frameElement => {
  932. const frameWindowId = frameElement.getAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  933. if (frameWindowId) {
  934. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  935. if (frameData) {
  936. this.options.framesData = this.options.framesData.filter(frame => frame.windowId != frameWindowId);
  937. if (frameData.processor) {
  938. this.stats.add("processed", "frames", 1);
  939. await frameData.processor.run();
  940. const pageData = await frameData.processor.getPageData();
  941. frameElement.removeAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  942. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  943. frameElement.setAttribute("sandbox", "allow-popups allow-scripts allow-same-origin");
  944. } else {
  945. frameElement.setAttribute("sandbox", "allow-popups");
  946. }
  947. if (frameElement.tagName == "OBJECT") {
  948. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  949. } else {
  950. if (frameElement.tagName == "FRAME") {
  951. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/#/g, "%23"));
  952. } else {
  953. frameElement.setAttribute("srcdoc", pageData.content);
  954. frameElement.removeAttribute("src");
  955. }
  956. }
  957. this.stats.addAll(pageData);
  958. } else {
  959. this.stats.add("discarded", "frames", 1);
  960. }
  961. }
  962. }
  963. }));
  964. }
  965. }
  966. async processHtmlImports() {
  967. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  968. await Promise.all(linkElements.map(async linkElement => {
  969. const processor = this.relImportProcessors.get(linkElement);
  970. if (processor) {
  971. this.stats.add("processed", "HTML imports", 1);
  972. this.relImportProcessors.delete(linkElement);
  973. const pageData = await processor.getPageData();
  974. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  975. this.stats.addAll(pageData);
  976. } else {
  977. this.stats.add("discarded", "HTML imports", 1);
  978. }
  979. }));
  980. }
  981. replaceStylesheets() {
  982. this.doc.querySelectorAll("style").forEach(styleElement => {
  983. const stylesheetInfo = this.stylesheets.get(styleElement);
  984. if (stylesheetInfo) {
  985. this.stylesheets.delete(styleElement);
  986. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  987. styleElement.textContent = stylesheetContent;
  988. if (stylesheetInfo.mediaText) {
  989. styleElement.media = stylesheetInfo.mediaText;
  990. }
  991. } else {
  992. styleElement.remove();
  993. }
  994. });
  995. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  996. const stylesheetInfo = this.stylesheets.get(linkElement);
  997. if (stylesheetInfo) {
  998. this.stylesheets.delete(linkElement);
  999. const styleElement = this.doc.createElement("style");
  1000. if (stylesheetInfo.mediaText) {
  1001. styleElement.media = stylesheetInfo.mediaText;
  1002. }
  1003. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1004. styleElement.textContent = stylesheetContent;
  1005. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1006. } else {
  1007. linkElement.remove();
  1008. }
  1009. });
  1010. }
  1011. replaceStyleAttributes() {
  1012. this.doc.querySelectorAll("[style]").forEach(async element => {
  1013. const declarations = this.styles.get(element);
  1014. if (declarations) {
  1015. this.styles.delete(element);
  1016. let styleContent = cssTree.generate(declarations);
  1017. element.setAttribute("style", styleContent);
  1018. } else {
  1019. element.setAttribute("style", "");
  1020. }
  1021. });
  1022. }
  1023. insertVariables() {
  1024. if (this.cssVariables.size) {
  1025. const styleElement = this.doc.createElement("style");
  1026. const firstStyleElement = this.doc.head.querySelector("style");
  1027. if (firstStyleElement) {
  1028. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1029. } else {
  1030. this.doc.head.appendChild(styleElement);
  1031. }
  1032. let stylesheetContent = "";
  1033. this.cssVariables.forEach((content, indexResource) => {
  1034. this.cssVariables.delete(indexResource);
  1035. if (stylesheetContent) {
  1036. stylesheetContent += ";";
  1037. }
  1038. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1039. });
  1040. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1041. }
  1042. }
  1043. compressHTML() {
  1044. let size;
  1045. if (this.options.displayStats) {
  1046. size = docUtil.getContentSize(this.doc.documentElement.outerHTML);
  1047. }
  1048. docUtil.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: docUtil.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1049. if (this.options.displayStats) {
  1050. this.stats.add("discarded", "HTML bytes", size - docUtil.getContentSize(this.doc.documentElement.outerHTML));
  1051. }
  1052. }
  1053. }
  1054. // ---------------
  1055. // ProcessorHelper
  1056. // ---------------
  1057. const DATA_URI_PREFIX = "data:";
  1058. const ABOUT_BLANK_URI = "about:blank";
  1059. const EMPTY_DATA_URI = "data:base64,";
  1060. const REGEXP_AMP = /&/g;
  1061. const REGEXP_NBSP = /\u00a0/g;
  1062. const REGEXP_START_TAG = /</g;
  1063. const REGEXP_END_TAG = />/g;
  1064. const REGEXP_URL_HASH = /(#.+?)$/;
  1065. const PREFIX_DATA_URI_VND = "data:application/vnd.";
  1066. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1067. class ProcessorHelper {
  1068. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1069. const date = options.saveDate;
  1070. const url = docUtil.parseURL(options.saveUrl);
  1071. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash);
  1072. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash);
  1073. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash);
  1074. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash);
  1075. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash);
  1076. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash);
  1077. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash);
  1078. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash);
  1079. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash);
  1080. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash);
  1081. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash);
  1082. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash);
  1083. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1084. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash);
  1085. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash);
  1086. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash);
  1087. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash);
  1088. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1089. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash);
  1090. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash);
  1091. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash);
  1092. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash);
  1093. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash);
  1094. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash);
  1095. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash);
  1096. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash);
  1097. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash);
  1098. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash);
  1099. template = await Util.evalTemplateVariable(template, "url-href", () => decodeURI(url.href) || "No href", dontReplaceSlash);
  1100. template = await Util.evalTemplateVariable(template, "url-referrer", () => decodeURI(options.referrer) || "No referrer", dontReplaceSlash);
  1101. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash);
  1102. template = await Util.evalTemplateVariable(template, "url-pathname", () => decodeURI(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash);
  1103. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash);
  1104. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash);
  1105. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash);
  1106. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1107. for (const [name, value] of params) {
  1108. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash);
  1109. }
  1110. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1111. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash);
  1112. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash);
  1113. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash);
  1114. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decodeURI(Util.getLastSegment(url)) || "No last segment", dontReplaceSlash);
  1115. if (content) {
  1116. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => docUtil.digest("SHA-256", content), dontReplaceSlash);
  1117. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => docUtil.digest("SHA-384", content), dontReplaceSlash);
  1118. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => docUtil.digest("SHA-512", content), dontReplaceSlash);
  1119. }
  1120. return template.trim();
  1121. }
  1122. static setBackgroundImage(element, url, style) {
  1123. element.style.setProperty("background-blend-mode", "normal", "important");
  1124. element.style.setProperty("background-clip", "content-box", "important");
  1125. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1126. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1127. element.style.setProperty("background-image", url, "important");
  1128. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1129. element.style.setProperty("background-origin", "content-box", "important");
  1130. element.style.setProperty("background-repeat", "no-repeat", "important");
  1131. }
  1132. static processShortcutIcons(doc) {
  1133. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1134. if (!shortcutIcon) {
  1135. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1136. if (shortcutIcon) {
  1137. shortcutIcon.rel = "icon";
  1138. }
  1139. }
  1140. if (shortcutIcon) {
  1141. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1142. if (linkElement != shortcutIcon) {
  1143. linkElement.remove();
  1144. }
  1145. });
  1146. }
  1147. }
  1148. static removeSingleLineCssComments(stylesheet) {
  1149. const removedRules = [];
  1150. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1151. const ruleData = cssRule.data;
  1152. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1153. removedRules.push(cssRule);
  1154. }
  1155. }
  1156. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1157. }
  1158. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet) {
  1159. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1160. const imports = Util.getImportFunctions(stylesheetContent);
  1161. await Promise.all(imports.map(async cssImport => {
  1162. const match = Util.matchImport(cssImport);
  1163. if (match) {
  1164. let resourceURL = Util.normalizeURL(match.resourceURL);
  1165. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1166. try {
  1167. resourceURL = docUtil.resolveURL(match.resourceURL, baseURI);
  1168. } catch (error) {
  1169. // ignored
  1170. }
  1171. if (Util.testValidURL(resourceURL)) {
  1172. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  1173. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1174. resourceURL = content.resourceURL;
  1175. let importedStylesheetContent = Util.removeCssComments(content.data);
  1176. if (options.compressCSS) {
  1177. importedStylesheetContent = docUtil.compressCSS(importedStylesheetContent);
  1178. }
  1179. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1180. if (stylesheetContent.includes(cssImport)) {
  1181. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet);
  1182. workStylesheet.textContent = importedStylesheetContent;
  1183. if (workStylesheet.sheet.cssRules.length) {
  1184. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), importedStylesheetContent);
  1185. } else {
  1186. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), "");
  1187. }
  1188. }
  1189. }
  1190. }
  1191. }
  1192. }));
  1193. return stylesheetContent;
  1194. }
  1195. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1196. const urlFunctions = Util.getUrlFunctions(stylesheetContent);
  1197. urlFunctions.map(urlFunction => {
  1198. const originalResourceURL = Util.matchURL(urlFunction);
  1199. const resourceURL = Util.normalizeURL(originalResourceURL);
  1200. if (!Util.testIgnoredPath(resourceURL)) {
  1201. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1202. let resolvedURL;
  1203. try {
  1204. resolvedURL = docUtil.resolveURL(resourceURL, baseURI);
  1205. } catch (error) {
  1206. // ignored
  1207. }
  1208. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1209. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  1210. }
  1211. } else {
  1212. let newUrlFunction;
  1213. if (originalResourceURL) {
  1214. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1215. } else {
  1216. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1217. }
  1218. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1219. }
  1220. } else {
  1221. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  1222. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  1223. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  1224. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  1225. }
  1226. }
  1227. }
  1228. });
  1229. return stylesheetContent;
  1230. }
  1231. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1232. resourceURL = Util.normalizeURL(resourceURL);
  1233. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1234. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charset: options.charset };
  1235. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1236. resourceURL = content.resourceURL;
  1237. let stylesheetContent = Util.removeCssComments(content.data);
  1238. if (options.compressCSS) {
  1239. stylesheetContent = docUtil.compressCSS(stylesheetContent);
  1240. }
  1241. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1242. return stylesheetContent;
  1243. }
  1244. }
  1245. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1246. const promises = [];
  1247. const removedRules = [];
  1248. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1249. const ruleData = cssRule.data;
  1250. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1251. removedRules.push(cssRule);
  1252. } else if (ruleData.block && ruleData.block.children) {
  1253. if (ruleData.type == "Rule") {
  1254. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1255. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1256. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1257. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1258. promises.push(processFontFaceRule(ruleData));
  1259. }
  1260. }
  1261. }
  1262. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1263. await Promise.all(promises);
  1264. async function processFontFaceRule(ruleData) {
  1265. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1266. if (declaration.type == "Declaration" && declaration.value.children) {
  1267. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1268. await Promise.all(urlFunctions.map(async urlFunction => {
  1269. const originalResourceURL = Util.matchURL(urlFunction);
  1270. const resourceURL = Util.normalizeURL(originalResourceURL);
  1271. if (!Util.testIgnoredPath(resourceURL)) {
  1272. if (Util.testValidURL(resourceURL)) {
  1273. let { content } = await batchRequest.addURL(resourceURL, true);
  1274. if (content != EMPTY_DATA_URI && !content.startsWith(PREFIX_DATA_URI_VND) && !content.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1275. const validResource = await docUtil.validFont(urlFunction);
  1276. if (!validResource) {
  1277. content = EMPTY_DATA_URI;
  1278. }
  1279. }
  1280. replaceURLs(declaration, originalResourceURL, content);
  1281. }
  1282. } else if (resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(PREFIX_DATA_URI_VND) && !resourceURL.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1283. const validResource = await docUtil.validFont(urlFunction);
  1284. if (!validResource) {
  1285. replaceURLs(declaration, originalResourceURL, EMPTY_DATA_URI);
  1286. }
  1287. }
  1288. }));
  1289. }
  1290. }));
  1291. function replaceURLs(declaration, oldURL, newURL) {
  1292. declaration.value.children.forEach(token => {
  1293. if (token.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1294. token.value.value = newURL;
  1295. }
  1296. });
  1297. }
  1298. }
  1299. }
  1300. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1301. await Promise.all(declarations.map(async declaration => {
  1302. if (declaration.type == "Declaration" && declaration.value.children) {
  1303. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1304. await Promise.all(urlFunctions.map(async urlFunction => {
  1305. const originalResourceURL = Util.matchURL(urlFunction);
  1306. const resourceURL = Util.normalizeURL(originalResourceURL);
  1307. if (!Util.testIgnoredPath(resourceURL)) {
  1308. if (Util.testValidURL(resourceURL)) {
  1309. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1310. let variableDefined;
  1311. const tokens = [];
  1312. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1313. if (duplicate && options.groupDuplicateImages && rootFunction) {
  1314. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1315. tokens.push({ parent, token, value });
  1316. variableDefined = true;
  1317. } else {
  1318. token.data.value.value = content;
  1319. }
  1320. });
  1321. if (variableDefined) {
  1322. cssVariables.set(indexResource, content);
  1323. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1324. }
  1325. }
  1326. }
  1327. }));
  1328. }
  1329. }));
  1330. function findURLToken(url, children, callback, depth = 0) {
  1331. for (let token = children.head; token; token = token.next) {
  1332. if (token.data.children) {
  1333. findURLToken(url, token.data.children, callback, depth + 1);
  1334. }
  1335. if (token.data.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1336. callback(token, children, depth == 0);
  1337. }
  1338. }
  1339. }
  1340. }
  1341. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1342. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1343. let resourceURL = resourceElement.getAttribute(attributeName);
  1344. resourceURL = Util.normalizeURL(resourceURL);
  1345. if (!Util.testIgnoredPath(resourceURL)) {
  1346. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1347. if (Util.testValidPath(resourceURL)) {
  1348. try {
  1349. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1350. } catch (error) {
  1351. // ignored
  1352. }
  1353. if (Util.testValidURL(resourceURL)) {
  1354. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1355. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1356. resourceElement.remove();
  1357. } else {
  1358. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1359. if (!forbiddenPrefixFound) {
  1360. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1361. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1362. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1363. cssVariables.set(indexResource, content);
  1364. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1365. styles.set(resourceElement, declarationList);
  1366. } else {
  1367. resourceElement.setAttribute(attributeName, content);
  1368. }
  1369. } else {
  1370. resourceElement.setAttribute(attributeName, content);
  1371. }
  1372. }
  1373. }
  1374. }
  1375. }
  1376. }
  1377. }));
  1378. }
  1379. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1380. const attributeName = "xlink:href";
  1381. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1382. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1383. let resourceURL = Util.normalizeURL(originalResourceURL);
  1384. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1385. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1386. try {
  1387. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1388. } catch (error) {
  1389. // ignored
  1390. }
  1391. if (Util.testValidURL(resourceURL)) {
  1392. try {
  1393. const { content } = await batchRequest.addURL(resourceURL);
  1394. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1395. if (hashMatch && hashMatch[0]) {
  1396. const svgDoc = docUtil.parseSVGContent(content);
  1397. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1398. if (symbolElement) {
  1399. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1400. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1401. }
  1402. } else {
  1403. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1404. }
  1405. } catch (error) {
  1406. /* ignored */
  1407. }
  1408. }
  1409. } else if (resourceURL == options.url) {
  1410. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1411. }
  1412. }));
  1413. }
  1414. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1415. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1416. const srcset = docUtil.parseSrcset(resourceElement.getAttribute(attributeName));
  1417. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1418. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1419. if (!Util.testIgnoredPath(resourceURL)) {
  1420. if (Util.testValidPath(resourceURL)) {
  1421. try {
  1422. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1423. } catch (error) {
  1424. // ignored
  1425. }
  1426. if (Util.testValidURL(resourceURL)) {
  1427. const { content } = await batchRequest.addURL(resourceURL, true);
  1428. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1429. if (forbiddenPrefixFound) {
  1430. return "";
  1431. }
  1432. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1433. } else {
  1434. return "";
  1435. }
  1436. } else {
  1437. return "";
  1438. }
  1439. } else {
  1440. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1441. }
  1442. }));
  1443. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1444. }));
  1445. }
  1446. static replaceImageSource(imgElement, variableName, options) {
  1447. const dataAttributeName = docUtil.IMAGE_ATTRIBUTE_NAME;
  1448. if (imgElement.getAttribute(dataAttributeName) != null) {
  1449. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1450. if (imgData.replaceable) {
  1451. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imgData.size.pxWidth}" height="${imgData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1452. const backgroundStyle = {};
  1453. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1454. if (backgroundSize) {
  1455. backgroundStyle["background-size"] = imgData.objectFit;
  1456. }
  1457. if (imgData.objectPosition) {
  1458. backgroundStyle["background-position"] = imgData.objectPosition;
  1459. }
  1460. if (imgData.backgroundColor) {
  1461. backgroundStyle["background-color"] = imgData.backgroundColor;
  1462. }
  1463. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1464. imgElement.removeAttribute(dataAttributeName);
  1465. return true;
  1466. }
  1467. }
  1468. }
  1469. }
  1470. // ----
  1471. // Util
  1472. // ----
  1473. const BLOB_URI_PREFIX = "blob:";
  1474. const HTTP_URI_PREFIX = /^https?:\/\//;
  1475. const FILE_URI_PREFIX = /^file:\/\//;
  1476. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1477. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1478. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1479. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1480. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1481. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1482. 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;
  1483. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1484. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1485. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1486. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1487. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1488. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1489. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1490. class Util {
  1491. static normalizeURL(url) {
  1492. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1493. return url;
  1494. } else {
  1495. return url.split("#")[0];
  1496. }
  1497. }
  1498. static getCSSValue(value) {
  1499. let result = "";
  1500. try {
  1501. result = cssTree.generate(value);
  1502. } catch (error) {
  1503. // ignored
  1504. }
  1505. return result;
  1506. }
  1507. static getOnEventAttributeNames(doc) {
  1508. const element = doc.createElement("div");
  1509. const attributeNames = [];
  1510. for (const propertyName in element) {
  1511. if (propertyName.startsWith("on")) {
  1512. attributeNames.push(propertyName);
  1513. }
  1514. }
  1515. return attributeNames;
  1516. }
  1517. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1518. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1519. if (template && template.match(replaceRegExp)) {
  1520. let value = await valueGetter();
  1521. if (!dontReplaceSlash) {
  1522. value = value.replace(/\/+/g, "_");
  1523. }
  1524. return template.replace(replaceRegExp, value);
  1525. }
  1526. return template;
  1527. }
  1528. static getLastSegment(url) {
  1529. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1530. if (!lastSegment) {
  1531. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1532. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1533. }
  1534. if (!lastSegment) {
  1535. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1536. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1537. }
  1538. if (!lastSegment) {
  1539. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1540. }
  1541. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1542. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1543. lastSegment = lastSegmentMatch[1];
  1544. }
  1545. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1546. return lastSegment;
  1547. }
  1548. static getRegExp(string) {
  1549. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1550. }
  1551. static getUrlFunctions(stylesheetContent) {
  1552. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1553. }
  1554. static getImportFunctions(stylesheetContent) {
  1555. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1556. }
  1557. static findShortcutIcon(shortcutIcons) {
  1558. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1559. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1560. return shortcutIcons[0];
  1561. }
  1562. static matchURL(stylesheetContent) {
  1563. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1564. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1565. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1566. return match && match[1];
  1567. }
  1568. static testIgnoredPath(resourceURL) {
  1569. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1570. }
  1571. static testValidPath(resourceURL) {
  1572. return resourceURL && !resourceURL.match(EMPTY_URL);
  1573. }
  1574. static testValidURL(resourceURL) {
  1575. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1576. }
  1577. static matchImport(stylesheetContent) {
  1578. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1579. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1580. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1581. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1582. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1583. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1584. if (match) {
  1585. const [, resourceURL, media] = match;
  1586. return { resourceURL, media };
  1587. }
  1588. }
  1589. static removeCssComments(stylesheetContent) {
  1590. let start, end;
  1591. do {
  1592. start = stylesheetContent.indexOf("/*");
  1593. end = stylesheetContent.indexOf("*/", start + 2);
  1594. if (start != -1 && end != -1) {
  1595. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1596. }
  1597. } while (start != -1 && end != -1);
  1598. return stylesheetContent;
  1599. }
  1600. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1601. if (mediaQuery) {
  1602. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1603. } else {
  1604. return stylesheetContent;
  1605. }
  1606. }
  1607. }
  1608. function log(...args) {
  1609. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1610. }
  1611. // -----
  1612. // Stats
  1613. // -----
  1614. const STATS_DEFAULT_VALUES = {
  1615. discarded: {
  1616. "HTML bytes": 0,
  1617. "hidden elements": 0,
  1618. "HTML imports": 0,
  1619. scripts: 0,
  1620. objects: 0,
  1621. "audio sources": 0,
  1622. "video sources": 0,
  1623. frames: 0,
  1624. "CSS rules": 0,
  1625. canvas: 0,
  1626. stylesheets: 0,
  1627. resources: 0,
  1628. medias: 0
  1629. },
  1630. processed: {
  1631. "HTML bytes": 0,
  1632. "hidden elements": 0,
  1633. "HTML imports": 0,
  1634. scripts: 0,
  1635. objects: 0,
  1636. "audio sources": 0,
  1637. "video sources": 0,
  1638. frames: 0,
  1639. "CSS rules": 0,
  1640. canvas: 0,
  1641. stylesheets: 0,
  1642. resources: 0,
  1643. medias: 0
  1644. }
  1645. };
  1646. class Stats {
  1647. constructor(options) {
  1648. this.options = options;
  1649. if (options.displayStats) {
  1650. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1651. }
  1652. }
  1653. set(type, subType, value) {
  1654. if (this.options.displayStats) {
  1655. this.data[type][subType] = value;
  1656. }
  1657. }
  1658. add(type, subType, value) {
  1659. if (this.options.displayStats) {
  1660. this.data[type][subType] += value;
  1661. }
  1662. }
  1663. addAll(pageData) {
  1664. if (this.options.displayStats) {
  1665. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1666. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1667. }
  1668. }
  1669. }
  1670. return { getClass };
  1671. })();