single-file-core.js 83 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019
  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.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree, sessionId = 0;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. options.insertFaviconLink = true;
  34. if (options.sessionId === undefined) {
  35. options.sessionId = sessionId;
  36. sessionId++;
  37. }
  38. }
  39. async run() {
  40. this.runner = new Runner(this.options, true);
  41. await this.runner.loadPage();
  42. await this.runner.initialize();
  43. await this.runner.run();
  44. }
  45. cancel() {
  46. this.cancelled = true;
  47. if (this.runner) {
  48. this.runner.cancel();
  49. }
  50. }
  51. async getPageData() {
  52. return this.runner.getPageData();
  53. }
  54. }
  55. // -------------
  56. // ProgressEvent
  57. // -------------
  58. const PAGE_LOADING = "page-loading";
  59. const PAGE_LOADED = "page-loaded";
  60. const RESOURCES_INITIALIZING = "resource-initializing";
  61. const RESOURCES_INITIALIZED = "resources-initialized";
  62. const RESOURCE_LOADED = "resource-loaded";
  63. const PAGE_ENDED = "page-ended";
  64. const STAGE_STARTED = "stage-started";
  65. const STAGE_ENDED = "stage-ended";
  66. const STAGE_TASK_STARTED = "stage-task-started";
  67. const STAGE_TASK_ENDED = "stage-task-ended";
  68. class ProgressEvent {
  69. constructor(type, detail) {
  70. 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 };
  71. }
  72. }
  73. // ------
  74. // Runner
  75. // ------
  76. const RESOLVE_URLS_STAGE = 0;
  77. const REPLACE_DATA_STAGE = 1;
  78. const REPLACE_DOCS_STAGE = 2;
  79. const POST_PROCESS_STAGE = 3;
  80. const STAGES = [{
  81. sequential: [
  82. { action: "preProcessPage" },
  83. { action: "replaceStyleContents" },
  84. { action: "resetCharsetMeta" },
  85. { option: "insertFaviconLink", action: "insertFaviconLink" },
  86. { action: "replaceCanvasElements" },
  87. { action: "insertFonts" },
  88. { action: "insertShadowRootContents" },
  89. { action: "setInputValues" },
  90. { option: "removeScripts", action: "removeScripts" },
  91. { option: "selected", action: "removeUnselectedElements" },
  92. { option: "removeVideoSrc", action: "insertVideoPosters" },
  93. { option: "removeFrames", action: "removeFrames" },
  94. { option: "removeVideoSrc", action: "removeVideoSources" },
  95. { option: "removeAudioSrc", action: "removeAudioSources" },
  96. { action: "removeDiscardedResources" },
  97. { option: "removeHiddenElements", action: "removeHiddenElements" },
  98. { action: "resolveHrefs" },
  99. { action: "resolveStyleAttributeURLs" }
  100. ],
  101. parallel: [
  102. { action: "resolveStylesheetURLs" },
  103. { option: "!removeFrames", action: "resolveFrameURLs" },
  104. { action: "resolveHtmlImportURLs" }
  105. ]
  106. }, {
  107. sequential: [
  108. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  109. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  110. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  111. ],
  112. parallel: [
  113. { action: "processStylesheets" },
  114. { action: "processStyleAttributes" },
  115. { action: "processPageResources" },
  116. { option: "!removeScripts", action: "processScripts" }
  117. ]
  118. }, {
  119. sequential: [
  120. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  121. ],
  122. parallel: [
  123. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  124. { option: "!removeFrames", action: "processFrames" },
  125. { option: "!removeImports", action: "processHtmlImports" },
  126. ]
  127. }, {
  128. sequential: [
  129. { action: "replaceStylesheets" },
  130. { action: "replaceStyleAttributes" },
  131. { action: "insertVariables" },
  132. { option: "compressHTML", action: "compressHTML" },
  133. { action: "cleanupPage" }
  134. ],
  135. parallel: [
  136. { option: "enableMaff", action: "insertMAFFMetaData" },
  137. { action: "setDocInfo" }
  138. ]
  139. }];
  140. class Runner {
  141. constructor(options, root) {
  142. this.root = root;
  143. this.options = options;
  144. this.options.url = this.options.url || (this.options.doc && this.options.doc.location.href);
  145. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  146. this.options.rootDocument = root;
  147. this.options.updatedResources = this.options.updatedResources || {};
  148. this.batchRequest = new BatchRequest();
  149. this.processor = new Processor(options, this.batchRequest);
  150. if (this.options.doc) {
  151. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  152. this.options.canvases = docData.canvases;
  153. this.options.fonts = docData.fonts;
  154. this.options.stylesheets = docData.stylesheets;
  155. this.options.images = docData.images;
  156. this.options.posters = docData.posters;
  157. this.options.usedFonts = docData.usedFonts;
  158. this.options.shadowRoots = docData.shadowRoots;
  159. this.options.imports = docData.imports;
  160. this.options.referrer = docData.referrer;
  161. }
  162. if (this.options.saveRawPage) {
  163. this.options.removeFrames = true;
  164. }
  165. this.options.content = this.options.content || (this.options.doc ? util.serialize(this.options.doc) : null);
  166. this.onprogress = options.onprogress || (() => { });
  167. }
  168. async loadPage() {
  169. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  170. await this.processor.loadPage(this.options.content);
  171. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  172. }
  173. async initialize() {
  174. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  175. await this.executeStage(RESOLVE_URLS_STAGE);
  176. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  177. if (this.options.doc) {
  178. util.postProcessDoc(this.options.doc, this.markedElements);
  179. }
  180. this.options.doc = null;
  181. this.options.win = null;
  182. }
  183. cancel() {
  184. this.cancelled = true;
  185. this.batchRequest.cancel();
  186. if (this.root) {
  187. if (this.options.frames) {
  188. this.options.frames.forEach(cancelRunner);
  189. }
  190. if (this.options.imports) {
  191. this.options.imports.forEach(cancelRunner);
  192. }
  193. }
  194. function cancelRunner(resourceData) {
  195. if (resourceData.runner) {
  196. resourceData.runner.cancel();
  197. }
  198. }
  199. }
  200. async run() {
  201. if (this.root) {
  202. this.processor.initialize(this.batchRequest);
  203. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  204. }
  205. await this.batchRequest.run(detail => {
  206. detail.pageURL = this.options.url;
  207. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  208. }, this.options);
  209. await this.pendingPromises;
  210. await this.executeStage(REPLACE_DOCS_STAGE);
  211. await this.executeStage(POST_PROCESS_STAGE);
  212. this.processor.finalize();
  213. }
  214. getDocument() {
  215. return this.processor.doc;
  216. }
  217. getStyleSheets() {
  218. return this.processor.stylesheets;
  219. }
  220. async getPageData() {
  221. if (this.root) {
  222. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  223. }
  224. return this.processor.getPageData();
  225. }
  226. async executeStage(step) {
  227. if (DEBUG) {
  228. log("**** STARTED STAGE", step, "****");
  229. }
  230. const frame = !this.root;
  231. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  232. STAGES[step].sequential.forEach(task => {
  233. let startTime;
  234. if (DEBUG) {
  235. startTime = Date.now();
  236. log(" -- STARTED task =", task.action);
  237. }
  238. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  239. if (!this.cancelled) {
  240. this.executeTask(task);
  241. }
  242. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  243. if (DEBUG) {
  244. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  245. }
  246. });
  247. let parallelTasksPromise;
  248. if (STAGES[step].parallel) {
  249. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  250. let startTime;
  251. if (DEBUG) {
  252. startTime = Date.now();
  253. log(" // STARTED task =", task.action);
  254. }
  255. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  256. if (!this.cancelled) {
  257. await this.executeTask(task);
  258. }
  259. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  260. if (DEBUG) {
  261. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  262. }
  263. }));
  264. } else {
  265. parallelTasksPromise = Promise.resolve();
  266. }
  267. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  268. if (DEBUG) {
  269. log("**** ENDED STAGE", step, "****");
  270. }
  271. return parallelTasksPromise;
  272. }
  273. executeTask(task) {
  274. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  275. return this.processor[task.action]();
  276. }
  277. }
  278. }
  279. // ------------
  280. // BatchRequest
  281. // ------------
  282. class BatchRequest {
  283. constructor() {
  284. this.requests = new Map();
  285. this.duplicates = new Map();
  286. }
  287. async addURL(resourceURL, asBinary, groupDuplicates) {
  288. return new Promise((resolve, reject) => {
  289. const requestKey = JSON.stringify([resourceURL, asBinary]);
  290. let resourceRequests = this.requests.get(requestKey);
  291. if (!resourceRequests) {
  292. resourceRequests = [];
  293. this.requests.set(requestKey, resourceRequests);
  294. }
  295. const callbacks = { resolve, reject };
  296. resourceRequests.push(callbacks);
  297. if (groupDuplicates) {
  298. let duplicateRequests = this.duplicates.get(requestKey);
  299. if (!duplicateRequests) {
  300. duplicateRequests = [];
  301. this.duplicates.set(requestKey, duplicateRequests);
  302. }
  303. duplicateRequests.push(callbacks);
  304. }
  305. });
  306. }
  307. getMaxResources() {
  308. return this.requests.size;
  309. }
  310. async run(onloadListener, options) {
  311. const resourceURLs = [...this.requests.keys()];
  312. let indexResource = 0;
  313. return Promise.all(resourceURLs.map(async requestKey => {
  314. const [resourceURL, asBinary] = JSON.parse(requestKey);
  315. const resourceRequests = this.requests.get(requestKey);
  316. try {
  317. const content = await util.getContent(resourceURL, {
  318. asBinary,
  319. maxResourceSize: options.maxResourceSize,
  320. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  321. frameId: options.windowId
  322. });
  323. indexResource = indexResource + 1;
  324. onloadListener({ url: resourceURL });
  325. if (!this.cancelled) {
  326. resourceRequests.forEach(callbacks => {
  327. const duplicateCallbacks = this.duplicates.get(requestKey);
  328. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  329. callbacks.resolve({ content: content.data, indexResource, duplicate });
  330. });
  331. }
  332. } catch (error) {
  333. indexResource = indexResource + 1;
  334. onloadListener({ url: resourceURL });
  335. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  336. }
  337. this.requests.delete(requestKey);
  338. }));
  339. }
  340. cancel() {
  341. this.cancelled = true;
  342. const resourceURLs = [...this.requests.keys()];
  343. resourceURLs.forEach(requestKey => {
  344. const resourceRequests = this.requests.get(requestKey);
  345. resourceRequests.forEach(callbacks => callbacks.reject());
  346. this.requests.delete(requestKey);
  347. });
  348. }
  349. }
  350. // ---------
  351. // Processor
  352. // ---------
  353. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  354. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  355. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  356. const SCRIPT_TAG_FOUND = /<script/gi;
  357. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  358. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  359. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  360. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  361. class Processor {
  362. constructor(options, batchRequest) {
  363. this.options = options;
  364. this.stats = new Stats(options);
  365. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  366. this.batchRequest = batchRequest;
  367. this.stylesheets = new Map();
  368. this.styles = new Map();
  369. this.cssVariables = new Map();
  370. }
  371. initialize() {
  372. this.options.saveDate = new Date();
  373. this.options.saveUrl = this.options.url;
  374. if (this.options.enableMaff) {
  375. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  376. }
  377. this.maxResources = this.batchRequest.getMaxResources();
  378. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  379. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  380. }
  381. if (!this.options.removeImports && this.options.imports) {
  382. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  383. }
  384. this.stats.set("processed", "resources", this.maxResources);
  385. }
  386. async loadPage(pageContent, charset) {
  387. let content;
  388. if (!pageContent || this.options.saveRawPage) {
  389. content = await util.getContent(this.baseURI, {
  390. maxResourceSize: this.options.maxResourceSize,
  391. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  392. charset,
  393. frameId: this.options.windowId
  394. });
  395. pageContent = content.data;
  396. }
  397. this.doc = util.parseDocContent(pageContent, this.baseURI);
  398. if (this.options.saveRawPage) {
  399. let charset;
  400. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  401. const charsetDeclaration = element.content.split(";")[1];
  402. if (charsetDeclaration && !charset) {
  403. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  404. }
  405. });
  406. if (charset && charset != content.charset) {
  407. return this.loadPage(pageContent, charset);
  408. }
  409. }
  410. this.workStyleElement = this.doc.createElement("style");
  411. this.doc.body.appendChild(this.workStyleElement);
  412. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  413. }
  414. finalize() {
  415. if (this.workStyleElement.parentNode) {
  416. this.workStyleElement.remove();
  417. }
  418. }
  419. async getPageData() {
  420. util.postProcessDoc(this.doc);
  421. const url = util.parseURL(this.baseURI);
  422. if (this.options.insertSingleFileComment) {
  423. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  424. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  425. " \n url: " + this.options.saveUrl +
  426. " \n saved date: " + this.options.saveDate +
  427. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  428. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  429. }
  430. let size;
  431. if (this.options.displayStats) {
  432. size = util.getContentSize(this.doc.documentElement.outerHTML);
  433. }
  434. const content = util.serialize(this.doc, this.options.compressHTML);
  435. if (this.options.displayStats) {
  436. const contentSize = util.getContentSize(content);
  437. this.stats.set("processed", "HTML bytes", contentSize);
  438. this.stats.add("discarded", "HTML bytes", size - contentSize);
  439. }
  440. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  441. const replacementCharacter = this.options.filenameReplacementCharacter;
  442. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  443. if (!this.options.backgroundSave) {
  444. filename = filename.replace(/\//g, replacementCharacter);
  445. }
  446. if (util.getContentSize(filename) > this.options.filenameMaxLength) {
  447. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  448. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  449. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  450. filename = filename + "…" + extension;
  451. }
  452. if (!filename) {
  453. filename = "Unnamed page";
  454. }
  455. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  456. return {
  457. stats: this.stats.data,
  458. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  459. filename,
  460. content
  461. };
  462. }
  463. preProcessPage() {
  464. if (this.options.win) {
  465. 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));
  466. }
  467. if (this.options.images && !this.options.saveRawPage) {
  468. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  469. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  470. if (attributeValue) {
  471. const imageData = this.options.images[Number(attributeValue)];
  472. if (imageData) {
  473. if (this.options.removeHiddenElements && imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) {
  474. imgElement.setAttribute("src", EMPTY_IMAGE);
  475. } else if (imageData.currentSrc) {
  476. imgElement.setAttribute("src", imageData.currentSrc);
  477. }
  478. if (this.options.loadDeferredImages) {
  479. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  480. imageData.src = imgElement.dataset.src;
  481. imgElement.setAttribute("src", imgElement.dataset.src);
  482. imgElement.removeAttribute("data-src");
  483. }
  484. }
  485. }
  486. }
  487. });
  488. if (this.options.loadDeferredImages) {
  489. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  490. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  491. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  492. imgElement.removeAttribute("data-srcset");
  493. }
  494. });
  495. }
  496. }
  497. }
  498. replaceStyleContents() {
  499. if (this.options.stylesheets) {
  500. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  501. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  502. if (attributeValue) {
  503. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  504. if (stylesheetContent) {
  505. styleElement.textContent = stylesheetContent;
  506. }
  507. }
  508. });
  509. }
  510. }
  511. removeUnselectedElements() {
  512. removeUnmarkedElements(this.doc.body);
  513. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  514. function removeUnmarkedElements(element) {
  515. let selectedElementFound = false;
  516. Array.from(element.childNodes).forEach(node => {
  517. if (node.nodeType == 1) {
  518. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  519. selectedElementFound = selectedElementFound || isSelectedElement;
  520. if (isSelectedElement) {
  521. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  522. removeUnmarkedElements(node);
  523. } else if (selectedElementFound) {
  524. removeNode(node);
  525. } else {
  526. hideNode(node);
  527. }
  528. }
  529. });
  530. }
  531. function removeNode(node) {
  532. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  533. node.remove();
  534. } else {
  535. hideNode(node);
  536. }
  537. }
  538. function hideNode(node) {
  539. if (canHideNode(node)) {
  540. node.style.setProperty("display", "none", "important");
  541. Array.from(node.childNodes).forEach(removeNode);
  542. }
  543. }
  544. function canHideNode(node) {
  545. if (node.nodeType == 1) {
  546. const tagName = node.tagName && node.tagName.toLowerCase();
  547. return (tagName != "svg" && tagName != "style" && tagName != "link");
  548. }
  549. }
  550. }
  551. insertVideoPosters() {
  552. if (this.options.posters) {
  553. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  554. let videoElement;
  555. if (element.tagName == "VIDEO") {
  556. videoElement = element;
  557. } else {
  558. videoElement = element.parentElement;
  559. }
  560. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  561. if (attributeValue) {
  562. const posterURL = this.options.posters[Number(attributeValue)];
  563. if (!videoElement.poster && posterURL) {
  564. videoElement.setAttribute("poster", posterURL);
  565. }
  566. }
  567. });
  568. }
  569. }
  570. removeFrames() {
  571. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  572. this.stats.set("discarded", "frames", frameElements.length);
  573. this.stats.set("processed", "frames", frameElements.length);
  574. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  575. }
  576. removeScripts() {
  577. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  578. this.doc.querySelectorAll("[href]").forEach(element => {
  579. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  580. element.setAttribute("href", "");
  581. }
  582. });
  583. this.doc.querySelectorAll("[src]").forEach(element => {
  584. if (element.src && element.src.match(/^\s*javascript:/)) {
  585. element.removeAttribute("src");
  586. }
  587. });
  588. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  589. this.stats.set("discarded", "scripts", scriptElements.length);
  590. this.stats.set("processed", "scripts", scriptElements.length);
  591. scriptElements.forEach(element => element.remove());
  592. }
  593. removeVideoSources() {
  594. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  595. this.stats.set("discarded", "video sources", videoSourceElements.length);
  596. this.stats.set("processed", "video sources", videoSourceElements.length);
  597. videoSourceElements.forEach(element => {
  598. if (element.tagName == "SOURCE") {
  599. element.remove();
  600. } else {
  601. videoSourceElements.forEach(element => element.removeAttribute("src"));
  602. }
  603. });
  604. }
  605. removeAudioSources() {
  606. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  607. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  608. this.stats.set("processed", "audio sources", audioSourceElements.length);
  609. audioSourceElements.forEach(element => {
  610. if (element.tagName == "SOURCE") {
  611. element.remove();
  612. } else {
  613. audioSourceElements.forEach(element => element.removeAttribute("src"));
  614. }
  615. });
  616. }
  617. removeDiscardedResources() {
  618. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  619. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  620. const objectElements = this.doc.querySelectorAll("applet, object[data]:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed[src]:not([src*=\".svg\"]):not([src*=\".pdf\"])");
  621. this.stats.set("discarded", "objects", objectElements.length);
  622. this.stats.set("processed", "objects", objectElements.length);
  623. objectElements.forEach(element => element.remove());
  624. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  625. replacedAttributeValue.forEach(element => {
  626. let regExp;
  627. if (this.options.removeScripts) {
  628. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  629. } else {
  630. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  631. }
  632. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  633. if (relValue.length) {
  634. element.setAttribute("rel", relValue);
  635. } else {
  636. element.remove();
  637. }
  638. });
  639. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  640. if (this.options.compressHTML) {
  641. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  642. }
  643. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  644. }
  645. resetCharsetMeta() {
  646. let charset;
  647. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  648. const charsetDeclaration = element.content.split(";")[1];
  649. if (charsetDeclaration && !charset) {
  650. charset = charsetDeclaration.split("=")[1];
  651. if (charset) {
  652. this.charset = charset.trim().toLowerCase();
  653. }
  654. }
  655. element.remove();
  656. });
  657. const metaElement = this.doc.createElement("meta");
  658. metaElement.setAttribute("charset", "utf-8");
  659. if (this.doc.head.firstChild) {
  660. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  661. } else {
  662. this.doc.head.appendChild(metaElement);
  663. }
  664. }
  665. setInputValues() {
  666. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  667. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  668. input.setAttribute("value", value || "");
  669. });
  670. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  671. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  672. if (value == "true") {
  673. input.setAttribute("checked", "");
  674. }
  675. });
  676. this.doc.querySelectorAll("textarea").forEach(textarea => {
  677. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  678. textarea.textContent = value || "";
  679. });
  680. this.doc.querySelectorAll("select").forEach(select => {
  681. select.querySelectorAll("option").forEach(option => {
  682. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  683. if (selected) {
  684. option.setAttribute("selected", "");
  685. }
  686. });
  687. });
  688. }
  689. insertFaviconLink() {
  690. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  691. if (!faviconElement) {
  692. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  693. }
  694. if (!faviconElement) {
  695. faviconElement = this.doc.createElement("link");
  696. faviconElement.setAttribute("type", "image/x-icon");
  697. faviconElement.setAttribute("rel", "shortcut icon");
  698. faviconElement.setAttribute("href", "/favicon.ico");
  699. }
  700. this.doc.head.appendChild(faviconElement);
  701. }
  702. replaceCanvasElements() {
  703. if (this.options.canvases) {
  704. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  705. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  706. if (attributeValue) {
  707. const canvasData = this.options.canvases[Number(attributeValue)];
  708. if (canvasData) {
  709. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  710. this.stats.add("processed", "canvas", 1);
  711. }
  712. }
  713. });
  714. }
  715. }
  716. insertFonts() {
  717. if (this.options.fonts && this.options.fonts.length) {
  718. let stylesheetContent = "";
  719. this.options.fonts.forEach(fontData => {
  720. if (fontData["font-family"] && fontData.src) {
  721. stylesheetContent += "@font-face{";
  722. let stylesContent = "";
  723. Object.keys(fontData).forEach(fontStyle => {
  724. if (stylesContent) {
  725. stylesContent += ";";
  726. }
  727. stylesContent += fontStyle + ":" + fontData[fontStyle];
  728. });
  729. stylesheetContent += stylesContent + "}";
  730. }
  731. });
  732. if (stylesheetContent) {
  733. const styleElement = this.doc.createElement("style");
  734. styleElement.textContent = stylesheetContent;
  735. const existingStyleElement = this.doc.querySelector("style");
  736. if (existingStyleElement) {
  737. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  738. } else {
  739. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  740. }
  741. }
  742. }
  743. }
  744. removeHiddenElements() {
  745. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  746. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  747. this.stats.set("discarded", "hidden elements", removedElements.length);
  748. this.stats.set("processed", "hidden elements", removedElements.length);
  749. if (hiddenElements.length) {
  750. const styleElement = this.doc.createElement("style");
  751. styleElement.textContent = ".sf-hidden{display:none!important;}";
  752. this.doc.head.appendChild(styleElement);
  753. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  754. }
  755. removedElements.forEach(element => element.remove());
  756. }
  757. resolveHrefs() {
  758. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  759. const href = element.getAttribute("href").trim();
  760. if (!Util.testIgnoredPath(href)) {
  761. let resolvedURL;
  762. try {
  763. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  764. } catch (error) {
  765. // ignored
  766. }
  767. if (resolvedURL) {
  768. const url = Util.normalizeURL(this.options.url);
  769. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  770. resolvedURL = resolvedURL.substring(url.length);
  771. }
  772. try {
  773. element.setAttribute("href", resolvedURL);
  774. } catch (error) {
  775. // ignored
  776. }
  777. }
  778. }
  779. });
  780. }
  781. resolveStyleAttributeURLs() {
  782. this.doc.querySelectorAll("[style]").forEach(element => {
  783. let styleContent = element.getAttribute("style");
  784. if (this.options.compressCSS) {
  785. styleContent = util.compressCSS(styleContent);
  786. }
  787. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  788. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  789. this.styles.set(element, declarationList);
  790. });
  791. }
  792. async resolveStylesheetURLs() {
  793. const options = {
  794. maxResourceSize: this.options.maxResourceSize,
  795. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  796. url: this.options.url,
  797. charset: this.charset,
  798. compressCSS: this.options.compressCSS,
  799. updatedResources: this.options.updatedResources,
  800. rootDocument: this.options.rootDocument,
  801. frameId: this.options.windowId
  802. };
  803. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  804. .map(async element => {
  805. let mediaText;
  806. if (element.media) {
  807. mediaText = element.media.toLowerCase();
  808. }
  809. const stylesheetInfo = { mediaText };
  810. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  811. stylesheetInfo.scoped = true;
  812. }
  813. if (element.tagName.toLowerCase() == "link") {
  814. if (element.charset) {
  815. options.charset = element.charset;
  816. }
  817. }
  818. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  819. }));
  820. if (options.rootDocument) {
  821. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  822. await Promise.all(newResources.map(async resource => {
  823. resource.retrieved = true;
  824. const stylesheetInfo = {};
  825. const element = this.doc.createElement("style");
  826. this.doc.body.appendChild(element);
  827. element.textContent = resource.content;
  828. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  829. }));
  830. }
  831. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  832. stylesheets.set(element, stylesheetInfo);
  833. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  834. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  835. if (match && match[1] && match[1] != options.charset) {
  836. options.charset = match[1];
  837. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  838. }
  839. let stylesheet;
  840. try {
  841. stylesheet = cssTree.parse(Util.removeCssComments(stylesheetContent));
  842. } catch (error) {
  843. // ignored
  844. }
  845. if (stylesheet && stylesheet.children) {
  846. if (options.compressCSS) {
  847. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  848. }
  849. stylesheetInfo.stylesheet = stylesheet;
  850. } else {
  851. stylesheets.delete(element);
  852. }
  853. }
  854. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  855. let content;
  856. if (element.tagName.toLowerCase() == "link") {
  857. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  858. } else {
  859. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  860. }
  861. return content || "";
  862. }
  863. }
  864. async resolveFrameURLs() {
  865. if (!this.options.saveRawPage && this.options.frames) {
  866. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  867. await Promise.all(frameElements.map(async frameElement => {
  868. if (frameElement.tagName == "OBJECT") {
  869. frameElement.setAttribute("data", "data:text/html,");
  870. } else {
  871. frameElement.removeAttribute("src");
  872. frameElement.removeAttribute("srcdoc");
  873. }
  874. Array.from(frameElement.childNodes).forEach(node => node.remove());
  875. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  876. if (frameWindowId) {
  877. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  878. if (frameData) {
  879. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  880. }
  881. }
  882. }));
  883. }
  884. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  885. options.insertSingleFileComment = false;
  886. options.insertFaviconLink = false;
  887. options.doc = null;
  888. options.win = null;
  889. options.url = frameData.baseURI;
  890. options.windowId = frameWindowId;
  891. if (frameData.content) {
  892. options.content = frameData.content;
  893. options.canvases = frameData.canvases;
  894. options.fonts = frameData.fonts;
  895. options.stylesheets = frameData.stylesheets;
  896. options.images = frameData.images;
  897. options.posters = frameData.posters;
  898. options.usedFonts = frameData.usedFonts;
  899. options.shadowRoots = frameData.shadowRoots;
  900. options.imports = frameData.imports;
  901. frameData.runner = new Runner(options);
  902. frameData.frameElement = frameElement;
  903. await frameData.runner.loadPage();
  904. await frameData.runner.initialize();
  905. frameData.maxResources = batchRequest.getMaxResources();
  906. }
  907. }
  908. }
  909. insertShadowRootContents() {
  910. const doc = this.doc;
  911. const options = this.options;
  912. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  913. processElement(this.doc);
  914. if (this.options.removeScripts) {
  915. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  916. }
  917. const scriptElement = doc.createElement("script");
  918. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  919. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{if (!element.parentElement.shadowRoot) {const shadowRoot=element.parentElement.attachShadow({mode:element.getAttribute("${SHADOW_MODE_ATTRIBUTE_NAME}"),delegatesFocus:Boolean(element.getAttribute("${SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME}"))});shadowRoot.innerHTML=element.innerHTML;element.remove();processNode(shadowRoot)}})}})()`;
  920. doc.body.appendChild(scriptElement);
  921. }
  922. function processElement(element) {
  923. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  924. shadowRootElements.forEach(element => {
  925. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  926. if (attributeValue) {
  927. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  928. if (shadowRootData) {
  929. const templateElement = doc.createElement("template");
  930. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  931. if (shadowRootData.delegatesFocus) {
  932. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  933. }
  934. if (shadowRootData.adoptedStyleSheets) {
  935. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  936. const styleElement = doc.createElement("style");
  937. styleElement.textContent = stylesheetContent;
  938. templateElement.appendChild(styleElement);
  939. });
  940. }
  941. const shadowDoc = util.parseDocContent(shadowRootData.content);
  942. if (shadowDoc.head) {
  943. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  944. if (metaCharset) {
  945. metaCharset.remove();
  946. }
  947. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  948. }
  949. if (shadowDoc.body) {
  950. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  951. }
  952. processElement(templateElement);
  953. if (element.firstChild) {
  954. element.insertBefore(templateElement, element.firstChild);
  955. } else {
  956. element.appendChild(templateElement);
  957. }
  958. }
  959. }
  960. });
  961. }
  962. }
  963. async resolveHtmlImportURLs() {
  964. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  965. await Promise.all(linkElements.map(async linkElement => {
  966. const resourceURL = linkElement.href;
  967. linkElement.removeAttribute("href");
  968. const options = Object.create(this.options);
  969. options.insertSingleFileComment = false;
  970. options.insertFaviconLink = false;
  971. options.removeUnusedStyles = false;
  972. options.removeAlternativeMedias = false;
  973. options.removeUnusedFonts = false;
  974. options.removeHiddenElements = false;
  975. options.doc = null;
  976. options.win = null;
  977. options.url = resourceURL;
  978. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  979. if (attributeValue) {
  980. const importData = options.imports[Number(attributeValue)];
  981. if (importData) {
  982. options.content = importData.content;
  983. importData.runner = new Runner(options);
  984. await importData.runner.loadPage();
  985. await importData.runner.initialize();
  986. if (!options.removeImports) {
  987. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  988. }
  989. importData.runner.getStyleSheets().forEach(stylesheet => {
  990. const importedStyleElement = this.doc.createElement("style");
  991. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  992. this.stylesheets.set(importedStyleElement, stylesheet);
  993. });
  994. }
  995. }
  996. if (options.removeImports) {
  997. linkElement.remove();
  998. this.stats.add("discarded", "HTML imports", 1);
  999. }
  1000. }));
  1001. }
  1002. removeUnusedStyles() {
  1003. if (!this.mediaAllInfo) {
  1004. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1005. }
  1006. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1007. this.stats.set("processed", "CSS rules", stats.processed);
  1008. this.stats.set("discarded", "CSS rules", stats.discarded);
  1009. }
  1010. removeUnusedFonts() {
  1011. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1012. }
  1013. removeAlternativeMedias() {
  1014. const stats = util.minifyMedias(this.stylesheets);
  1015. this.stats.set("processed", "medias", stats.processed);
  1016. this.stats.set("discarded", "medias", stats.discarded);
  1017. }
  1018. async processStylesheets() {
  1019. await Promise.all(Array.from(this.stylesheets).map(([, stylesheetInfo]) =>
  1020. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1021. ));
  1022. }
  1023. async processStyleAttributes() {
  1024. return Promise.all(Array.from(this.styles).map(([, declarationList]) =>
  1025. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1026. ));
  1027. }
  1028. async processPageResources() {
  1029. const processAttributeArgs = [
  1030. ["link[href][rel*=\"icon\"]", "href", false, true],
  1031. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1032. ["img[src], input[src][type=image]", "src", true],
  1033. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1034. ["video[poster]", "poster"],
  1035. ["*[background]", "background"],
  1036. ["image", "xlink:href"]
  1037. ];
  1038. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1039. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1040. );
  1041. resourcePromises = resourcePromises.concat([
  1042. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1043. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1044. ]);
  1045. if (!this.options.removeAudioSrc) {
  1046. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1047. }
  1048. if (!this.options.removeVideoSrc) {
  1049. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1050. }
  1051. await Promise.all(resourcePromises);
  1052. ProcessorHelper.processShortcutIcons(this.doc);
  1053. }
  1054. async processScripts() {
  1055. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1056. let resourceURL;
  1057. let scriptSrc;
  1058. if (element.tagName == "SCRIPT") {
  1059. scriptSrc = element.getAttribute("src");
  1060. element.removeAttribute("src");
  1061. } else {
  1062. scriptSrc = element.getAttribute("href");
  1063. element.removeAttribute("href");
  1064. }
  1065. element.removeAttribute("integrity");
  1066. element.textContent = "";
  1067. try {
  1068. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1069. } catch (error) {
  1070. // ignored
  1071. }
  1072. if (Util.testValidURL(resourceURL)) {
  1073. this.stats.add("processed", "scripts", 1);
  1074. const content = await util.getContent(resourceURL, {
  1075. asBinary: true,
  1076. maxResourceSize: this.options.maxResourceSize,
  1077. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1078. frameId: this.options.windowId
  1079. });
  1080. content.data = Util.getUpdatedResourceContent(resourceURL, content, this.options);
  1081. if (element.tagName == "SCRIPT") {
  1082. element.setAttribute("src", content.data);
  1083. } else {
  1084. const scriptElement = this.doc.createElement("script");
  1085. scriptElement.setAttribute("src", content.data);
  1086. element.parentElement.replaceChild(scriptElement, element);
  1087. }
  1088. }
  1089. }));
  1090. }
  1091. removeAlternativeImages() {
  1092. util.removeAlternativeImages(this.doc);
  1093. }
  1094. async removeAlternativeFonts() {
  1095. await util.removeAlternativeFonts(this.doc, this.stylesheets);
  1096. }
  1097. async processFrames() {
  1098. if (this.options.frames) {
  1099. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1100. await Promise.all(frameElements.map(async frameElement => {
  1101. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1102. if (frameWindowId) {
  1103. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1104. if (frameData) {
  1105. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1106. if (frameData.runner) {
  1107. this.stats.add("processed", "frames", 1);
  1108. await frameData.runner.run();
  1109. const pageData = await frameData.runner.getPageData();
  1110. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1111. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1112. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1113. sandbox += " allow-scripts allow-same-origin";
  1114. }
  1115. frameElement.setAttribute("sandbox", sandbox);
  1116. if (frameElement.tagName == "OBJECT") {
  1117. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1118. } else {
  1119. if (frameElement.tagName == "FRAME") {
  1120. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/#/g, "%23"));
  1121. } else {
  1122. frameElement.setAttribute("srcdoc", pageData.content);
  1123. frameElement.removeAttribute("src");
  1124. }
  1125. }
  1126. this.stats.addAll(pageData);
  1127. } else {
  1128. this.stats.add("discarded", "frames", 1);
  1129. }
  1130. }
  1131. }
  1132. }));
  1133. }
  1134. }
  1135. async processHtmlImports() {
  1136. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1137. await Promise.all(linkElements.map(async linkElement => {
  1138. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1139. if (attributeValue) {
  1140. const importData = this.options.imports[Number(attributeValue)];
  1141. if (importData.runner) {
  1142. this.stats.add("processed", "HTML imports", 1);
  1143. await importData.runner.run();
  1144. const pageData = await importData.runner.getPageData();
  1145. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1146. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1147. this.stats.addAll(pageData);
  1148. } else {
  1149. this.stats.add("discarded", "HTML imports", 1);
  1150. }
  1151. }
  1152. }));
  1153. }
  1154. replaceStylesheets() {
  1155. this.doc.querySelectorAll("style").forEach(styleElement => {
  1156. const stylesheetInfo = this.stylesheets.get(styleElement);
  1157. if (stylesheetInfo) {
  1158. this.stylesheets.delete(styleElement);
  1159. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1160. styleElement.textContent = stylesheetContent;
  1161. if (stylesheetInfo.mediaText) {
  1162. styleElement.media = stylesheetInfo.mediaText;
  1163. }
  1164. } else {
  1165. styleElement.remove();
  1166. }
  1167. });
  1168. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1169. const stylesheetInfo = this.stylesheets.get(linkElement);
  1170. if (stylesheetInfo) {
  1171. this.stylesheets.delete(linkElement);
  1172. const styleElement = this.doc.createElement("style");
  1173. if (stylesheetInfo.mediaText) {
  1174. styleElement.media = stylesheetInfo.mediaText;
  1175. }
  1176. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1177. styleElement.textContent = stylesheetContent;
  1178. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1179. } else {
  1180. linkElement.remove();
  1181. }
  1182. });
  1183. }
  1184. replaceStyleAttributes() {
  1185. this.doc.querySelectorAll("[style]").forEach(async element => {
  1186. const declarations = this.styles.get(element);
  1187. if (declarations) {
  1188. this.styles.delete(element);
  1189. let styleContent = cssTree.generate(declarations);
  1190. element.setAttribute("style", styleContent);
  1191. } else {
  1192. element.setAttribute("style", "");
  1193. }
  1194. });
  1195. }
  1196. insertVariables() {
  1197. if (this.cssVariables.size) {
  1198. const styleElement = this.doc.createElement("style");
  1199. const firstStyleElement = this.doc.head.querySelector("style");
  1200. if (firstStyleElement) {
  1201. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1202. } else {
  1203. this.doc.head.appendChild(styleElement);
  1204. }
  1205. let stylesheetContent = "";
  1206. this.cssVariables.forEach((content, indexResource) => {
  1207. this.cssVariables.delete(indexResource);
  1208. if (stylesheetContent) {
  1209. stylesheetContent += ";";
  1210. }
  1211. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1212. });
  1213. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1214. }
  1215. }
  1216. compressHTML() {
  1217. let size;
  1218. if (this.options.displayStats) {
  1219. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1220. }
  1221. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1222. if (this.options.displayStats) {
  1223. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1224. }
  1225. }
  1226. cleanupPage() {
  1227. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1228. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1229. if (metaCharset) {
  1230. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1231. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1232. this.doc.head.querySelector("meta[charset]").remove();
  1233. }
  1234. }
  1235. }
  1236. async insertMAFFMetaData() {
  1237. const maffMetaData = await this.maffMetaDataPromise;
  1238. if (maffMetaData && maffMetaData.content) {
  1239. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1240. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1241. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1242. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1243. if (originalURLElement) {
  1244. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1245. }
  1246. if (archiveTimeElement) {
  1247. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1248. if (value) {
  1249. const date = new Date(value);
  1250. if (!isNaN(date.getTime())) {
  1251. this.options.saveDate = new Date(value);
  1252. }
  1253. }
  1254. }
  1255. }
  1256. }
  1257. async setDocInfo() {
  1258. const titleElement = this.doc.querySelector("title");
  1259. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1260. const authorElement = this.doc.querySelector("meta[name=author]");
  1261. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1262. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1263. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1264. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1265. this.options.info = {
  1266. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1267. lang: this.doc.documentElement.lang,
  1268. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1269. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1270. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : ""
  1271. };
  1272. }
  1273. }
  1274. // ---------------
  1275. // ProcessorHelper
  1276. // ---------------
  1277. const DATA_URI_PREFIX = "data:";
  1278. const ABOUT_BLANK_URI = "about:blank";
  1279. const EMPTY_DATA_URI = "data:base64,";
  1280. const REGEXP_URL_HASH = /(#.+?)$/;
  1281. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1282. class ProcessorHelper {
  1283. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1284. const date = options.saveDate;
  1285. const url = util.parseURL(options.saveUrl);
  1286. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1287. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1288. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1289. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1290. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1291. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1292. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1293. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1294. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1295. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1296. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1297. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1298. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1299. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1300. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1301. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1302. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1303. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1304. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1305. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1306. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1307. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1308. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1309. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1310. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1311. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1312. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1313. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1314. template = await Util.evalTemplateVariable(template, "url-href", () => decodeURI(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1315. template = await Util.evalTemplateVariable(template, "url-referrer", () => decodeURI(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1316. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1317. template = await Util.evalTemplateVariable(template, "url-pathname", () => decodeURI(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1318. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1319. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1320. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1321. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1322. for (const [name, value] of params) {
  1323. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1324. }
  1325. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1326. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1327. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1328. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1329. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decodeURI(Util.getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1330. if (content) {
  1331. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1332. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1333. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1334. }
  1335. return template.trim();
  1336. }
  1337. static setBackgroundImage(element, url, style) {
  1338. element.style.setProperty("background-blend-mode", "normal", "important");
  1339. element.style.setProperty("background-clip", "content-box", "important");
  1340. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1341. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1342. element.style.setProperty("background-image", url, "important");
  1343. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1344. element.style.setProperty("background-origin", "content-box", "important");
  1345. element.style.setProperty("background-repeat", "no-repeat", "important");
  1346. }
  1347. static processShortcutIcons(doc) {
  1348. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1349. if (!shortcutIcon) {
  1350. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1351. if (shortcutIcon) {
  1352. shortcutIcon.rel = "icon";
  1353. }
  1354. }
  1355. if (shortcutIcon) {
  1356. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1357. if (linkElement != shortcutIcon) {
  1358. linkElement.remove();
  1359. }
  1360. });
  1361. }
  1362. }
  1363. static removeSingleLineCssComments(stylesheet) {
  1364. const removedRules = [];
  1365. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1366. const ruleData = cssRule.data;
  1367. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1368. removedRules.push(cssRule);
  1369. }
  1370. }
  1371. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1372. }
  1373. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1374. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1375. const imports = Util.getImportFunctions(stylesheetContent);
  1376. await Promise.all(imports.map(async cssImport => {
  1377. const match = Util.matchImport(cssImport);
  1378. if (match) {
  1379. const regExpCssImport = Util.getRegExp(cssImport);
  1380. let resourceURL = Util.normalizeURL(match.resourceURL);
  1381. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1382. try {
  1383. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1384. } catch (error) {
  1385. // ignored
  1386. }
  1387. if (Util.testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1388. const content = await util.getContent(resourceURL, {
  1389. maxResourceSize: options.maxResourceSize,
  1390. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1391. validateTextContentType: true,
  1392. frameId: options.windowId
  1393. });
  1394. resourceURL = content.resourceURL;
  1395. content.data = Util.getUpdatedResourceContent(resourceURL, content, options);
  1396. let importedStylesheetContent = Util.removeCssComments(content.data);
  1397. if (options.compressCSS) {
  1398. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1399. }
  1400. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1401. if (stylesheetContent.includes(cssImport)) {
  1402. const ancestorStyleSheets = new Set(importedStyleSheets);
  1403. ancestorStyleSheets.add(resourceURL);
  1404. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1405. workStylesheet.textContent = importedStylesheetContent;
  1406. if (workStylesheet.sheet.cssRules.length) {
  1407. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1408. } else {
  1409. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1410. }
  1411. }
  1412. } else {
  1413. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1414. }
  1415. } else {
  1416. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1417. }
  1418. }
  1419. }));
  1420. return stylesheetContent;
  1421. }
  1422. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1423. const urlFunctions = Util.getUrlFunctions(stylesheetContent, true);
  1424. urlFunctions.map(urlFunction => {
  1425. const originalResourceURL = Util.matchURL(urlFunction);
  1426. const resourceURL = Util.normalizeURL(originalResourceURL);
  1427. if (!Util.testIgnoredPath(resourceURL)) {
  1428. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1429. let resolvedURL;
  1430. if (!originalResourceURL.startsWith("#")) {
  1431. try {
  1432. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1433. } catch (error) {
  1434. // ignored
  1435. }
  1436. }
  1437. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1438. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1439. }
  1440. } else {
  1441. let newUrlFunction;
  1442. if (originalResourceURL) {
  1443. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1444. } else {
  1445. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1446. }
  1447. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1448. }
  1449. }
  1450. });
  1451. return stylesheetContent;
  1452. }
  1453. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1454. resourceURL = Util.normalizeURL(resourceURL);
  1455. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1456. const content = await util.getContent(resourceURL, {
  1457. maxResourceSize: options.maxResourceSize,
  1458. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1459. charset: options.charset,
  1460. frameId: options.windowId
  1461. });
  1462. resourceURL = content.resourceURL;
  1463. content.data = Util.getUpdatedResourceContent(content.resourceURL, content, options);
  1464. let stylesheetContent = Util.removeCssComments(content.data);
  1465. if (options.compressCSS) {
  1466. stylesheetContent = util.compressCSS(stylesheetContent);
  1467. }
  1468. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1469. return stylesheetContent;
  1470. }
  1471. }
  1472. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1473. const promises = [];
  1474. const removedRules = [];
  1475. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1476. const ruleData = cssRule.data;
  1477. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1478. removedRules.push(cssRule);
  1479. } else if (ruleData.block && ruleData.block.children) {
  1480. if (ruleData.type == "Rule") {
  1481. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1482. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1483. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1484. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1485. promises.push(processFontFaceRule(ruleData));
  1486. }
  1487. }
  1488. }
  1489. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1490. await Promise.all(promises);
  1491. async function processFontFaceRule(ruleData) {
  1492. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1493. if (declaration.type == "Declaration" && declaration.value.children) {
  1494. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value), true);
  1495. await Promise.all(urlFunctions.map(async urlFunction => {
  1496. const originalResourceURL = Util.matchURL(urlFunction);
  1497. const resourceURL = Util.normalizeURL(originalResourceURL);
  1498. if (!Util.testIgnoredPath(resourceURL)) {
  1499. if (Util.testValidURL(resourceURL)) {
  1500. let { content } = await batchRequest.addURL(resourceURL, true);
  1501. replaceURLs(declaration, originalResourceURL, content);
  1502. }
  1503. }
  1504. }));
  1505. }
  1506. }));
  1507. function replaceURLs(declaration, oldURL, newURL) {
  1508. declaration.value.children.forEach(token => {
  1509. if (token.type == "Url" && util.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1510. token.value.value = newURL;
  1511. }
  1512. });
  1513. }
  1514. }
  1515. }
  1516. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1517. await Promise.all(declarations.map(async declaration => {
  1518. if (declaration.type == "Declaration" && declaration.value.children) {
  1519. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1520. await Promise.all(urlFunctions.map(async urlFunction => {
  1521. const originalResourceURL = Util.matchURL(urlFunction);
  1522. const resourceURL = Util.normalizeURL(originalResourceURL);
  1523. if (!Util.testIgnoredPath(resourceURL)) {
  1524. if (Util.testValidURL(resourceURL)) {
  1525. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1526. let variableDefined;
  1527. const tokens = [];
  1528. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1529. if (!originalResourceURL.startsWith("#")) {
  1530. if (duplicate && options.groupDuplicateImages && rootFunction) {
  1531. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1532. tokens.push({ parent, token, value });
  1533. variableDefined = true;
  1534. } else {
  1535. token.data.value.value = content;
  1536. }
  1537. }
  1538. });
  1539. if (variableDefined) {
  1540. cssVariables.set(indexResource, content);
  1541. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1542. }
  1543. }
  1544. }
  1545. }));
  1546. }
  1547. }));
  1548. function findURLToken(url, children, callback, depth = 0) {
  1549. for (let token = children.head; token; token = token.next) {
  1550. if (token.data.children) {
  1551. findURLToken(url, token.data.children, callback, depth + 1);
  1552. }
  1553. if (token.data.type == "Url" && util.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1554. callback(token, children, depth == 0);
  1555. }
  1556. }
  1557. }
  1558. }
  1559. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1560. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1561. let resourceURL = resourceElement.getAttribute(attributeName);
  1562. resourceURL = Util.normalizeURL(resourceURL);
  1563. if (!Util.testIgnoredPath(resourceURL)) {
  1564. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1565. if (Util.testValidPath(resourceURL)) {
  1566. try {
  1567. resourceURL = util.resolveURL(resourceURL, baseURI);
  1568. } catch (error) {
  1569. // ignored
  1570. }
  1571. if (Util.testValidURL(resourceURL)) {
  1572. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1573. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1574. resourceElement.remove();
  1575. } else {
  1576. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1577. if (!forbiddenPrefixFound) {
  1578. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1579. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1580. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1581. cssVariables.set(indexResource, content);
  1582. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1583. styles.set(resourceElement, declarationList);
  1584. } else {
  1585. resourceElement.setAttribute(attributeName, content);
  1586. }
  1587. } else {
  1588. resourceElement.setAttribute(attributeName, content);
  1589. }
  1590. }
  1591. }
  1592. }
  1593. }
  1594. }
  1595. }));
  1596. }
  1597. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1598. const attributeName = "xlink:href";
  1599. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1600. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1601. let resourceURL = Util.normalizeURL(originalResourceURL);
  1602. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1603. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1604. try {
  1605. resourceURL = util.resolveURL(resourceURL, baseURI);
  1606. } catch (error) {
  1607. // ignored
  1608. }
  1609. if (Util.testValidURL(resourceURL)) {
  1610. const { content } = await batchRequest.addURL(resourceURL);
  1611. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1612. if (hashMatch && hashMatch[0]) {
  1613. let symbolElement;
  1614. try {
  1615. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1616. } catch (error) {
  1617. // ignored
  1618. }
  1619. if (symbolElement) {
  1620. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1621. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1622. }
  1623. } else {
  1624. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1625. }
  1626. }
  1627. } else if (resourceURL == options.url) {
  1628. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1629. }
  1630. }));
  1631. }
  1632. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1633. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1634. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1635. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1636. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1637. if (!Util.testIgnoredPath(resourceURL)) {
  1638. if (Util.testValidPath(resourceURL)) {
  1639. try {
  1640. resourceURL = util.resolveURL(resourceURL, baseURI);
  1641. } catch (error) {
  1642. // ignored
  1643. }
  1644. if (Util.testValidURL(resourceURL)) {
  1645. const { content } = await batchRequest.addURL(resourceURL, true);
  1646. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1647. if (forbiddenPrefixFound) {
  1648. return "";
  1649. }
  1650. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1651. } else {
  1652. return "";
  1653. }
  1654. } else {
  1655. return "";
  1656. }
  1657. } else {
  1658. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1659. }
  1660. }));
  1661. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1662. }));
  1663. }
  1664. static replaceImageSource(imgElement, variableName, options) {
  1665. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1666. if (attributeValue) {
  1667. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1668. if (imageData && imageData.replaceable) {
  1669. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1670. const backgroundStyle = {};
  1671. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1672. if (backgroundSize) {
  1673. backgroundStyle["background-size"] = imageData.objectFit;
  1674. }
  1675. if (imageData.objectPosition) {
  1676. backgroundStyle["background-position"] = imageData.objectPosition;
  1677. }
  1678. if (imageData.backgroundColor) {
  1679. backgroundStyle["background-color"] = imageData.backgroundColor;
  1680. }
  1681. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1682. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1683. return true;
  1684. }
  1685. }
  1686. }
  1687. }
  1688. // ----
  1689. // Util
  1690. // ----
  1691. const BLOB_URI_PREFIX = "blob:";
  1692. const HTTP_URI_PREFIX = /^https?:\/\//;
  1693. const FILE_URI_PREFIX = /^file:\/\//;
  1694. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1695. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1696. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1697. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1698. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1699. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1700. 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;
  1701. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1702. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1703. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1704. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1705. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1706. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1707. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1708. class Util {
  1709. static getUpdatedResourceContent(resourceURL, content, options) {
  1710. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1711. options.updatedResources[resourceURL].retrieved = true;
  1712. return options.updatedResources[resourceURL].content;
  1713. } else {
  1714. return content.data || "";
  1715. }
  1716. }
  1717. static normalizeURL(url) {
  1718. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1719. return url;
  1720. } else {
  1721. return url.split("#")[0];
  1722. }
  1723. }
  1724. static getCSSValue(value) {
  1725. let result = "";
  1726. try {
  1727. result = cssTree.generate(value);
  1728. } catch (error) {
  1729. // ignored
  1730. }
  1731. return result;
  1732. }
  1733. static getOnEventAttributeNames(doc) {
  1734. const element = doc.createElement("div");
  1735. const attributeNames = [];
  1736. for (const propertyName in element) {
  1737. if (propertyName.startsWith("on")) {
  1738. attributeNames.push(propertyName);
  1739. }
  1740. }
  1741. return attributeNames;
  1742. }
  1743. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1744. const replaceRegExp = new RegExp("{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}", "g");
  1745. if (template && template.match(replaceRegExp)) {
  1746. let value = await valueGetter();
  1747. if (!dontReplaceSlash) {
  1748. value = value.replace(/\/+/g, replacementCharacter);
  1749. }
  1750. return template.replace(replaceRegExp, value);
  1751. }
  1752. return template;
  1753. }
  1754. static getLastSegment(url, replacementCharacter) {
  1755. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1756. if (!lastSegment) {
  1757. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1758. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1759. }
  1760. if (!lastSegment) {
  1761. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1762. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1763. }
  1764. if (!lastSegment) {
  1765. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1766. }
  1767. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1768. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1769. lastSegment = lastSegmentMatch[1];
  1770. }
  1771. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1772. return lastSegment;
  1773. }
  1774. static getRegExp(string) {
  1775. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1776. }
  1777. static getUrlFunctions(stylesheetContent, unique) {
  1778. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1779. if (unique) {
  1780. return [...new Set(result)];
  1781. } else {
  1782. return result;
  1783. }
  1784. }
  1785. static getImportFunctions(stylesheetContent) {
  1786. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1787. }
  1788. static findShortcutIcon(shortcutIcons) {
  1789. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1790. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1791. return shortcutIcons[0];
  1792. }
  1793. static matchURL(stylesheetContent) {
  1794. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1795. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1796. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1797. return match && match[1];
  1798. }
  1799. static testIgnoredPath(resourceURL) {
  1800. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1801. }
  1802. static testValidPath(resourceURL) {
  1803. return resourceURL && !resourceURL.match(EMPTY_URL);
  1804. }
  1805. static testValidURL(resourceURL) {
  1806. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1807. }
  1808. static matchImport(stylesheetContent) {
  1809. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1810. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1811. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1812. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1813. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1814. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1815. if (match) {
  1816. const [, resourceURL, media] = match;
  1817. return { resourceURL, media };
  1818. }
  1819. }
  1820. static removeCssComments(stylesheetContent) {
  1821. let start, end;
  1822. do {
  1823. start = stylesheetContent.indexOf("/*");
  1824. end = stylesheetContent.indexOf("*/", start + 2);
  1825. if (start != -1 && end != -1) {
  1826. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1827. }
  1828. } while (start != -1 && end != -1);
  1829. return stylesheetContent;
  1830. }
  1831. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1832. if (mediaQuery) {
  1833. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1834. } else {
  1835. return stylesheetContent;
  1836. }
  1837. }
  1838. }
  1839. function log(...args) {
  1840. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1841. }
  1842. // -----
  1843. // Stats
  1844. // -----
  1845. const STATS_DEFAULT_VALUES = {
  1846. discarded: {
  1847. "HTML bytes": 0,
  1848. "hidden elements": 0,
  1849. "HTML imports": 0,
  1850. scripts: 0,
  1851. objects: 0,
  1852. "audio sources": 0,
  1853. "video sources": 0,
  1854. frames: 0,
  1855. "CSS rules": 0,
  1856. canvas: 0,
  1857. stylesheets: 0,
  1858. resources: 0,
  1859. medias: 0
  1860. },
  1861. processed: {
  1862. "HTML bytes": 0,
  1863. "hidden elements": 0,
  1864. "HTML imports": 0,
  1865. scripts: 0,
  1866. objects: 0,
  1867. "audio sources": 0,
  1868. "video sources": 0,
  1869. frames: 0,
  1870. "CSS rules": 0,
  1871. canvas: 0,
  1872. stylesheets: 0,
  1873. resources: 0,
  1874. medias: 0
  1875. }
  1876. };
  1877. class Stats {
  1878. constructor(options) {
  1879. this.options = options;
  1880. if (options.displayStats) {
  1881. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1882. }
  1883. }
  1884. set(type, subType, value) {
  1885. if (this.options.displayStats) {
  1886. this.data[type][subType] = value;
  1887. }
  1888. }
  1889. add(type, subType, value) {
  1890. if (this.options.displayStats) {
  1891. this.data[type][subType] += value;
  1892. }
  1893. }
  1894. addAll(pageData) {
  1895. if (this.options.displayStats) {
  1896. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1897. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1898. }
  1899. }
  1900. }
  1901. return { getClass };
  1902. })();