single-file-core.js 85 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067
  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. if (this.options.userScriptEnabled) {
  41. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  42. }
  43. this.runner = new Runner(this.options, true);
  44. await this.runner.loadPage();
  45. await this.runner.initialize();
  46. if (this.options.userScriptEnabled) {
  47. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  48. }
  49. await this.runner.run();
  50. }
  51. cancel() {
  52. this.cancelled = true;
  53. if (this.runner) {
  54. this.runner.cancel();
  55. }
  56. }
  57. async getPageData() {
  58. return this.runner.getPageData();
  59. }
  60. }
  61. // -------------
  62. // ProgressEvent
  63. // -------------
  64. const PAGE_LOADING = "page-loading";
  65. const PAGE_LOADED = "page-loaded";
  66. const RESOURCES_INITIALIZING = "resource-initializing";
  67. const RESOURCES_INITIALIZED = "resources-initialized";
  68. const RESOURCE_LOADED = "resource-loaded";
  69. const PAGE_ENDED = "page-ended";
  70. const STAGE_STARTED = "stage-started";
  71. const STAGE_ENDED = "stage-ended";
  72. const STAGE_TASK_STARTED = "stage-task-started";
  73. const STAGE_TASK_ENDED = "stage-task-ended";
  74. class ProgressEvent {
  75. constructor(type, detail) {
  76. 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 };
  77. }
  78. }
  79. // ------
  80. // Runner
  81. // ------
  82. const RESOLVE_URLS_STAGE = 0;
  83. const REPLACE_DATA_STAGE = 1;
  84. const REPLACE_DOCS_STAGE = 2;
  85. const POST_PROCESS_STAGE = 3;
  86. const STAGES = [{
  87. sequential: [
  88. { action: "preProcessPage" },
  89. { action: "replaceStyleContents" },
  90. { action: "resetCharsetMeta" },
  91. { option: "insertFaviconLink", action: "insertFaviconLink" },
  92. { action: "replaceCanvasElements" },
  93. { action: "insertFonts" },
  94. { action: "insertShadowRootContents" },
  95. { action: "setInputValues" },
  96. { option: "removeScripts", action: "removeScripts" },
  97. { option: "selected", action: "removeUnselectedElements" },
  98. { option: "removeVideoSrc", action: "insertVideoPosters" },
  99. { option: "removeFrames", action: "removeFrames" },
  100. { option: "removeVideoSrc", action: "removeVideoSources" },
  101. { option: "removeAudioSrc", action: "removeAudioSources" },
  102. { action: "removeDiscardedResources" },
  103. { option: "removeHiddenElements", action: "removeHiddenElements" },
  104. { action: "resolveHrefs" },
  105. { action: "resolveStyleAttributeURLs" }
  106. ],
  107. parallel: [
  108. { action: "resolveStylesheetURLs" },
  109. { option: "!removeFrames", action: "resolveFrameURLs" },
  110. { action: "resolveHtmlImportURLs" }
  111. ]
  112. }, {
  113. sequential: [
  114. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  115. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  116. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  117. ],
  118. parallel: [
  119. { action: "processStylesheets" },
  120. { action: "processStyleAttributes" },
  121. { action: "processPageResources" },
  122. { option: "!removeScripts", action: "processScripts" }
  123. ]
  124. }, {
  125. sequential: [
  126. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  127. ],
  128. parallel: [
  129. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  130. { option: "!removeFrames", action: "processFrames" },
  131. { option: "!removeImports", action: "processHtmlImports" },
  132. ]
  133. }, {
  134. sequential: [
  135. { action: "replaceStylesheets" },
  136. { action: "replaceStyleAttributes" },
  137. { action: "insertVariables" },
  138. { option: "compressHTML", action: "compressHTML" },
  139. { action: "cleanupPage" }
  140. ],
  141. parallel: [
  142. { option: "enableMaff", action: "insertMAFFMetaData" },
  143. { action: "setDocInfo" }
  144. ]
  145. }];
  146. class Runner {
  147. constructor(options, root) {
  148. this.root = root;
  149. this.options = options;
  150. this.options.url = this.options.url || (this.options.doc && this.options.doc.location.href);
  151. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  152. this.options.rootDocument = root;
  153. this.options.updatedResources = this.options.updatedResources || {};
  154. this.batchRequest = new BatchRequest();
  155. this.processor = new Processor(options, this.batchRequest);
  156. if (this.options.doc) {
  157. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  158. this.options.canvases = docData.canvases;
  159. this.options.fonts = docData.fonts;
  160. this.options.stylesheets = docData.stylesheets;
  161. this.options.images = docData.images;
  162. this.options.posters = docData.posters;
  163. this.options.usedFonts = docData.usedFonts;
  164. this.options.shadowRoots = docData.shadowRoots;
  165. this.options.imports = docData.imports;
  166. this.options.referrer = docData.referrer;
  167. this.markedElements = docData.markedElements;
  168. }
  169. if (this.options.saveRawPage) {
  170. this.options.removeFrames = true;
  171. }
  172. this.options.content = this.options.content || (this.options.doc ? util.serialize(this.options.doc) : null);
  173. this.onprogress = options.onprogress || (() => { });
  174. }
  175. async loadPage() {
  176. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  177. await this.processor.loadPage(this.options.content);
  178. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  179. }
  180. async initialize() {
  181. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  182. await this.executeStage(RESOLVE_URLS_STAGE);
  183. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  184. if (this.options.doc) {
  185. util.postProcessDoc(this.options.doc, this.markedElements);
  186. }
  187. this.options.doc = null;
  188. this.options.win = null;
  189. }
  190. cancel() {
  191. this.cancelled = true;
  192. this.batchRequest.cancel();
  193. if (this.root) {
  194. if (this.options.frames) {
  195. this.options.frames.forEach(cancelRunner);
  196. }
  197. if (this.options.imports) {
  198. this.options.imports.forEach(cancelRunner);
  199. }
  200. }
  201. function cancelRunner(resourceData) {
  202. if (resourceData.runner) {
  203. resourceData.runner.cancel();
  204. }
  205. }
  206. }
  207. async run() {
  208. if (this.root) {
  209. this.processor.initialize(this.batchRequest);
  210. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  211. }
  212. await this.batchRequest.run(detail => {
  213. detail.pageURL = this.options.url;
  214. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  215. }, this.options);
  216. await this.pendingPromises;
  217. await this.executeStage(REPLACE_DOCS_STAGE);
  218. await this.executeStage(POST_PROCESS_STAGE);
  219. this.processor.finalize();
  220. }
  221. getDocument() {
  222. return this.processor.doc;
  223. }
  224. getStyleSheets() {
  225. return this.processor.stylesheets;
  226. }
  227. async getPageData() {
  228. if (this.root) {
  229. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  230. }
  231. return this.processor.getPageData();
  232. }
  233. async executeStage(step) {
  234. if (DEBUG) {
  235. log("**** STARTED STAGE", step, "****");
  236. }
  237. const frame = !this.root;
  238. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  239. STAGES[step].sequential.forEach(task => {
  240. let startTime;
  241. if (DEBUG) {
  242. startTime = Date.now();
  243. log(" -- STARTED task =", task.action);
  244. }
  245. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  246. if (!this.cancelled) {
  247. this.executeTask(task);
  248. }
  249. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  250. if (DEBUG) {
  251. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  252. }
  253. });
  254. let parallelTasksPromise;
  255. if (STAGES[step].parallel) {
  256. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  257. let startTime;
  258. if (DEBUG) {
  259. startTime = Date.now();
  260. log(" // STARTED task =", task.action);
  261. }
  262. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  263. if (!this.cancelled) {
  264. await this.executeTask(task);
  265. }
  266. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  267. if (DEBUG) {
  268. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  269. }
  270. }));
  271. } else {
  272. parallelTasksPromise = Promise.resolve();
  273. }
  274. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  275. if (DEBUG) {
  276. log("**** ENDED STAGE", step, "****");
  277. }
  278. return parallelTasksPromise;
  279. }
  280. executeTask(task) {
  281. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  282. return this.processor[task.action]();
  283. }
  284. }
  285. }
  286. // ------------
  287. // BatchRequest
  288. // ------------
  289. class BatchRequest {
  290. constructor() {
  291. this.requests = new Map();
  292. this.duplicates = new Map();
  293. }
  294. async addURL(resourceURL, asBinary, groupDuplicates) {
  295. return new Promise((resolve, reject) => {
  296. const requestKey = JSON.stringify([resourceURL, asBinary]);
  297. let resourceRequests = this.requests.get(requestKey);
  298. if (!resourceRequests) {
  299. resourceRequests = [];
  300. this.requests.set(requestKey, resourceRequests);
  301. }
  302. const callbacks = { resolve, reject };
  303. resourceRequests.push(callbacks);
  304. if (groupDuplicates) {
  305. let duplicateRequests = this.duplicates.get(requestKey);
  306. if (!duplicateRequests) {
  307. duplicateRequests = [];
  308. this.duplicates.set(requestKey, duplicateRequests);
  309. }
  310. duplicateRequests.push(callbacks);
  311. }
  312. });
  313. }
  314. getMaxResources() {
  315. return this.requests.size;
  316. }
  317. async run(onloadListener, options) {
  318. const resourceURLs = [...this.requests.keys()];
  319. let indexResource = 0;
  320. return Promise.all(resourceURLs.map(async requestKey => {
  321. const [resourceURL, asBinary] = JSON.parse(requestKey);
  322. const resourceRequests = this.requests.get(requestKey);
  323. try {
  324. const content = await util.getContent(resourceURL, {
  325. asBinary,
  326. maxResourceSize: options.maxResourceSize,
  327. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  328. frameId: options.windowId
  329. });
  330. indexResource = indexResource + 1;
  331. onloadListener({ url: resourceURL });
  332. if (!this.cancelled) {
  333. resourceRequests.forEach(callbacks => {
  334. const duplicateCallbacks = this.duplicates.get(requestKey);
  335. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  336. callbacks.resolve({ content: content.data, indexResource, duplicate });
  337. });
  338. }
  339. } catch (error) {
  340. indexResource = indexResource + 1;
  341. onloadListener({ url: resourceURL });
  342. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  343. }
  344. this.requests.delete(requestKey);
  345. }));
  346. }
  347. cancel() {
  348. this.cancelled = true;
  349. const resourceURLs = [...this.requests.keys()];
  350. resourceURLs.forEach(requestKey => {
  351. const resourceRequests = this.requests.get(requestKey);
  352. resourceRequests.forEach(callbacks => callbacks.reject());
  353. this.requests.delete(requestKey);
  354. });
  355. }
  356. }
  357. // ---------
  358. // Processor
  359. // ---------
  360. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  361. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  362. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  363. const SCRIPT_TAG_FOUND = /<script/gi;
  364. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  365. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  366. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  367. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  368. class Processor {
  369. constructor(options, batchRequest) {
  370. this.options = options;
  371. this.stats = new Stats(options);
  372. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  373. this.batchRequest = batchRequest;
  374. this.stylesheets = new Map();
  375. this.styles = new Map();
  376. this.cssVariables = new Map();
  377. }
  378. initialize() {
  379. this.options.saveDate = new Date();
  380. this.options.saveUrl = this.options.url;
  381. if (this.options.enableMaff) {
  382. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  383. }
  384. this.maxResources = this.batchRequest.getMaxResources();
  385. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  386. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  387. }
  388. if (!this.options.removeImports && this.options.imports) {
  389. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  390. }
  391. this.stats.set("processed", "resources", this.maxResources);
  392. }
  393. async loadPage(pageContent, charset) {
  394. let content;
  395. if (!pageContent || this.options.saveRawPage) {
  396. content = await util.getContent(this.baseURI, {
  397. maxResourceSize: this.options.maxResourceSize,
  398. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  399. charset,
  400. frameId: this.options.windowId
  401. });
  402. pageContent = content.data;
  403. }
  404. this.doc = util.parseDocContent(pageContent, this.baseURI);
  405. if (this.options.saveRawPage) {
  406. let charset;
  407. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  408. const charsetDeclaration = element.content.split(";")[1];
  409. if (charsetDeclaration && !charset) {
  410. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  411. }
  412. });
  413. if (charset && charset != content.charset) {
  414. return this.loadPage(pageContent, charset);
  415. }
  416. }
  417. this.workStyleElement = this.doc.createElement("style");
  418. this.doc.body.appendChild(this.workStyleElement);
  419. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  420. }
  421. finalize() {
  422. if (this.workStyleElement.parentNode) {
  423. this.workStyleElement.remove();
  424. }
  425. }
  426. async getPageData() {
  427. util.postProcessDoc(this.doc);
  428. const url = util.parseURL(this.baseURI);
  429. if (this.options.insertSingleFileComment) {
  430. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  431. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  432. " \n url: " + this.options.saveUrl +
  433. " \n saved date: " + this.options.saveDate +
  434. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  435. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  436. }
  437. let size;
  438. if (this.options.displayStats) {
  439. size = util.getContentSize(this.doc.documentElement.outerHTML);
  440. }
  441. const content = util.serialize(this.doc, this.options.compressHTML);
  442. if (this.options.displayStats) {
  443. const contentSize = util.getContentSize(content);
  444. this.stats.set("processed", "HTML bytes", contentSize);
  445. this.stats.add("discarded", "HTML bytes", size - contentSize);
  446. }
  447. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  448. const replacementCharacter = this.options.filenameReplacementCharacter;
  449. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  450. if (!this.options.backgroundSave) {
  451. filename = filename.replace(/\//g, replacementCharacter);
  452. }
  453. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  454. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  455. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  456. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  457. filename = filename + "…" + extension;
  458. }
  459. if (!filename) {
  460. filename = "Unnamed page";
  461. }
  462. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  463. const pageData = {
  464. stats: this.stats.data,
  465. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  466. filename,
  467. content
  468. };
  469. if (this.options.addProof) {
  470. pageData.hash = await util.digest("SHA-256", content);
  471. }
  472. return pageData;
  473. }
  474. preProcessPage() {
  475. if (this.options.win) {
  476. 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));
  477. }
  478. if (this.options.images && !this.options.saveRawPage) {
  479. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  480. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  481. if (attributeValue) {
  482. const imageData = this.options.images[Number(attributeValue)];
  483. if (imageData) {
  484. if (this.options.removeHiddenElements && (
  485. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  486. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  487. )) {
  488. imgElement.setAttribute("src", EMPTY_IMAGE);
  489. } else {
  490. if (imageData.currentSrc) {
  491. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  492. imgElement.setAttribute("src", imageData.currentSrc);
  493. }
  494. if (this.options.loadDeferredImages) {
  495. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  496. imageData.src = imgElement.dataset.src;
  497. imgElement.setAttribute("src", imgElement.dataset.src);
  498. imgElement.removeAttribute("data-src");
  499. }
  500. }
  501. }
  502. }
  503. }
  504. });
  505. if (this.options.loadDeferredImages) {
  506. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  507. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  508. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  509. imgElement.removeAttribute("data-srcset");
  510. }
  511. });
  512. }
  513. }
  514. }
  515. replaceStyleContents() {
  516. if (this.options.stylesheets) {
  517. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  518. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  519. if (attributeValue) {
  520. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  521. if (stylesheetContent) {
  522. styleElement.textContent = stylesheetContent;
  523. }
  524. }
  525. });
  526. }
  527. }
  528. removeUnselectedElements() {
  529. removeUnmarkedElements(this.doc.body);
  530. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  531. function removeUnmarkedElements(element) {
  532. let selectedElementFound = false;
  533. Array.from(element.childNodes).forEach(node => {
  534. if (node.nodeType == 1) {
  535. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  536. selectedElementFound = selectedElementFound || isSelectedElement;
  537. if (isSelectedElement) {
  538. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  539. removeUnmarkedElements(node);
  540. } else if (selectedElementFound) {
  541. removeNode(node);
  542. } else {
  543. hideNode(node);
  544. }
  545. }
  546. });
  547. }
  548. function removeNode(node) {
  549. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  550. node.remove();
  551. } else {
  552. hideNode(node);
  553. }
  554. }
  555. function hideNode(node) {
  556. if (canHideNode(node)) {
  557. node.style.setProperty("display", "none", "important");
  558. Array.from(node.childNodes).forEach(removeNode);
  559. }
  560. }
  561. function canHideNode(node) {
  562. if (node.nodeType == 1) {
  563. const tagName = node.tagName && node.tagName.toLowerCase();
  564. return (tagName != "svg" && tagName != "style" && tagName != "link");
  565. }
  566. }
  567. }
  568. insertVideoPosters() {
  569. if (this.options.posters) {
  570. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  571. let videoElement;
  572. if (element.tagName == "VIDEO") {
  573. videoElement = element;
  574. } else {
  575. videoElement = element.parentElement;
  576. }
  577. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  578. if (attributeValue) {
  579. const posterURL = this.options.posters[Number(attributeValue)];
  580. if (!videoElement.poster && posterURL) {
  581. videoElement.setAttribute("poster", posterURL);
  582. }
  583. }
  584. });
  585. }
  586. }
  587. removeFrames() {
  588. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  589. this.stats.set("discarded", "frames", frameElements.length);
  590. this.stats.set("processed", "frames", frameElements.length);
  591. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  592. }
  593. removeScripts() {
  594. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  595. this.doc.querySelectorAll("[href]").forEach(element => {
  596. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  597. element.setAttribute("href", "");
  598. }
  599. });
  600. this.doc.querySelectorAll("[src]").forEach(element => {
  601. if (element.src && element.src.match(/^\s*javascript:/)) {
  602. element.removeAttribute("src");
  603. }
  604. });
  605. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  606. this.stats.set("discarded", "scripts", scriptElements.length);
  607. this.stats.set("processed", "scripts", scriptElements.length);
  608. scriptElements.forEach(element => element.remove());
  609. }
  610. removeVideoSources() {
  611. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  612. this.stats.set("discarded", "video sources", videoSourceElements.length);
  613. this.stats.set("processed", "video sources", videoSourceElements.length);
  614. videoSourceElements.forEach(element => {
  615. if (element.tagName == "SOURCE") {
  616. element.remove();
  617. } else {
  618. videoSourceElements.forEach(element => element.removeAttribute("src"));
  619. }
  620. });
  621. }
  622. removeAudioSources() {
  623. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  624. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  625. this.stats.set("processed", "audio sources", audioSourceElements.length);
  626. audioSourceElements.forEach(element => {
  627. if (element.tagName == "SOURCE") {
  628. element.remove();
  629. } else {
  630. audioSourceElements.forEach(element => element.removeAttribute("src"));
  631. }
  632. });
  633. }
  634. removeDiscardedResources() {
  635. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  636. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  637. 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\"])");
  638. this.stats.set("discarded", "objects", objectElements.length);
  639. this.stats.set("processed", "objects", objectElements.length);
  640. objectElements.forEach(element => element.remove());
  641. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  642. replacedAttributeValue.forEach(element => {
  643. let regExp;
  644. if (this.options.removeScripts) {
  645. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  646. } else {
  647. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  648. }
  649. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  650. if (relValue.length) {
  651. element.setAttribute("rel", relValue);
  652. } else {
  653. element.remove();
  654. }
  655. });
  656. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  657. if (this.options.compressHTML) {
  658. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  659. }
  660. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  661. }
  662. resetCharsetMeta() {
  663. let charset;
  664. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  665. const charsetDeclaration = element.content.split(";")[1];
  666. if (charsetDeclaration && !charset) {
  667. charset = charsetDeclaration.split("=")[1];
  668. if (charset) {
  669. this.charset = charset.trim().toLowerCase();
  670. }
  671. }
  672. element.remove();
  673. });
  674. const metaElement = this.doc.createElement("meta");
  675. metaElement.setAttribute("charset", "utf-8");
  676. if (this.doc.head.firstChild) {
  677. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  678. } else {
  679. this.doc.head.appendChild(metaElement);
  680. }
  681. }
  682. setInputValues() {
  683. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  684. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  685. input.setAttribute("value", value || "");
  686. });
  687. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  688. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  689. if (value == "true") {
  690. input.setAttribute("checked", "");
  691. }
  692. });
  693. this.doc.querySelectorAll("textarea").forEach(textarea => {
  694. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  695. textarea.textContent = value || "";
  696. });
  697. this.doc.querySelectorAll("select").forEach(select => {
  698. select.querySelectorAll("option").forEach(option => {
  699. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  700. if (selected) {
  701. option.setAttribute("selected", "");
  702. }
  703. });
  704. });
  705. }
  706. insertFaviconLink() {
  707. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  708. if (!faviconElement) {
  709. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  710. }
  711. if (!faviconElement) {
  712. faviconElement = this.doc.createElement("link");
  713. faviconElement.setAttribute("type", "image/x-icon");
  714. faviconElement.setAttribute("rel", "shortcut icon");
  715. faviconElement.setAttribute("href", "/favicon.ico");
  716. }
  717. this.doc.head.appendChild(faviconElement);
  718. }
  719. replaceCanvasElements() {
  720. if (this.options.canvases) {
  721. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  722. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  723. if (attributeValue) {
  724. const canvasData = this.options.canvases[Number(attributeValue)];
  725. if (canvasData) {
  726. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  727. this.stats.add("processed", "canvas", 1);
  728. }
  729. }
  730. });
  731. }
  732. }
  733. insertFonts() {
  734. if (this.options.fonts && this.options.fonts.length) {
  735. let stylesheetContent = "";
  736. this.options.fonts.forEach(fontData => {
  737. if (fontData["font-family"] && fontData.src) {
  738. stylesheetContent += "@font-face{";
  739. let stylesContent = "";
  740. Object.keys(fontData).forEach(fontStyle => {
  741. if (stylesContent) {
  742. stylesContent += ";";
  743. }
  744. stylesContent += fontStyle + ":" + fontData[fontStyle];
  745. });
  746. stylesheetContent += stylesContent + "}";
  747. }
  748. });
  749. if (stylesheetContent) {
  750. const styleElement = this.doc.createElement("style");
  751. styleElement.textContent = stylesheetContent;
  752. const existingStyleElement = this.doc.querySelector("style");
  753. if (existingStyleElement) {
  754. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  755. } else {
  756. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  757. }
  758. }
  759. }
  760. }
  761. removeHiddenElements() {
  762. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  763. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  764. this.stats.set("discarded", "hidden elements", removedElements.length);
  765. this.stats.set("processed", "hidden elements", removedElements.length);
  766. if (hiddenElements.length) {
  767. const styleElement = this.doc.createElement("style");
  768. styleElement.textContent = ".sf-hidden{display:none!important;}";
  769. this.doc.head.appendChild(styleElement);
  770. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  771. }
  772. removedElements.forEach(element => element.remove());
  773. }
  774. resolveHrefs() {
  775. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  776. const href = element.getAttribute("href").trim();
  777. if (!Util.testIgnoredPath(href)) {
  778. let resolvedURL;
  779. try {
  780. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  781. } catch (error) {
  782. // ignored
  783. }
  784. if (resolvedURL) {
  785. const url = Util.normalizeURL(this.options.url);
  786. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  787. resolvedURL = resolvedURL.substring(url.length);
  788. }
  789. try {
  790. element.setAttribute("href", resolvedURL);
  791. } catch (error) {
  792. // ignored
  793. }
  794. }
  795. }
  796. });
  797. }
  798. resolveStyleAttributeURLs() {
  799. this.doc.querySelectorAll("[style]").forEach(element => {
  800. let styleContent = element.getAttribute("style");
  801. if (this.options.compressCSS) {
  802. styleContent = util.compressCSS(styleContent);
  803. }
  804. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  805. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  806. this.styles.set(element, declarationList);
  807. });
  808. }
  809. async resolveStylesheetURLs() {
  810. const options = {
  811. maxResourceSize: this.options.maxResourceSize,
  812. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  813. url: this.options.url,
  814. charset: this.charset,
  815. compressCSS: this.options.compressCSS,
  816. updatedResources: this.options.updatedResources,
  817. rootDocument: this.options.rootDocument,
  818. frameId: this.options.windowId
  819. };
  820. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  821. .map(async element => {
  822. let mediaText;
  823. if (element.media) {
  824. mediaText = element.media.toLowerCase();
  825. }
  826. const stylesheetInfo = { mediaText };
  827. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  828. stylesheetInfo.scoped = true;
  829. }
  830. if (element.tagName.toLowerCase() == "link") {
  831. if (element.charset) {
  832. options.charset = element.charset;
  833. }
  834. }
  835. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  836. }));
  837. if (options.rootDocument) {
  838. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  839. await Promise.all(newResources.map(async resource => {
  840. resource.retrieved = true;
  841. const stylesheetInfo = {};
  842. const element = this.doc.createElement("style");
  843. this.doc.body.appendChild(element);
  844. element.textContent = resource.content;
  845. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  846. }));
  847. }
  848. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  849. stylesheets.set(element, stylesheetInfo);
  850. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  851. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  852. if (match && match[1] && match[1] != options.charset) {
  853. options.charset = match[1];
  854. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  855. }
  856. let stylesheet;
  857. try {
  858. stylesheet = cssTree.parse(Util.removeCssComments(stylesheetContent));
  859. } catch (error) {
  860. // ignored
  861. }
  862. if (stylesheet && stylesheet.children) {
  863. if (options.compressCSS) {
  864. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  865. }
  866. stylesheetInfo.stylesheet = stylesheet;
  867. } else {
  868. stylesheets.delete(element);
  869. }
  870. }
  871. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  872. let content;
  873. if (element.tagName.toLowerCase() == "link") {
  874. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  875. } else {
  876. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  877. }
  878. return content || "";
  879. }
  880. }
  881. async resolveFrameURLs() {
  882. if (!this.options.saveRawPage && this.options.frames) {
  883. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  884. await Promise.all(frameElements.map(async frameElement => {
  885. if (frameElement.tagName == "OBJECT") {
  886. frameElement.setAttribute("data", "data:text/html,");
  887. } else {
  888. frameElement.removeAttribute("src");
  889. frameElement.removeAttribute("srcdoc");
  890. }
  891. Array.from(frameElement.childNodes).forEach(node => node.remove());
  892. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  893. if (frameWindowId) {
  894. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  895. if (frameData) {
  896. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  897. }
  898. }
  899. }));
  900. }
  901. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  902. options.insertSingleFileComment = false;
  903. options.insertFaviconLink = false;
  904. options.doc = null;
  905. options.win = null;
  906. options.url = frameData.baseURI;
  907. options.windowId = frameWindowId;
  908. if (frameData.content) {
  909. options.content = frameData.content;
  910. options.canvases = frameData.canvases;
  911. options.fonts = frameData.fonts;
  912. options.stylesheets = frameData.stylesheets;
  913. options.images = frameData.images;
  914. options.posters = frameData.posters;
  915. options.usedFonts = frameData.usedFonts;
  916. options.shadowRoots = frameData.shadowRoots;
  917. options.imports = frameData.imports;
  918. frameData.runner = new Runner(options);
  919. frameData.frameElement = frameElement;
  920. await frameData.runner.loadPage();
  921. await frameData.runner.initialize();
  922. frameData.maxResources = batchRequest.getMaxResources();
  923. }
  924. }
  925. }
  926. insertShadowRootContents() {
  927. const doc = this.doc;
  928. const options = this.options;
  929. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  930. processElement(this.doc);
  931. if (this.options.removeScripts) {
  932. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  933. }
  934. const scriptElement = doc.createElement("script");
  935. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  936. 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)}})}})()`;
  937. doc.body.appendChild(scriptElement);
  938. }
  939. function processElement(element) {
  940. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  941. shadowRootElements.forEach(element => {
  942. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  943. if (attributeValue) {
  944. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  945. if (shadowRootData) {
  946. const templateElement = doc.createElement("template");
  947. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  948. if (shadowRootData.delegatesFocus) {
  949. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  950. }
  951. if (shadowRootData.adoptedStyleSheets) {
  952. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  953. const styleElement = doc.createElement("style");
  954. styleElement.textContent = stylesheetContent;
  955. templateElement.appendChild(styleElement);
  956. });
  957. }
  958. const shadowDoc = util.parseDocContent(shadowRootData.content);
  959. if (shadowDoc.head) {
  960. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  961. if (metaCharset) {
  962. metaCharset.remove();
  963. }
  964. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  965. }
  966. if (shadowDoc.body) {
  967. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  968. }
  969. processElement(templateElement);
  970. if (element.firstChild) {
  971. element.insertBefore(templateElement, element.firstChild);
  972. } else {
  973. element.appendChild(templateElement);
  974. }
  975. }
  976. }
  977. });
  978. }
  979. }
  980. async resolveHtmlImportURLs() {
  981. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  982. await Promise.all(linkElements.map(async linkElement => {
  983. const resourceURL = linkElement.href;
  984. linkElement.removeAttribute("href");
  985. const options = Object.create(this.options);
  986. options.insertSingleFileComment = false;
  987. options.insertFaviconLink = false;
  988. options.removeUnusedStyles = false;
  989. options.removeAlternativeMedias = false;
  990. options.removeUnusedFonts = false;
  991. options.removeHiddenElements = false;
  992. options.doc = null;
  993. options.win = null;
  994. options.url = resourceURL;
  995. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  996. if (attributeValue) {
  997. const importData = options.imports[Number(attributeValue)];
  998. if (importData) {
  999. options.content = importData.content;
  1000. importData.runner = new Runner(options);
  1001. await importData.runner.loadPage();
  1002. await importData.runner.initialize();
  1003. if (!options.removeImports) {
  1004. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1005. }
  1006. importData.runner.getStyleSheets().forEach(stylesheet => {
  1007. const importedStyleElement = this.doc.createElement("style");
  1008. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1009. this.stylesheets.set(importedStyleElement, stylesheet);
  1010. });
  1011. }
  1012. }
  1013. if (options.removeImports) {
  1014. linkElement.remove();
  1015. this.stats.add("discarded", "HTML imports", 1);
  1016. }
  1017. }));
  1018. }
  1019. removeUnusedStyles() {
  1020. if (!this.mediaAllInfo) {
  1021. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1022. }
  1023. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1024. this.stats.set("processed", "CSS rules", stats.processed);
  1025. this.stats.set("discarded", "CSS rules", stats.discarded);
  1026. }
  1027. removeUnusedFonts() {
  1028. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1029. }
  1030. removeAlternativeMedias() {
  1031. const stats = util.minifyMedias(this.stylesheets);
  1032. this.stats.set("processed", "medias", stats.processed);
  1033. this.stats.set("discarded", "medias", stats.discarded);
  1034. }
  1035. async processStylesheets() {
  1036. await Promise.all(Array.from(this.stylesheets).map(([, stylesheetInfo]) =>
  1037. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1038. ));
  1039. }
  1040. async processStyleAttributes() {
  1041. return Promise.all(Array.from(this.styles).map(([, declarationList]) =>
  1042. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1043. ));
  1044. }
  1045. async processPageResources() {
  1046. const processAttributeArgs = [
  1047. ["link[href][rel*=\"icon\"]", "href", false, true],
  1048. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1049. ["img[src], input[src][type=image]", "src", true],
  1050. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1051. ["video[poster]", "poster"],
  1052. ["*[background]", "background"],
  1053. ["image", "xlink:href"],
  1054. ["image", "href"]
  1055. ];
  1056. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1057. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1058. );
  1059. resourcePromises = resourcePromises.concat([
  1060. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1061. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1062. ]);
  1063. if (!this.options.removeAudioSrc) {
  1064. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1065. }
  1066. if (!this.options.removeVideoSrc) {
  1067. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1068. }
  1069. await Promise.all(resourcePromises);
  1070. ProcessorHelper.processShortcutIcons(this.doc);
  1071. }
  1072. async processScripts() {
  1073. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1074. let resourceURL;
  1075. let scriptSrc;
  1076. if (element.tagName == "SCRIPT") {
  1077. scriptSrc = element.getAttribute("src");
  1078. element.removeAttribute("src");
  1079. } else {
  1080. scriptSrc = element.getAttribute("href");
  1081. element.removeAttribute("href");
  1082. }
  1083. element.removeAttribute("integrity");
  1084. element.textContent = "";
  1085. try {
  1086. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1087. } catch (error) {
  1088. // ignored
  1089. }
  1090. if (Util.testValidURL(resourceURL)) {
  1091. this.stats.add("processed", "scripts", 1);
  1092. const content = await util.getContent(resourceURL, {
  1093. asBinary: true,
  1094. maxResourceSize: this.options.maxResourceSize,
  1095. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1096. frameId: this.options.windowId
  1097. });
  1098. content.data = Util.getUpdatedResourceContent(resourceURL, content, this.options);
  1099. if (element.tagName == "SCRIPT") {
  1100. element.setAttribute("src", content.data);
  1101. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1102. element.setAttribute("async", "");
  1103. }
  1104. } else {
  1105. const scriptElement = this.doc.createElement("script");
  1106. scriptElement.setAttribute("src", content.data);
  1107. scriptElement.setAttribute("async", "");
  1108. element.parentElement.replaceChild(scriptElement, element);
  1109. }
  1110. }
  1111. }));
  1112. }
  1113. removeAlternativeImages() {
  1114. util.removeAlternativeImages(this.doc);
  1115. }
  1116. async removeAlternativeFonts() {
  1117. await util.removeAlternativeFonts(this.doc, this.stylesheets);
  1118. }
  1119. async processFrames() {
  1120. if (this.options.frames) {
  1121. const frameElements = Array.from(this.doc.querySelectorAll("iframe:not([" + util.HIDDEN_FRAME_ATTRIBUTE_NAME + "]), frame, object[type=\"text/html\"][data]"));
  1122. await Promise.all(frameElements.map(async frameElement => {
  1123. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1124. if (frameWindowId) {
  1125. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1126. if (frameData) {
  1127. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1128. if (frameData.runner) {
  1129. this.stats.add("processed", "frames", 1);
  1130. await frameData.runner.run();
  1131. const pageData = await frameData.runner.getPageData();
  1132. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1133. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1134. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1135. sandbox += " allow-scripts allow-same-origin";
  1136. }
  1137. frameElement.setAttribute("sandbox", sandbox);
  1138. if (frameElement.tagName == "OBJECT") {
  1139. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1140. } else {
  1141. if (frameElement.tagName == "FRAME") {
  1142. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1143. } else {
  1144. frameElement.setAttribute("srcdoc", pageData.content);
  1145. frameElement.removeAttribute("src");
  1146. }
  1147. }
  1148. this.stats.addAll(pageData);
  1149. } else {
  1150. this.stats.add("discarded", "frames", 1);
  1151. }
  1152. }
  1153. }
  1154. }));
  1155. }
  1156. }
  1157. async processHtmlImports() {
  1158. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1159. await Promise.all(linkElements.map(async linkElement => {
  1160. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1161. if (attributeValue) {
  1162. const importData = this.options.imports[Number(attributeValue)];
  1163. if (importData.runner) {
  1164. this.stats.add("processed", "HTML imports", 1);
  1165. await importData.runner.run();
  1166. const pageData = await importData.runner.getPageData();
  1167. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1168. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1169. this.stats.addAll(pageData);
  1170. } else {
  1171. this.stats.add("discarded", "HTML imports", 1);
  1172. }
  1173. }
  1174. }));
  1175. }
  1176. replaceStylesheets() {
  1177. this.doc.querySelectorAll("style").forEach(styleElement => {
  1178. const stylesheetInfo = this.stylesheets.get(styleElement);
  1179. if (stylesheetInfo) {
  1180. this.stylesheets.delete(styleElement);
  1181. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1182. styleElement.textContent = stylesheetContent;
  1183. if (stylesheetInfo.mediaText) {
  1184. styleElement.media = stylesheetInfo.mediaText;
  1185. }
  1186. } else {
  1187. styleElement.remove();
  1188. }
  1189. });
  1190. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1191. const stylesheetInfo = this.stylesheets.get(linkElement);
  1192. if (stylesheetInfo) {
  1193. this.stylesheets.delete(linkElement);
  1194. const styleElement = this.doc.createElement("style");
  1195. if (stylesheetInfo.mediaText) {
  1196. styleElement.media = stylesheetInfo.mediaText;
  1197. }
  1198. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1199. styleElement.textContent = stylesheetContent;
  1200. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1201. } else {
  1202. linkElement.remove();
  1203. }
  1204. });
  1205. }
  1206. replaceStyleAttributes() {
  1207. this.doc.querySelectorAll("[style]").forEach(async element => {
  1208. const declarations = this.styles.get(element);
  1209. if (declarations) {
  1210. this.styles.delete(element);
  1211. let styleContent = cssTree.generate(declarations);
  1212. element.setAttribute("style", styleContent);
  1213. } else {
  1214. element.setAttribute("style", "");
  1215. }
  1216. });
  1217. }
  1218. insertVariables() {
  1219. if (this.cssVariables.size) {
  1220. const styleElement = this.doc.createElement("style");
  1221. const firstStyleElement = this.doc.head.querySelector("style");
  1222. if (firstStyleElement) {
  1223. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1224. } else {
  1225. this.doc.head.appendChild(styleElement);
  1226. }
  1227. let stylesheetContent = "";
  1228. this.cssVariables.forEach((content, indexResource) => {
  1229. this.cssVariables.delete(indexResource);
  1230. if (stylesheetContent) {
  1231. stylesheetContent += ";";
  1232. }
  1233. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1234. });
  1235. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1236. }
  1237. }
  1238. compressHTML() {
  1239. let size;
  1240. if (this.options.displayStats) {
  1241. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1242. }
  1243. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1244. if (this.options.displayStats) {
  1245. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1246. }
  1247. }
  1248. cleanupPage() {
  1249. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1250. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1251. if (metaCharset) {
  1252. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1253. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1254. this.doc.head.querySelector("meta[charset]").remove();
  1255. }
  1256. }
  1257. }
  1258. async insertMAFFMetaData() {
  1259. const maffMetaData = await this.maffMetaDataPromise;
  1260. if (maffMetaData && maffMetaData.content) {
  1261. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1262. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1263. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1264. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1265. if (originalURLElement) {
  1266. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1267. }
  1268. if (archiveTimeElement) {
  1269. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1270. if (value) {
  1271. const date = new Date(value);
  1272. if (!isNaN(date.getTime())) {
  1273. this.options.saveDate = new Date(value);
  1274. }
  1275. }
  1276. }
  1277. }
  1278. }
  1279. async setDocInfo() {
  1280. const titleElement = this.doc.querySelector("title");
  1281. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1282. const authorElement = this.doc.querySelector("meta[name=author]");
  1283. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1284. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1285. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1286. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1287. this.options.info = {
  1288. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1289. lang: this.doc.documentElement.lang,
  1290. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1291. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1292. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : ""
  1293. };
  1294. }
  1295. }
  1296. // ---------------
  1297. // ProcessorHelper
  1298. // ---------------
  1299. const DATA_URI_PREFIX = "data:";
  1300. const ABOUT_BLANK_URI = "about:blank";
  1301. const EMPTY_DATA_URI = "data:null;base64,";
  1302. const REGEXP_URL_HASH = /(#.+?)$/;
  1303. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1304. const VAR_MAX_SIZE = 1024 * 1024 * 2;
  1305. class ProcessorHelper {
  1306. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1307. const date = options.saveDate;
  1308. const url = util.parseURL(options.saveUrl);
  1309. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1310. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1311. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1312. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1313. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1314. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1315. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1316. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1317. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1318. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1319. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1320. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1321. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1322. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1323. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1324. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1325. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1326. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1327. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1328. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1329. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1330. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1331. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1332. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1333. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1334. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1335. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1336. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1337. template = await Util.evalTemplateVariable(template, "url-href", () => decodeURI(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1338. template = await Util.evalTemplateVariable(template, "url-referrer", () => decodeURI(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1339. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1340. template = await Util.evalTemplateVariable(template, "url-pathname", () => decodeURI(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1341. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1342. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1343. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1344. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1345. for (const [name, value] of params) {
  1346. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1347. }
  1348. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1349. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1350. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1351. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1352. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decodeURI(Util.getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1353. if (content) {
  1354. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1357. }
  1358. return template.trim();
  1359. }
  1360. static setBackgroundImage(element, url, style) {
  1361. element.style.setProperty("background-blend-mode", "normal", "important");
  1362. element.style.setProperty("background-clip", "content-box", "important");
  1363. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1364. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1365. element.style.setProperty("background-image", url, "important");
  1366. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1367. element.style.setProperty("background-origin", "content-box", "important");
  1368. element.style.setProperty("background-repeat", "no-repeat", "important");
  1369. }
  1370. static processShortcutIcons(doc) {
  1371. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1372. if (!shortcutIcon) {
  1373. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1374. if (shortcutIcon) {
  1375. shortcutIcon.rel = "icon";
  1376. }
  1377. }
  1378. if (shortcutIcon) {
  1379. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1380. if (linkElement != shortcutIcon) {
  1381. linkElement.remove();
  1382. }
  1383. });
  1384. }
  1385. }
  1386. static removeSingleLineCssComments(stylesheet) {
  1387. const removedRules = [];
  1388. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1389. const ruleData = cssRule.data;
  1390. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1391. removedRules.push(cssRule);
  1392. }
  1393. }
  1394. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1395. }
  1396. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1397. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1398. const imports = Util.getImportFunctions(stylesheetContent);
  1399. await Promise.all(imports.map(async cssImport => {
  1400. const match = Util.matchImport(cssImport);
  1401. if (match) {
  1402. const regExpCssImport = Util.getRegExp(cssImport);
  1403. let resourceURL = Util.normalizeURL(match.resourceURL);
  1404. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1405. try {
  1406. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1407. } catch (error) {
  1408. // ignored
  1409. }
  1410. if (Util.testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1411. const content = await util.getContent(resourceURL, {
  1412. maxResourceSize: options.maxResourceSize,
  1413. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1414. validateTextContentType: true,
  1415. frameId: options.windowId
  1416. });
  1417. resourceURL = content.resourceURL;
  1418. content.data = Util.getUpdatedResourceContent(resourceURL, content, options);
  1419. let importedStylesheetContent = Util.removeCssComments(content.data);
  1420. if (options.compressCSS) {
  1421. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1422. }
  1423. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1424. if (stylesheetContent.includes(cssImport)) {
  1425. const ancestorStyleSheets = new Set(importedStyleSheets);
  1426. ancestorStyleSheets.add(resourceURL);
  1427. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1428. workStylesheet.textContent = importedStylesheetContent;
  1429. if (workStylesheet.sheet.cssRules.length) {
  1430. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1431. } else {
  1432. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1433. }
  1434. }
  1435. } else {
  1436. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1437. }
  1438. } else {
  1439. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1440. }
  1441. }
  1442. }));
  1443. return stylesheetContent;
  1444. }
  1445. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1446. const urlFunctions = Util.getUrlFunctions(stylesheetContent, true);
  1447. urlFunctions.map(urlFunction => {
  1448. const originalResourceURL = Util.matchURL(urlFunction);
  1449. const resourceURL = Util.normalizeURL(originalResourceURL);
  1450. if (!Util.testIgnoredPath(resourceURL)) {
  1451. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1452. let resolvedURL;
  1453. if (!originalResourceURL.startsWith("#")) {
  1454. try {
  1455. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1456. } catch (error) {
  1457. // ignored
  1458. }
  1459. }
  1460. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1461. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1462. }
  1463. } else {
  1464. let newUrlFunction;
  1465. if (originalResourceURL) {
  1466. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1467. } else {
  1468. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1469. }
  1470. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1471. }
  1472. }
  1473. });
  1474. return stylesheetContent;
  1475. }
  1476. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1477. resourceURL = Util.normalizeURL(resourceURL);
  1478. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1479. const content = await util.getContent(resourceURL, {
  1480. maxResourceSize: options.maxResourceSize,
  1481. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1482. charset: options.charset,
  1483. frameId: options.windowId
  1484. });
  1485. resourceURL = content.resourceURL;
  1486. content.data = Util.getUpdatedResourceContent(content.resourceURL, content, options);
  1487. let stylesheetContent = Util.removeCssComments(content.data);
  1488. if (options.compressCSS) {
  1489. stylesheetContent = util.compressCSS(stylesheetContent);
  1490. }
  1491. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1492. return stylesheetContent;
  1493. }
  1494. }
  1495. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1496. const promises = [];
  1497. const removedRules = [];
  1498. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1499. const ruleData = cssRule.data;
  1500. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1501. removedRules.push(cssRule);
  1502. } else if (ruleData.block && ruleData.block.children) {
  1503. if (ruleData.type == "Rule") {
  1504. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1505. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1506. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1507. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1508. promises.push(processFontFaceRule(ruleData));
  1509. }
  1510. }
  1511. }
  1512. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1513. await Promise.all(promises);
  1514. async function processFontFaceRule(ruleData) {
  1515. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1516. if (declaration.type == "Declaration" && declaration.value.children) {
  1517. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value), true);
  1518. await Promise.all(urlFunctions.map(async urlFunction => {
  1519. const originalResourceURL = Util.matchURL(urlFunction);
  1520. const resourceURL = Util.normalizeURL(originalResourceURL);
  1521. if (!Util.testIgnoredPath(resourceURL)) {
  1522. if (Util.testValidURL(resourceURL)) {
  1523. let { content } = await batchRequest.addURL(resourceURL, true);
  1524. replaceURLs(declaration, originalResourceURL, content);
  1525. }
  1526. }
  1527. }));
  1528. }
  1529. }));
  1530. function replaceURLs(declaration, oldURL, newURL) {
  1531. declaration.value.children.forEach(token => {
  1532. if (token.type == "Url" && util.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1533. token.value.value = newURL;
  1534. }
  1535. });
  1536. }
  1537. }
  1538. }
  1539. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1540. await Promise.all(declarations.map(async declaration => {
  1541. if (declaration.type == "Declaration" && declaration.value.children) {
  1542. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1543. await Promise.all(urlFunctions.map(async urlFunction => {
  1544. const originalResourceURL = Util.matchURL(urlFunction);
  1545. const resourceURL = Util.normalizeURL(originalResourceURL);
  1546. if (!Util.testIgnoredPath(resourceURL)) {
  1547. if (Util.testValidURL(resourceURL)) {
  1548. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1549. let variableDefined;
  1550. const tokens = [];
  1551. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1552. if (!originalResourceURL.startsWith("#")) {
  1553. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < VAR_MAX_SIZE) {
  1554. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1555. tokens.push({ parent, token, value });
  1556. variableDefined = true;
  1557. } else {
  1558. token.data.value.value = content;
  1559. }
  1560. }
  1561. });
  1562. if (variableDefined) {
  1563. cssVariables.set(indexResource, content);
  1564. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1565. }
  1566. }
  1567. }
  1568. }));
  1569. }
  1570. }));
  1571. function findURLToken(url, children, callback, depth = 0) {
  1572. for (let token = children.head; token; token = token.next) {
  1573. if (token.data.children) {
  1574. findURLToken(url, token.data.children, callback, depth + 1);
  1575. }
  1576. if (token.data.type == "Url" && util.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1577. callback(token, children, depth == 0);
  1578. }
  1579. }
  1580. }
  1581. }
  1582. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1583. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1584. let resourceURL = resourceElement.getAttribute(attributeName);
  1585. if (resourceURL != null) {
  1586. resourceURL = Util.normalizeURL(resourceURL);
  1587. if (!Util.testIgnoredPath(resourceURL)) {
  1588. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1589. if (Util.testValidPath(resourceURL)) {
  1590. try {
  1591. resourceURL = util.resolveURL(resourceURL, baseURI);
  1592. } catch (error) {
  1593. // ignored
  1594. }
  1595. if (Util.testValidURL(resourceURL)) {
  1596. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1597. if (resourceElement.dataset.singleFileOriginURL) {
  1598. let originURL = resourceElement.dataset.singleFileOriginURL;
  1599. delete resourceElement.dataset.singleFileOriginURL;
  1600. if (content == EMPTY_DATA_URI) {
  1601. try {
  1602. originURL = util.resolveURL(originURL, baseURI);
  1603. } catch (error) {
  1604. // ignored
  1605. }
  1606. try {
  1607. resourceURL = originURL;
  1608. content = (await util.getContent(resourceURL, {
  1609. asBinary: true,
  1610. maxResourceSize: options.maxResourceSize,
  1611. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1612. frameId: options.windowId
  1613. })).data;
  1614. } catch (error) {
  1615. // ignored
  1616. }
  1617. }
  1618. }
  1619. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1620. resourceElement.remove();
  1621. } else {
  1622. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1623. if (!forbiddenPrefixFound) {
  1624. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1625. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < VAR_MAX_SIZE) {
  1626. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1627. cssVariables.set(indexResource, content);
  1628. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1629. styles.set(resourceElement, declarationList);
  1630. } else {
  1631. resourceElement.setAttribute(attributeName, content);
  1632. }
  1633. } else {
  1634. resourceElement.setAttribute(attributeName, content);
  1635. }
  1636. }
  1637. }
  1638. }
  1639. }
  1640. }
  1641. }
  1642. }));
  1643. }
  1644. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1645. const attributeName = "xlink:href";
  1646. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1647. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1648. let resourceURL = Util.normalizeURL(originalResourceURL);
  1649. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1650. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1651. try {
  1652. resourceURL = util.resolveURL(resourceURL, baseURI);
  1653. } catch (error) {
  1654. // ignored
  1655. }
  1656. if (Util.testValidURL(resourceURL)) {
  1657. const { content } = await batchRequest.addURL(resourceURL);
  1658. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1659. if (hashMatch && hashMatch[0]) {
  1660. let symbolElement;
  1661. try {
  1662. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1663. } catch (error) {
  1664. // ignored
  1665. }
  1666. if (symbolElement) {
  1667. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1668. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1669. }
  1670. } else {
  1671. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1672. }
  1673. }
  1674. } else if (resourceURL == options.url) {
  1675. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1676. }
  1677. }));
  1678. }
  1679. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1680. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1681. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1682. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1683. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1684. if (!Util.testIgnoredPath(resourceURL)) {
  1685. if (Util.testValidPath(resourceURL)) {
  1686. try {
  1687. resourceURL = util.resolveURL(resourceURL, baseURI);
  1688. } catch (error) {
  1689. // ignored
  1690. }
  1691. if (Util.testValidURL(resourceURL)) {
  1692. const { content } = await batchRequest.addURL(resourceURL, true);
  1693. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1694. if (forbiddenPrefixFound) {
  1695. return "";
  1696. }
  1697. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1698. } else {
  1699. return "";
  1700. }
  1701. } else {
  1702. return "";
  1703. }
  1704. } else {
  1705. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1706. }
  1707. }));
  1708. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1709. }));
  1710. }
  1711. static replaceImageSource(imgElement, variableName, options) {
  1712. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1713. if (attributeValue) {
  1714. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1715. if (imageData && imageData.replaceable) {
  1716. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1717. const backgroundStyle = {};
  1718. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1719. if (backgroundSize) {
  1720. backgroundStyle["background-size"] = imageData.objectFit;
  1721. }
  1722. if (imageData.objectPosition) {
  1723. backgroundStyle["background-position"] = imageData.objectPosition;
  1724. }
  1725. if (imageData.backgroundColor) {
  1726. backgroundStyle["background-color"] = imageData.backgroundColor;
  1727. }
  1728. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1729. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1730. return true;
  1731. }
  1732. }
  1733. }
  1734. }
  1735. // ----
  1736. // Util
  1737. // ----
  1738. const BLOB_URI_PREFIX = "blob:";
  1739. const HTTP_URI_PREFIX = /^https?:\/\//;
  1740. const FILE_URI_PREFIX = /^file:\/\//;
  1741. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1742. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1743. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1744. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1745. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1746. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1747. 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;
  1748. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1749. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1750. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1751. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1752. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1753. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1754. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1755. class Util {
  1756. static getUpdatedResourceContent(resourceURL, content, options) {
  1757. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1758. options.updatedResources[resourceURL].retrieved = true;
  1759. return options.updatedResources[resourceURL].content;
  1760. } else {
  1761. return content.data || "";
  1762. }
  1763. }
  1764. static normalizeURL(url) {
  1765. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1766. return url;
  1767. } else {
  1768. return url.split("#")[0];
  1769. }
  1770. }
  1771. static getCSSValue(value) {
  1772. let result = "";
  1773. try {
  1774. result = cssTree.generate(value);
  1775. } catch (error) {
  1776. // ignored
  1777. }
  1778. return result;
  1779. }
  1780. static getOnEventAttributeNames(doc) {
  1781. const element = doc.createElement("div");
  1782. const attributeNames = [];
  1783. for (const propertyName in element) {
  1784. if (propertyName.startsWith("on")) {
  1785. attributeNames.push(propertyName);
  1786. }
  1787. }
  1788. attributeNames.push("onunload");
  1789. return attributeNames;
  1790. }
  1791. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1792. const replaceRegExp = new RegExp("{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}", "g");
  1793. if (template && template.match(replaceRegExp)) {
  1794. let value = await valueGetter();
  1795. if (!dontReplaceSlash) {
  1796. value = value.replace(/\/+/g, replacementCharacter);
  1797. }
  1798. return template.replace(replaceRegExp, value);
  1799. }
  1800. return template;
  1801. }
  1802. static getLastSegment(url, replacementCharacter) {
  1803. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1804. if (!lastSegment) {
  1805. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1806. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1807. }
  1808. if (!lastSegment) {
  1809. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1810. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1811. }
  1812. if (!lastSegment) {
  1813. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1814. }
  1815. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1816. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1817. lastSegment = lastSegmentMatch[1];
  1818. }
  1819. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1820. return lastSegment;
  1821. }
  1822. static getRegExp(string) {
  1823. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1824. }
  1825. static getUrlFunctions(stylesheetContent, unique) {
  1826. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1827. if (unique) {
  1828. return [...new Set(result)];
  1829. } else {
  1830. return result;
  1831. }
  1832. }
  1833. static getImportFunctions(stylesheetContent) {
  1834. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1835. }
  1836. static findShortcutIcon(shortcutIcons) {
  1837. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1838. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1839. return shortcutIcons[0];
  1840. }
  1841. static matchURL(stylesheetContent) {
  1842. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1843. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1844. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1845. return match && match[1];
  1846. }
  1847. static testIgnoredPath(resourceURL) {
  1848. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1849. }
  1850. static testValidPath(resourceURL) {
  1851. return resourceURL && !resourceURL.match(EMPTY_URL);
  1852. }
  1853. static testValidURL(resourceURL) {
  1854. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1855. }
  1856. static matchImport(stylesheetContent) {
  1857. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1858. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1859. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1860. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1861. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1862. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1863. if (match) {
  1864. const [, resourceURL, media] = match;
  1865. return { resourceURL, media };
  1866. }
  1867. }
  1868. static removeCssComments(stylesheetContent) {
  1869. let start, end;
  1870. do {
  1871. start = stylesheetContent.indexOf("/*");
  1872. end = stylesheetContent.indexOf("*/", start + 2);
  1873. if (start != -1 && end != -1) {
  1874. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1875. }
  1876. } while (start != -1 && end != -1);
  1877. return stylesheetContent;
  1878. }
  1879. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1880. if (mediaQuery) {
  1881. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1882. } else {
  1883. return stylesheetContent;
  1884. }
  1885. }
  1886. }
  1887. function log(...args) {
  1888. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1889. }
  1890. // -----
  1891. // Stats
  1892. // -----
  1893. const STATS_DEFAULT_VALUES = {
  1894. discarded: {
  1895. "HTML bytes": 0,
  1896. "hidden elements": 0,
  1897. "HTML imports": 0,
  1898. scripts: 0,
  1899. objects: 0,
  1900. "audio sources": 0,
  1901. "video sources": 0,
  1902. frames: 0,
  1903. "CSS rules": 0,
  1904. canvas: 0,
  1905. stylesheets: 0,
  1906. resources: 0,
  1907. medias: 0
  1908. },
  1909. processed: {
  1910. "HTML bytes": 0,
  1911. "hidden elements": 0,
  1912. "HTML imports": 0,
  1913. scripts: 0,
  1914. objects: 0,
  1915. "audio sources": 0,
  1916. "video sources": 0,
  1917. frames: 0,
  1918. "CSS rules": 0,
  1919. canvas: 0,
  1920. stylesheets: 0,
  1921. resources: 0,
  1922. medias: 0
  1923. }
  1924. };
  1925. class Stats {
  1926. constructor(options) {
  1927. this.options = options;
  1928. if (options.displayStats) {
  1929. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1930. }
  1931. }
  1932. set(type, subType, value) {
  1933. if (this.options.displayStats) {
  1934. this.data[type][subType] = value;
  1935. }
  1936. }
  1937. add(type, subType, value) {
  1938. if (this.options.displayStats) {
  1939. this.data[type][subType] += value;
  1940. }
  1941. }
  1942. addAll(pageData) {
  1943. if (this.options.displayStats) {
  1944. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1945. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1946. }
  1947. }
  1948. }
  1949. return { getClass };
  1950. })();