single-file-core.js 87 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139
  1. /*
  2. * Copyright 2010-2020 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. if (options.sessionId === undefined) {
  34. options.sessionId = sessionId;
  35. sessionId++;
  36. }
  37. }
  38. async run() {
  39. if (this.options.userScriptEnabled) {
  40. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  41. }
  42. this.runner = new Runner(this.options, true);
  43. await this.runner.loadPage();
  44. await this.runner.initialize();
  45. if (this.options.userScriptEnabled) {
  46. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  47. }
  48. await this.runner.run();
  49. }
  50. cancel() {
  51. this.cancelled = true;
  52. if (this.runner) {
  53. this.runner.cancel();
  54. }
  55. }
  56. getPageData() {
  57. return this.runner.getPageData();
  58. }
  59. }
  60. // -------------
  61. // ProgressEvent
  62. // -------------
  63. const PAGE_LOADING = "page-loading";
  64. const PAGE_LOADED = "page-loaded";
  65. const RESOURCES_INITIALIZING = "resource-initializing";
  66. const RESOURCES_INITIALIZED = "resources-initialized";
  67. const RESOURCE_LOADED = "resource-loaded";
  68. const PAGE_ENDED = "page-ended";
  69. const STAGE_STARTED = "stage-started";
  70. const STAGE_ENDED = "stage-ended";
  71. const STAGE_TASK_STARTED = "stage-task-started";
  72. const STAGE_TASK_ENDED = "stage-task-ended";
  73. class ProgressEvent {
  74. constructor(type, detail) {
  75. 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 };
  76. }
  77. }
  78. // ------
  79. // Runner
  80. // ------
  81. const RESOLVE_URLS_STAGE = 0;
  82. const REPLACE_DATA_STAGE = 1;
  83. const REPLACE_DOCS_STAGE = 2;
  84. const POST_PROCESS_STAGE = 3;
  85. const STAGES = [{
  86. sequential: [
  87. { action: "preProcessPage" },
  88. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  89. { action: "replaceStyleContents" },
  90. { action: "resetCharsetMeta" },
  91. { option: "saveFavicon", action: "saveFavicon" },
  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. const rootDocDefined = root && options.doc;
  149. this.root = root;
  150. this.options = options;
  151. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  152. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  153. this.options.rootDocument = root;
  154. this.options.updatedResources = this.options.updatedResources || {};
  155. this.batchRequest = new BatchRequest();
  156. this.processor = new Processor(options, this.batchRequest);
  157. if (rootDocDefined) {
  158. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  159. this.options.canvases = docData.canvases;
  160. this.options.fonts = docData.fonts;
  161. this.options.stylesheets = docData.stylesheets;
  162. this.options.images = docData.images;
  163. this.options.posters = docData.posters;
  164. this.options.usedFonts = docData.usedFonts;
  165. this.options.shadowRoots = docData.shadowRoots;
  166. this.options.imports = docData.imports;
  167. this.options.referrer = docData.referrer;
  168. this.markedElements = docData.markedElements;
  169. }
  170. if (this.options.saveRawPage) {
  171. this.options.removeFrames = true;
  172. }
  173. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  174. this.onprogress = options.onprogress || (() => { });
  175. }
  176. async loadPage() {
  177. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  178. await this.processor.loadPage(this.options.content);
  179. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  180. }
  181. async initialize() {
  182. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  183. await this.executeStage(RESOLVE_URLS_STAGE);
  184. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  185. if (this.root && this.options.doc) {
  186. util.postProcessDoc(this.options.doc, this.markedElements);
  187. }
  188. }
  189. cancel() {
  190. this.cancelled = true;
  191. this.batchRequest.cancel();
  192. if (this.root) {
  193. if (this.options.frames) {
  194. this.options.frames.forEach(cancelRunner);
  195. }
  196. if (this.options.imports) {
  197. this.options.imports.forEach(cancelRunner);
  198. }
  199. }
  200. function cancelRunner(resourceData) {
  201. if (resourceData.runner) {
  202. resourceData.runner.cancel();
  203. }
  204. }
  205. }
  206. async run() {
  207. if (this.root) {
  208. this.processor.initialize(this.batchRequest);
  209. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  210. }
  211. await this.batchRequest.run(detail => {
  212. detail.pageURL = this.options.url;
  213. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  214. }, this.options);
  215. await this.pendingPromises;
  216. this.options.doc = null;
  217. this.options.win = null;
  218. await this.executeStage(REPLACE_DOCS_STAGE);
  219. await this.executeStage(POST_PROCESS_STAGE);
  220. this.processor.finalize();
  221. }
  222. getDocument() {
  223. return this.processor.doc;
  224. }
  225. getStyleSheets() {
  226. return this.processor.stylesheets;
  227. }
  228. getPageData() {
  229. if (this.root) {
  230. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  231. }
  232. return this.processor.getPageData();
  233. }
  234. async executeStage(step) {
  235. if (DEBUG) {
  236. log("**** STARTED STAGE", step, "****");
  237. }
  238. const frame = !this.root;
  239. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  240. STAGES[step].sequential.forEach(task => {
  241. let startTime;
  242. if (DEBUG) {
  243. startTime = Date.now();
  244. log(" -- STARTED task =", task.action);
  245. }
  246. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  247. if (!this.cancelled) {
  248. this.executeTask(task);
  249. }
  250. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  251. if (DEBUG) {
  252. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  253. }
  254. });
  255. let parallelTasksPromise;
  256. if (STAGES[step].parallel) {
  257. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  258. let startTime;
  259. if (DEBUG) {
  260. startTime = Date.now();
  261. log(" // STARTED task =", task.action);
  262. }
  263. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  264. if (!this.cancelled) {
  265. await this.executeTask(task);
  266. }
  267. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  268. if (DEBUG) {
  269. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  270. }
  271. }));
  272. } else {
  273. parallelTasksPromise = Promise.resolve();
  274. }
  275. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  276. if (DEBUG) {
  277. log("**** ENDED STAGE", step, "****");
  278. }
  279. return parallelTasksPromise;
  280. }
  281. executeTask(task) {
  282. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  283. return this.processor[task.action]();
  284. }
  285. }
  286. }
  287. // ------------
  288. // BatchRequest
  289. // ------------
  290. class BatchRequest {
  291. constructor() {
  292. this.requests = new Map();
  293. this.duplicates = new Map();
  294. }
  295. addURL(resourceURL, asBinary, groupDuplicates) {
  296. return new Promise((resolve, reject) => {
  297. const requestKey = JSON.stringify([resourceURL, asBinary]);
  298. let resourceRequests = this.requests.get(requestKey);
  299. if (!resourceRequests) {
  300. resourceRequests = [];
  301. this.requests.set(requestKey, resourceRequests);
  302. }
  303. const callbacks = { resolve, reject };
  304. resourceRequests.push(callbacks);
  305. if (groupDuplicates) {
  306. let duplicateRequests = this.duplicates.get(requestKey);
  307. if (!duplicateRequests) {
  308. duplicateRequests = [];
  309. this.duplicates.set(requestKey, duplicateRequests);
  310. }
  311. duplicateRequests.push(callbacks);
  312. }
  313. });
  314. }
  315. getMaxResources() {
  316. return this.requests.size;
  317. }
  318. run(onloadListener, options) {
  319. const resourceURLs = [...this.requests.keys()];
  320. let indexResource = 0;
  321. return Promise.all(resourceURLs.map(async requestKey => {
  322. const [resourceURL, asBinary] = JSON.parse(requestKey);
  323. const resourceRequests = this.requests.get(requestKey);
  324. try {
  325. const content = await util.getContent(resourceURL, {
  326. asBinary,
  327. maxResourceSize: options.maxResourceSize,
  328. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  329. frameId: options.windowId
  330. });
  331. indexResource = indexResource + 1;
  332. onloadListener({ url: resourceURL });
  333. if (!this.cancelled) {
  334. resourceRequests.forEach(callbacks => {
  335. const duplicateCallbacks = this.duplicates.get(requestKey);
  336. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  337. callbacks.resolve({ content: content.data, indexResource, duplicate });
  338. });
  339. }
  340. } catch (error) {
  341. indexResource = indexResource + 1;
  342. onloadListener({ url: resourceURL });
  343. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  344. }
  345. this.requests.delete(requestKey);
  346. }));
  347. }
  348. cancel() {
  349. this.cancelled = true;
  350. const resourceURLs = [...this.requests.keys()];
  351. resourceURLs.forEach(requestKey => {
  352. const resourceRequests = this.requests.get(requestKey);
  353. resourceRequests.forEach(callbacks => callbacks.reject());
  354. this.requests.delete(requestKey);
  355. });
  356. }
  357. }
  358. // ---------
  359. // Processor
  360. // ---------
  361. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  362. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  363. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  364. const SCRIPT_TAG_FOUND = /<script/gi;
  365. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  366. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  367. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  368. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  369. const UTF8_CHARSET = "utf-8";
  370. class Processor {
  371. constructor(options, batchRequest) {
  372. this.options = options;
  373. this.stats = new Stats(options);
  374. this.baseURI = normalizeURL(options.baseURI || options.url);
  375. this.batchRequest = batchRequest;
  376. this.stylesheets = new Map();
  377. this.styles = new Map();
  378. this.cssVariables = new Map();
  379. }
  380. initialize() {
  381. this.options.saveDate = new Date();
  382. this.options.saveUrl = this.options.url;
  383. if (this.options.enableMaff) {
  384. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  385. }
  386. this.maxResources = this.batchRequest.getMaxResources();
  387. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  388. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  389. }
  390. if (!this.options.removeImports && this.options.imports) {
  391. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  392. }
  393. this.stats.set("processed", "resources", this.maxResources);
  394. }
  395. async loadPage(pageContent, charset) {
  396. let content;
  397. if (!pageContent || this.options.saveRawPage) {
  398. content = await util.getContent(this.baseURI, {
  399. maxResourceSize: this.options.maxResourceSize,
  400. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  401. charset,
  402. frameId: this.options.windowId
  403. });
  404. pageContent = content.data;
  405. }
  406. this.doc = util.parseDocContent(pageContent, this.baseURI);
  407. if (this.options.saveRawPage) {
  408. let charset;
  409. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  410. const charsetDeclaration = element.content.split(";")[1];
  411. if (charsetDeclaration && !charset) {
  412. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  413. }
  414. });
  415. if (charset && charset != content.charset) {
  416. return this.loadPage(pageContent, charset);
  417. }
  418. }
  419. this.workStyleElement = this.doc.createElement("style");
  420. this.doc.body.appendChild(this.workStyleElement);
  421. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  422. }
  423. finalize() {
  424. if (this.workStyleElement.parentNode) {
  425. this.workStyleElement.remove();
  426. }
  427. }
  428. async getPageData() {
  429. util.postProcessDoc(this.doc);
  430. const url = util.parseURL(this.baseURI);
  431. if (this.options.insertSingleFileComment) {
  432. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  433. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  434. " \n url: " + this.options.saveUrl +
  435. " \n saved date: " + this.options.saveDate +
  436. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  437. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  438. }
  439. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  440. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  441. if (!canonicalLink) {
  442. canonicalLink = this.doc.createElement("link");
  443. canonicalLink.setAttribute("rel", "canonical");
  444. this.doc.head.appendChild(canonicalLink);
  445. }
  446. if (canonicalLink && !canonicalLink.href) {
  447. canonicalLink.href = this.options.saveUrl;
  448. }
  449. }
  450. let size;
  451. if (this.options.displayStats) {
  452. size = util.getContentSize(this.doc.documentElement.outerHTML);
  453. }
  454. const content = util.serialize(this.doc, this.options.compressHTML);
  455. if (this.options.displayStats) {
  456. const contentSize = util.getContentSize(content);
  457. this.stats.set("processed", "HTML bytes", contentSize);
  458. this.stats.add("discarded", "HTML bytes", size - contentSize);
  459. }
  460. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  461. const replacementCharacter = this.options.filenameReplacementCharacter;
  462. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  463. if (!this.options.backgroundSave) {
  464. filename = filename.replace(/\//g, replacementCharacter);
  465. }
  466. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  467. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  468. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  469. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  470. filename = filename + "…" + extension;
  471. }
  472. if (!filename) {
  473. filename = "Unnamed page";
  474. }
  475. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  476. const pageData = {
  477. stats: this.stats.data,
  478. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  479. filename,
  480. content
  481. };
  482. if (this.options.addProof) {
  483. pageData.hash = await util.digest("SHA-256", content);
  484. }
  485. if (this.options.retrieveLinks) {
  486. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  487. }
  488. return pageData;
  489. }
  490. preProcessPage() {
  491. if (this.options.win) {
  492. 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));
  493. }
  494. if (this.options.images && !this.options.saveRawPage) {
  495. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  496. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  497. if (attributeValue) {
  498. const imageData = this.options.images[Number(attributeValue)];
  499. if (imageData) {
  500. if (this.options.removeHiddenElements && (
  501. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  502. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  503. )) {
  504. imgElement.setAttribute("src", EMPTY_IMAGE);
  505. } else {
  506. if (imageData.currentSrc) {
  507. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  508. imgElement.setAttribute("src", imageData.currentSrc);
  509. }
  510. if (this.options.loadDeferredImages) {
  511. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  512. imageData.src = imgElement.dataset.src;
  513. imgElement.setAttribute("src", imgElement.dataset.src);
  514. imgElement.removeAttribute("data-src");
  515. }
  516. }
  517. }
  518. }
  519. }
  520. });
  521. if (this.options.loadDeferredImages) {
  522. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  523. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  524. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  525. imgElement.removeAttribute("data-srcset");
  526. }
  527. });
  528. }
  529. }
  530. }
  531. replaceStyleContents() {
  532. if (this.options.stylesheets) {
  533. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  534. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  535. if (attributeValue) {
  536. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  537. if (stylesheetContent) {
  538. styleElement.textContent = stylesheetContent;
  539. }
  540. }
  541. });
  542. }
  543. }
  544. removeUnselectedElements() {
  545. removeUnmarkedElements(this.doc.body);
  546. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  547. function removeUnmarkedElements(element) {
  548. let selectedElementFound = false;
  549. Array.from(element.childNodes).forEach(node => {
  550. if (node.nodeType == 1) {
  551. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  552. selectedElementFound = selectedElementFound || isSelectedElement;
  553. if (isSelectedElement) {
  554. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  555. removeUnmarkedElements(node);
  556. } else if (selectedElementFound) {
  557. removeNode(node);
  558. } else {
  559. hideNode(node);
  560. }
  561. }
  562. });
  563. }
  564. function removeNode(node) {
  565. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  566. node.remove();
  567. } else {
  568. hideNode(node);
  569. }
  570. }
  571. function hideNode(node) {
  572. if (canHideNode(node)) {
  573. node.style.setProperty("display", "none", "important");
  574. Array.from(node.childNodes).forEach(removeNode);
  575. }
  576. }
  577. function canHideNode(node) {
  578. if (node.nodeType == 1) {
  579. const tagName = node.tagName && node.tagName.toLowerCase();
  580. return (tagName != "svg" && tagName != "style" && tagName != "link");
  581. }
  582. }
  583. }
  584. insertVideoPosters() {
  585. if (this.options.posters) {
  586. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  587. let videoElement;
  588. if (element.tagName == "VIDEO") {
  589. videoElement = element;
  590. } else {
  591. videoElement = element.parentElement;
  592. }
  593. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  594. if (attributeValue) {
  595. const posterURL = this.options.posters[Number(attributeValue)];
  596. if (!videoElement.poster && posterURL) {
  597. videoElement.setAttribute("poster", posterURL);
  598. }
  599. }
  600. });
  601. }
  602. }
  603. removeFrames() {
  604. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  605. this.stats.set("discarded", "frames", frameElements.length);
  606. this.stats.set("processed", "frames", frameElements.length);
  607. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  608. }
  609. removeScripts() {
  610. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  611. this.doc.querySelectorAll("[href]").forEach(element => {
  612. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  613. element.setAttribute("href", "");
  614. }
  615. });
  616. this.doc.querySelectorAll("[src]").forEach(element => {
  617. if (element.src && element.src.match(/^\s*javascript:/)) {
  618. element.removeAttribute("src");
  619. }
  620. });
  621. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  622. this.stats.set("discarded", "scripts", scriptElements.length);
  623. this.stats.set("processed", "scripts", scriptElements.length);
  624. scriptElements.forEach(element => element.remove());
  625. }
  626. removeVideoSources() {
  627. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  628. this.stats.set("discarded", "video sources", videoSourceElements.length);
  629. this.stats.set("processed", "video sources", videoSourceElements.length);
  630. videoSourceElements.forEach(element => {
  631. if (element.tagName == "SOURCE") {
  632. element.remove();
  633. } else {
  634. videoSourceElements.forEach(element => element.removeAttribute("src"));
  635. }
  636. });
  637. }
  638. removeAudioSources() {
  639. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  640. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  641. this.stats.set("processed", "audio sources", audioSourceElements.length);
  642. audioSourceElements.forEach(element => {
  643. if (element.tagName == "SOURCE") {
  644. element.remove();
  645. } else {
  646. audioSourceElements.forEach(element => element.removeAttribute("src"));
  647. }
  648. });
  649. }
  650. removeDiscardedResources() {
  651. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  652. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  653. 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\"])");
  654. this.stats.set("discarded", "objects", objectElements.length);
  655. this.stats.set("processed", "objects", objectElements.length);
  656. objectElements.forEach(element => element.remove());
  657. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  658. replacedAttributeValue.forEach(element => {
  659. let regExp;
  660. if (this.options.removeScripts) {
  661. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  662. } else {
  663. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  664. }
  665. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  666. if (relValue.length) {
  667. element.setAttribute("rel", relValue);
  668. } else {
  669. element.remove();
  670. }
  671. });
  672. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  673. if (this.options.compressHTML) {
  674. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  675. }
  676. if (!this.options.saveFavicon) {
  677. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  678. }
  679. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  680. }
  681. resetCharsetMeta() {
  682. let charset;
  683. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  684. const charsetDeclaration = element.content.split(";")[1];
  685. if (charsetDeclaration && !charset) {
  686. charset = charsetDeclaration.split("=")[1];
  687. if (charset) {
  688. this.charset = charset.trim().toLowerCase();
  689. }
  690. }
  691. element.remove();
  692. });
  693. const metaElement = this.doc.createElement("meta");
  694. metaElement.setAttribute("charset", UTF8_CHARSET);
  695. if (this.doc.head.firstChild) {
  696. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  697. } else {
  698. this.doc.head.appendChild(metaElement);
  699. }
  700. }
  701. setInputValues() {
  702. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  703. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  704. input.setAttribute("value", value || "");
  705. });
  706. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  707. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  708. if (value == "true") {
  709. input.setAttribute("checked", "");
  710. }
  711. });
  712. this.doc.querySelectorAll("textarea").forEach(textarea => {
  713. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  714. textarea.textContent = value || "";
  715. });
  716. this.doc.querySelectorAll("select").forEach(select => {
  717. select.querySelectorAll("option").forEach(option => {
  718. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  719. if (selected) {
  720. option.setAttribute("selected", "");
  721. }
  722. });
  723. });
  724. }
  725. saveFavicon() {
  726. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  727. if (!faviconElement) {
  728. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  729. }
  730. if (!faviconElement) {
  731. faviconElement = this.doc.createElement("link");
  732. faviconElement.setAttribute("type", "image/x-icon");
  733. faviconElement.setAttribute("rel", "shortcut icon");
  734. faviconElement.setAttribute("href", "/favicon.ico");
  735. }
  736. this.doc.head.appendChild(faviconElement);
  737. }
  738. replaceCanvasElements() {
  739. if (this.options.canvases) {
  740. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  741. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  742. if (attributeValue) {
  743. const canvasData = this.options.canvases[Number(attributeValue)];
  744. if (canvasData) {
  745. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  746. this.stats.add("processed", "canvas", 1);
  747. }
  748. }
  749. });
  750. }
  751. }
  752. insertFonts() {
  753. if (this.options.fonts && this.options.fonts.length) {
  754. let stylesheetContent = "";
  755. this.options.fonts.forEach(fontData => {
  756. if (fontData["font-family"] && fontData.src) {
  757. stylesheetContent += "@font-face{";
  758. let stylesContent = "";
  759. Object.keys(fontData).forEach(fontStyle => {
  760. if (stylesContent) {
  761. stylesContent += ";";
  762. }
  763. stylesContent += fontStyle + ":" + fontData[fontStyle];
  764. });
  765. stylesheetContent += stylesContent + "}";
  766. }
  767. });
  768. if (stylesheetContent) {
  769. const styleElement = this.doc.createElement("style");
  770. styleElement.textContent = stylesheetContent;
  771. const existingStyleElement = this.doc.querySelector("style");
  772. if (existingStyleElement) {
  773. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  774. } else {
  775. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  776. }
  777. }
  778. }
  779. }
  780. removeHiddenElements() {
  781. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  782. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  783. this.stats.set("discarded", "hidden elements", removedElements.length);
  784. this.stats.set("processed", "hidden elements", removedElements.length);
  785. if (hiddenElements.length) {
  786. const styleElement = this.doc.createElement("style");
  787. styleElement.textContent = ".sf-hidden{display:none!important;}";
  788. this.doc.head.appendChild(styleElement);
  789. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  790. }
  791. removedElements.forEach(element => element.remove());
  792. }
  793. resolveHrefs() {
  794. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  795. const href = element.getAttribute("href").trim();
  796. if (!testIgnoredPath(href)) {
  797. let resolvedURL;
  798. try {
  799. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  800. } catch (error) {
  801. // ignored
  802. }
  803. if (resolvedURL) {
  804. const url = normalizeURL(this.options.url);
  805. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  806. resolvedURL = resolvedURL.substring(url.length);
  807. }
  808. try {
  809. element.setAttribute("href", resolvedURL);
  810. } catch (error) {
  811. // ignored
  812. }
  813. }
  814. }
  815. });
  816. }
  817. resolveStyleAttributeURLs() {
  818. this.doc.querySelectorAll("[style]").forEach(element => {
  819. let styleContent = element.getAttribute("style");
  820. if (this.options.compressCSS) {
  821. styleContent = util.compressCSS(styleContent);
  822. }
  823. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  824. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  825. this.styles.set(element, declarationList);
  826. });
  827. }
  828. async resolveStylesheetURLs() {
  829. const options = {
  830. maxResourceSize: this.options.maxResourceSize,
  831. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  832. url: this.options.url,
  833. charset: this.charset,
  834. compressCSS: this.options.compressCSS,
  835. updatedResources: this.options.updatedResources,
  836. rootDocument: this.options.rootDocument,
  837. frameId: this.options.windowId
  838. };
  839. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  840. .map(async element => {
  841. let mediaText;
  842. if (element.media) {
  843. mediaText = element.media.toLowerCase();
  844. }
  845. const stylesheetInfo = { mediaText };
  846. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  847. stylesheetInfo.scoped = true;
  848. }
  849. if (element.tagName.toLowerCase() == "link") {
  850. if (element.charset) {
  851. options.charset = element.charset;
  852. }
  853. }
  854. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  855. }));
  856. if (options.rootDocument) {
  857. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  858. await Promise.all(newResources.map(async resource => {
  859. resource.retrieved = true;
  860. const stylesheetInfo = {};
  861. const element = this.doc.createElement("style");
  862. this.doc.body.appendChild(element);
  863. element.textContent = resource.content;
  864. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  865. }));
  866. }
  867. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  868. stylesheets.set(element, stylesheetInfo);
  869. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  870. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  871. if (match && match[1] && match[1] != options.charset) {
  872. options.charset = match[1];
  873. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  874. }
  875. let stylesheet;
  876. try {
  877. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  878. } catch (error) {
  879. // ignored
  880. }
  881. if (stylesheet && stylesheet.children) {
  882. if (options.compressCSS) {
  883. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  884. }
  885. stylesheetInfo.stylesheet = stylesheet;
  886. } else {
  887. stylesheets.delete(element);
  888. }
  889. }
  890. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  891. let content;
  892. if (element.tagName.toLowerCase() == "link") {
  893. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  894. } else {
  895. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  896. }
  897. return content || "";
  898. }
  899. }
  900. async resolveFrameURLs() {
  901. if (!this.options.saveRawPage) {
  902. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  903. await Promise.all(frameElements.map(async frameElement => {
  904. if (frameElement.tagName == "OBJECT") {
  905. frameElement.setAttribute("data", "data:text/html,");
  906. } else {
  907. frameElement.removeAttribute("src");
  908. frameElement.removeAttribute("srcdoc");
  909. }
  910. Array.from(frameElement.childNodes).forEach(node => node.remove());
  911. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  912. if (this.options.frames && frameWindowId) {
  913. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  914. if (frameData) {
  915. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  916. }
  917. }
  918. }));
  919. }
  920. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  921. options.insertSingleFileComment = false;
  922. options.insertCanonicalLink = false;
  923. options.saveFavicon = false;
  924. options.url = frameData.baseURI;
  925. options.windowId = frameWindowId;
  926. if (frameData.content) {
  927. options.content = frameData.content;
  928. options.canvases = frameData.canvases;
  929. options.fonts = frameData.fonts;
  930. options.stylesheets = frameData.stylesheets;
  931. options.images = frameData.images;
  932. options.posters = frameData.posters;
  933. options.usedFonts = frameData.usedFonts;
  934. options.shadowRoots = frameData.shadowRoots;
  935. options.imports = frameData.imports;
  936. frameData.runner = new Runner(options);
  937. frameData.frameElement = frameElement;
  938. await frameData.runner.loadPage();
  939. await frameData.runner.initialize();
  940. frameData.maxResources = batchRequest.getMaxResources();
  941. }
  942. }
  943. }
  944. insertShadowRootContents() {
  945. const doc = this.doc;
  946. const options = this.options;
  947. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  948. processElement(this.doc);
  949. if (this.options.removeScripts) {
  950. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  951. }
  952. const scriptElement = doc.createElement("script");
  953. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  954. 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)}})}})()`;
  955. doc.body.appendChild(scriptElement);
  956. }
  957. function processElement(element) {
  958. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  959. shadowRootElements.forEach(element => {
  960. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  961. if (attributeValue) {
  962. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  963. if (shadowRootData) {
  964. const templateElement = doc.createElement("template");
  965. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  966. if (shadowRootData.delegatesFocus) {
  967. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  968. }
  969. if (shadowRootData.adoptedStyleSheets) {
  970. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  971. const styleElement = doc.createElement("style");
  972. styleElement.textContent = stylesheetContent;
  973. templateElement.appendChild(styleElement);
  974. });
  975. }
  976. const shadowDoc = util.parseDocContent(shadowRootData.content);
  977. if (shadowDoc.head) {
  978. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  979. if (metaCharset) {
  980. metaCharset.remove();
  981. }
  982. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  983. }
  984. if (shadowDoc.body) {
  985. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  986. }
  987. processElement(templateElement);
  988. if (element.firstChild) {
  989. element.insertBefore(templateElement, element.firstChild);
  990. } else {
  991. element.appendChild(templateElement);
  992. }
  993. }
  994. }
  995. });
  996. }
  997. }
  998. async resolveHtmlImportURLs() {
  999. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1000. await Promise.all(linkElements.map(async linkElement => {
  1001. const resourceURL = linkElement.href;
  1002. linkElement.removeAttribute("href");
  1003. const options = Object.create(this.options);
  1004. options.insertSingleFileComment = false;
  1005. options.insertCanonicalLink = false;
  1006. options.saveFavicon = false;
  1007. options.removeUnusedStyles = false;
  1008. options.removeAlternativeMedias = false;
  1009. options.removeUnusedFonts = false;
  1010. options.removeHiddenElements = false;
  1011. options.url = resourceURL;
  1012. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1013. if (attributeValue) {
  1014. const importData = options.imports[Number(attributeValue)];
  1015. if (importData) {
  1016. options.content = importData.content;
  1017. importData.runner = new Runner(options);
  1018. await importData.runner.loadPage();
  1019. await importData.runner.initialize();
  1020. if (!options.removeImports) {
  1021. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1022. }
  1023. importData.runner.getStyleSheets().forEach(stylesheet => {
  1024. const importedStyleElement = this.doc.createElement("style");
  1025. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1026. this.stylesheets.set(importedStyleElement, stylesheet);
  1027. });
  1028. }
  1029. }
  1030. if (options.removeImports) {
  1031. linkElement.remove();
  1032. this.stats.add("discarded", "HTML imports", 1);
  1033. }
  1034. }));
  1035. }
  1036. removeUnusedStyles() {
  1037. if (!this.mediaAllInfo) {
  1038. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1039. }
  1040. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1041. this.stats.set("processed", "CSS rules", stats.processed);
  1042. this.stats.set("discarded", "CSS rules", stats.discarded);
  1043. }
  1044. removeUnusedFonts() {
  1045. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1046. }
  1047. removeAlternativeMedias() {
  1048. const stats = util.minifyMedias(this.stylesheets);
  1049. this.stats.set("processed", "medias", stats.processed);
  1050. this.stats.set("discarded", "medias", stats.discarded);
  1051. }
  1052. async processStylesheets() {
  1053. this.options.fontURLs = new Map();
  1054. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1055. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1056. ));
  1057. }
  1058. async processStyleAttributes() {
  1059. return Promise.all([...this.styles].map(([, declarationList]) =>
  1060. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1061. ));
  1062. }
  1063. async processPageResources() {
  1064. const processAttributeArgs = [
  1065. ["link[href][rel*=\"icon\"]", "href", false, true],
  1066. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1067. ["img[src], input[src][type=image]", "src", true],
  1068. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1069. ["video[poster]", "poster"],
  1070. ["*[background]", "background"],
  1071. ["image", "xlink:href"],
  1072. ["image", "href"]
  1073. ];
  1074. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1075. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1076. );
  1077. resourcePromises = resourcePromises.concat([
  1078. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1079. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1080. ]);
  1081. if (!this.options.removeAudioSrc) {
  1082. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1083. }
  1084. if (!this.options.removeVideoSrc) {
  1085. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1086. }
  1087. await Promise.all(resourcePromises);
  1088. if (this.options.saveFavicon) {
  1089. ProcessorHelper.processShortcutIcons(this.doc);
  1090. }
  1091. }
  1092. async processScripts() {
  1093. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1094. let resourceURL;
  1095. let scriptSrc;
  1096. if (element.tagName == "SCRIPT") {
  1097. scriptSrc = element.getAttribute("src");
  1098. element.removeAttribute("src");
  1099. } else {
  1100. scriptSrc = element.getAttribute("href");
  1101. element.removeAttribute("href");
  1102. }
  1103. element.removeAttribute("integrity");
  1104. element.textContent = "";
  1105. try {
  1106. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1107. } catch (error) {
  1108. // ignored
  1109. }
  1110. if (testValidURL(resourceURL)) {
  1111. this.stats.add("processed", "scripts", 1);
  1112. const content = await util.getContent(resourceURL, {
  1113. asBinary: true,
  1114. charset: this.charset != UTF8_CHARSET && this.charset,
  1115. maxResourceSize: this.options.maxResourceSize,
  1116. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1117. frameId: this.options.windowId
  1118. });
  1119. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1120. if (element.tagName == "SCRIPT") {
  1121. element.setAttribute("src", content.data);
  1122. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1123. element.setAttribute("async", "");
  1124. }
  1125. } else {
  1126. const scriptElement = this.doc.createElement("script");
  1127. scriptElement.setAttribute("src", content.data);
  1128. scriptElement.setAttribute("async", "");
  1129. element.parentElement.replaceChild(scriptElement, element);
  1130. }
  1131. }
  1132. }));
  1133. }
  1134. removeAlternativeImages() {
  1135. util.removeAlternativeImages(this.doc);
  1136. }
  1137. async removeAlternativeFonts() {
  1138. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs);
  1139. }
  1140. async processFrames() {
  1141. if (this.options.frames) {
  1142. const frameElements = Array.from(this.doc.querySelectorAll("iframe:not([" + util.HIDDEN_FRAME_ATTRIBUTE_NAME + "]), frame, object[type=\"text/html\"][data]"));
  1143. await Promise.all(frameElements.map(async frameElement => {
  1144. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1145. if (frameWindowId) {
  1146. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1147. if (frameData) {
  1148. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1149. if (frameData.runner) {
  1150. this.stats.add("processed", "frames", 1);
  1151. await frameData.runner.run();
  1152. const pageData = await frameData.runner.getPageData();
  1153. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1154. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1155. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1156. sandbox += " allow-scripts allow-same-origin";
  1157. }
  1158. frameElement.setAttribute("sandbox", sandbox);
  1159. if (frameElement.tagName == "OBJECT") {
  1160. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1161. } else {
  1162. if (frameElement.tagName == "FRAME") {
  1163. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1164. } else {
  1165. frameElement.setAttribute("srcdoc", pageData.content);
  1166. frameElement.removeAttribute("src");
  1167. }
  1168. }
  1169. this.stats.addAll(pageData);
  1170. } else {
  1171. this.stats.add("discarded", "frames", 1);
  1172. }
  1173. }
  1174. }
  1175. }));
  1176. }
  1177. }
  1178. async processHtmlImports() {
  1179. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1180. await Promise.all(linkElements.map(async linkElement => {
  1181. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1182. if (attributeValue) {
  1183. const importData = this.options.imports[Number(attributeValue)];
  1184. if (importData.runner) {
  1185. this.stats.add("processed", "HTML imports", 1);
  1186. await importData.runner.run();
  1187. const pageData = await importData.runner.getPageData();
  1188. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1189. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1190. this.stats.addAll(pageData);
  1191. } else {
  1192. this.stats.add("discarded", "HTML imports", 1);
  1193. }
  1194. }
  1195. }));
  1196. }
  1197. replaceStylesheets() {
  1198. this.doc.querySelectorAll("style").forEach(styleElement => {
  1199. const stylesheetInfo = this.stylesheets.get(styleElement);
  1200. if (stylesheetInfo) {
  1201. this.stylesheets.delete(styleElement);
  1202. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1203. styleElement.textContent = stylesheetContent;
  1204. if (stylesheetInfo.mediaText) {
  1205. styleElement.media = stylesheetInfo.mediaText;
  1206. }
  1207. } else {
  1208. styleElement.remove();
  1209. }
  1210. });
  1211. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1212. const stylesheetInfo = this.stylesheets.get(linkElement);
  1213. if (stylesheetInfo) {
  1214. this.stylesheets.delete(linkElement);
  1215. const styleElement = this.doc.createElement("style");
  1216. if (stylesheetInfo.mediaText) {
  1217. styleElement.media = stylesheetInfo.mediaText;
  1218. }
  1219. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1220. styleElement.textContent = stylesheetContent;
  1221. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1222. } else {
  1223. linkElement.remove();
  1224. }
  1225. });
  1226. }
  1227. replaceStyleAttributes() {
  1228. this.doc.querySelectorAll("[style]").forEach(element => {
  1229. const declarations = this.styles.get(element);
  1230. if (declarations) {
  1231. this.styles.delete(element);
  1232. let styleContent = cssTree.generate(declarations);
  1233. element.setAttribute("style", styleContent);
  1234. } else {
  1235. element.setAttribute("style", "");
  1236. }
  1237. });
  1238. }
  1239. insertVariables() {
  1240. if (this.cssVariables.size) {
  1241. const styleElement = this.doc.createElement("style");
  1242. const firstStyleElement = this.doc.head.querySelector("style");
  1243. if (firstStyleElement) {
  1244. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1245. } else {
  1246. this.doc.head.appendChild(styleElement);
  1247. }
  1248. let stylesheetContent = "";
  1249. this.cssVariables.forEach((content, indexResource) => {
  1250. this.cssVariables.delete(indexResource);
  1251. if (stylesheetContent) {
  1252. stylesheetContent += ";";
  1253. }
  1254. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1255. });
  1256. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1257. }
  1258. }
  1259. compressHTML() {
  1260. let size;
  1261. if (this.options.displayStats) {
  1262. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1263. }
  1264. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1265. if (this.options.displayStats) {
  1266. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1267. }
  1268. }
  1269. cleanupPage() {
  1270. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1271. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1272. if (metaCharset) {
  1273. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1274. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1275. this.doc.head.querySelector("meta[charset]").remove();
  1276. }
  1277. }
  1278. }
  1279. resetZoomLevel() {
  1280. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1281. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1282. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1283. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1284. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1285. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1286. this.doc.documentElement.style.removeProperty("-sf-transform");
  1287. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1288. }
  1289. async insertMAFFMetaData() {
  1290. const maffMetaData = await this.maffMetaDataPromise;
  1291. if (maffMetaData && maffMetaData.content) {
  1292. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1293. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1294. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1295. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1296. if (originalURLElement) {
  1297. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1298. }
  1299. if (archiveTimeElement) {
  1300. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1301. if (value) {
  1302. const date = new Date(value);
  1303. if (!isNaN(date.getTime())) {
  1304. this.options.saveDate = new Date(value);
  1305. }
  1306. }
  1307. }
  1308. }
  1309. }
  1310. async setDocInfo() {
  1311. const titleElement = this.doc.querySelector("title");
  1312. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1313. const authorElement = this.doc.querySelector("meta[name=author]");
  1314. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1315. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1316. const headingElement = this.doc.querySelector("h1");
  1317. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1318. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1319. this.options.info = {
  1320. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1321. lang: this.doc.documentElement.lang,
  1322. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1323. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1324. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1325. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1326. };
  1327. }
  1328. }
  1329. // ---------------
  1330. // ProcessorHelper
  1331. // ---------------
  1332. const DATA_URI_PREFIX = "data:";
  1333. const ABOUT_BLANK_URI = "about:blank";
  1334. const EMPTY_DATA_URI = "data:null;base64,";
  1335. const REGEXP_URL_HASH = /(#.+?)$/;
  1336. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1337. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1338. class ProcessorHelper {
  1339. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1340. const date = options.saveDate;
  1341. const url = util.parseURL(options.saveUrl);
  1342. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1343. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1344. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1345. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1346. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1347. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1348. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1349. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1350. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1351. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1352. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1353. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1354. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1357. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1358. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1359. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1360. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1361. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1362. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1363. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1364. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1365. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1366. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1367. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1368. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1370. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1371. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1376. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1377. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1378. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1379. for (const [name, value] of params) {
  1380. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1381. }
  1382. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1383. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1387. if (content) {
  1388. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. }
  1392. return template.trim();
  1393. function decode(value) {
  1394. try {
  1395. return decodeURI(value);
  1396. } catch (error) {
  1397. return value;
  1398. }
  1399. }
  1400. }
  1401. static setBackgroundImage(element, url, style) {
  1402. element.style.setProperty("background-blend-mode", "normal", "important");
  1403. element.style.setProperty("background-clip", "content-box", "important");
  1404. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1405. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1406. element.style.setProperty("background-image", url, "important");
  1407. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1408. element.style.setProperty("background-origin", "content-box", "important");
  1409. element.style.setProperty("background-repeat", "no-repeat", "important");
  1410. }
  1411. static processShortcutIcons(doc) {
  1412. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1413. if (!shortcutIcon) {
  1414. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1415. if (shortcutIcon) {
  1416. shortcutIcon.rel = "icon";
  1417. }
  1418. }
  1419. if (shortcutIcon) {
  1420. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1421. if (linkElement != shortcutIcon) {
  1422. linkElement.remove();
  1423. }
  1424. });
  1425. }
  1426. }
  1427. static removeSingleLineCssComments(stylesheet) {
  1428. const removedRules = [];
  1429. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1430. const ruleData = cssRule.data;
  1431. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1432. removedRules.push(cssRule);
  1433. }
  1434. }
  1435. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1436. }
  1437. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1438. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1439. const imports = getImportFunctions(stylesheetContent);
  1440. await Promise.all(imports.map(async cssImport => {
  1441. const match = matchImport(cssImport);
  1442. if (match) {
  1443. const regExpCssImport = getRegExp(cssImport);
  1444. let resourceURL = normalizeURL(match.resourceURL);
  1445. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1446. try {
  1447. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1448. } catch (error) {
  1449. // ignored
  1450. }
  1451. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1452. const content = await util.getContent(resourceURL, {
  1453. maxResourceSize: options.maxResourceSize,
  1454. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1455. validateTextContentType: true,
  1456. frameId: options.frameId
  1457. });
  1458. resourceURL = content.resourceURL;
  1459. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1460. let importedStylesheetContent = removeCssComments(content.data);
  1461. if (options.compressCSS) {
  1462. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1463. }
  1464. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1465. if (stylesheetContent.includes(cssImport)) {
  1466. const ancestorStyleSheets = new Set(importedStyleSheets);
  1467. ancestorStyleSheets.add(resourceURL);
  1468. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1469. workStylesheet.textContent = importedStylesheetContent;
  1470. if (workStylesheet.sheet.cssRules.length) {
  1471. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1472. } else {
  1473. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1474. }
  1475. }
  1476. } else {
  1477. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1478. }
  1479. } else {
  1480. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1481. }
  1482. }
  1483. }));
  1484. return stylesheetContent;
  1485. }
  1486. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1487. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1488. urlFunctions.map(urlFunction => {
  1489. const originalResourceURL = matchURL(urlFunction);
  1490. const resourceURL = normalizeURL(originalResourceURL);
  1491. if (!testIgnoredPath(resourceURL)) {
  1492. if (!resourceURL || testValidPath(resourceURL)) {
  1493. let resolvedURL;
  1494. if (!originalResourceURL.startsWith("#")) {
  1495. try {
  1496. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1497. } catch (error) {
  1498. // ignored
  1499. }
  1500. }
  1501. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1502. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1503. }
  1504. } else {
  1505. let newUrlFunction;
  1506. if (originalResourceURL) {
  1507. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1508. } else {
  1509. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1510. }
  1511. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1512. }
  1513. }
  1514. });
  1515. return stylesheetContent;
  1516. }
  1517. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1518. resourceURL = normalizeURL(resourceURL);
  1519. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1520. const content = await util.getContent(resourceURL, {
  1521. maxResourceSize: options.maxResourceSize,
  1522. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1523. charset: options.charset,
  1524. frameId: options.frameId,
  1525. validateTextContentType: true
  1526. });
  1527. resourceURL = content.resourceURL;
  1528. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1529. let stylesheetContent = removeCssComments(content.data);
  1530. if (options.compressCSS) {
  1531. stylesheetContent = util.compressCSS(stylesheetContent);
  1532. }
  1533. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1534. return stylesheetContent;
  1535. }
  1536. }
  1537. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1538. const promises = [];
  1539. const removedRules = [];
  1540. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1541. const ruleData = cssRule.data;
  1542. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1543. removedRules.push(cssRule);
  1544. } else if (ruleData.block && ruleData.block.children) {
  1545. if (ruleData.type == "Rule") {
  1546. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1547. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1548. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1549. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1550. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1551. }
  1552. }
  1553. }
  1554. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1555. await Promise.all(promises);
  1556. async function processFontFaceRule(ruleData, fontURLs) {
  1557. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1558. if (declaration.type == "Declaration" && declaration.value.children) {
  1559. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1560. await Promise.all(urlFunctions.map(async urlFunction => {
  1561. const originalResourceURL = matchURL(urlFunction);
  1562. const resourceURL = normalizeURL(originalResourceURL);
  1563. if (!testIgnoredPath(resourceURL)) {
  1564. if (testValidURL(resourceURL)) {
  1565. let { content } = await batchRequest.addURL(resourceURL, true);
  1566. let resourceURLs = fontURLs.get(declaration);
  1567. if (!resourceURLs) {
  1568. resourceURLs = [];
  1569. fontURLs.set(declaration, resourceURLs);
  1570. }
  1571. resourceURLs.push(resourceURL);
  1572. replaceURLs(declaration, originalResourceURL, content);
  1573. }
  1574. }
  1575. }));
  1576. }
  1577. }));
  1578. function replaceURLs(declaration, oldURL, newURL) {
  1579. declaration.value.children.forEach(token => {
  1580. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1581. token.value.value = newURL;
  1582. }
  1583. });
  1584. }
  1585. }
  1586. }
  1587. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1588. await Promise.all(declarations.map(async declaration => {
  1589. if (declaration.type == "Declaration" && declaration.value.children) {
  1590. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1591. await Promise.all(urlFunctions.map(async urlFunction => {
  1592. const originalResourceURL = matchURL(urlFunction);
  1593. const resourceURL = normalizeURL(originalResourceURL);
  1594. if (!testIgnoredPath(resourceURL)) {
  1595. if (testValidURL(resourceURL)) {
  1596. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1597. let variableDefined;
  1598. const tokens = [];
  1599. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1600. if (!originalResourceURL.startsWith("#")) {
  1601. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1602. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1603. tokens.push({ parent, token, value });
  1604. variableDefined = true;
  1605. } else {
  1606. token.data.value.value = content;
  1607. }
  1608. }
  1609. });
  1610. if (variableDefined) {
  1611. cssVariables.set(indexResource, content);
  1612. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1613. }
  1614. }
  1615. }
  1616. }));
  1617. }
  1618. }));
  1619. function findURLToken(url, children, callback, depth = 0) {
  1620. for (let token = children.head; token; token = token.next) {
  1621. if (token.data.children) {
  1622. findURLToken(url, token.data.children, callback, depth + 1);
  1623. }
  1624. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1625. callback(token, children, depth == 0);
  1626. }
  1627. }
  1628. }
  1629. }
  1630. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1631. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1632. let resourceURL = resourceElement.getAttribute(attributeName);
  1633. if (resourceURL != null) {
  1634. resourceURL = normalizeURL(resourceURL);
  1635. if (!testIgnoredPath(resourceURL)) {
  1636. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1637. if (testValidPath(resourceURL)) {
  1638. try {
  1639. resourceURL = util.resolveURL(resourceURL, baseURI);
  1640. } catch (error) {
  1641. // ignored
  1642. }
  1643. if (testValidURL(resourceURL)) {
  1644. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1645. if (resourceElement.dataset.singleFileOriginURL) {
  1646. let originURL = resourceElement.dataset.singleFileOriginURL;
  1647. delete resourceElement.dataset.singleFileOriginURL;
  1648. if (content == EMPTY_DATA_URI) {
  1649. try {
  1650. originURL = util.resolveURL(originURL, baseURI);
  1651. } catch (error) {
  1652. // ignored
  1653. }
  1654. try {
  1655. resourceURL = originURL;
  1656. content = (await util.getContent(resourceURL, {
  1657. asBinary: true,
  1658. maxResourceSize: options.maxResourceSize,
  1659. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1660. frameId: options.windowId
  1661. })).data;
  1662. } catch (error) {
  1663. // ignored
  1664. }
  1665. }
  1666. }
  1667. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1668. resourceElement.remove();
  1669. } else {
  1670. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1671. if (!forbiddenPrefixFound) {
  1672. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1673. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1674. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1675. cssVariables.set(indexResource, content);
  1676. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1677. styles.set(resourceElement, declarationList);
  1678. } else {
  1679. resourceElement.setAttribute(attributeName, content);
  1680. }
  1681. } else {
  1682. resourceElement.setAttribute(attributeName, content);
  1683. }
  1684. }
  1685. }
  1686. }
  1687. }
  1688. }
  1689. }
  1690. }));
  1691. }
  1692. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1693. let attributeName = "xlink:href";
  1694. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1695. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1696. if (originalResourceURL == null) {
  1697. attributeName = "href";
  1698. originalResourceURL = resourceElement.getAttribute(attributeName);
  1699. }
  1700. let resourceURL = normalizeURL(originalResourceURL);
  1701. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1702. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1703. try {
  1704. resourceURL = util.resolveURL(resourceURL, baseURI);
  1705. } catch (error) {
  1706. // ignored
  1707. }
  1708. if (testValidURL(resourceURL)) {
  1709. const { content } = await batchRequest.addURL(resourceURL);
  1710. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1711. if (hashMatch && hashMatch[0]) {
  1712. let symbolElement;
  1713. try {
  1714. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1715. } catch (error) {
  1716. // ignored
  1717. }
  1718. if (symbolElement) {
  1719. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1720. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1721. }
  1722. } else {
  1723. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1724. }
  1725. }
  1726. } else if (resourceURL == options.url) {
  1727. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1728. }
  1729. }));
  1730. }
  1731. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1732. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1733. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1734. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1735. let resourceURL = normalizeURL(srcsetValue.url);
  1736. if (!testIgnoredPath(resourceURL)) {
  1737. if (testValidPath(resourceURL)) {
  1738. try {
  1739. resourceURL = util.resolveURL(resourceURL, baseURI);
  1740. } catch (error) {
  1741. // ignored
  1742. }
  1743. if (testValidURL(resourceURL)) {
  1744. const { content } = await batchRequest.addURL(resourceURL, true);
  1745. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1746. if (forbiddenPrefixFound) {
  1747. return "";
  1748. }
  1749. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1750. } else {
  1751. return "";
  1752. }
  1753. } else {
  1754. return "";
  1755. }
  1756. } else {
  1757. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1758. }
  1759. }));
  1760. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1761. }));
  1762. }
  1763. static replaceImageSource(imgElement, variableName, options) {
  1764. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1765. if (attributeValue) {
  1766. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1767. if (imageData && imageData.replaceable) {
  1768. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1769. const backgroundStyle = {};
  1770. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1771. if (backgroundSize) {
  1772. backgroundStyle["background-size"] = imageData.objectFit;
  1773. }
  1774. if (imageData.objectPosition) {
  1775. backgroundStyle["background-position"] = imageData.objectPosition;
  1776. }
  1777. if (imageData.backgroundColor) {
  1778. backgroundStyle["background-color"] = imageData.backgroundColor;
  1779. }
  1780. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1781. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1782. return true;
  1783. }
  1784. }
  1785. }
  1786. }
  1787. // ----
  1788. // Util
  1789. // ----
  1790. const BLOB_URI_PREFIX = "blob:";
  1791. const HTTP_URI_PREFIX = /^https?:\/\//;
  1792. const FILE_URI_PREFIX = /^file:\/\//;
  1793. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1794. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1795. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1796. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1797. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1798. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1799. 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;
  1800. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1801. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1802. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1803. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1804. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1805. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1806. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1807. function getUpdatedResourceContent(resourceURL, content, options) {
  1808. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1809. options.updatedResources[resourceURL].retrieved = true;
  1810. return options.updatedResources[resourceURL].content;
  1811. } else {
  1812. return content.data || "";
  1813. }
  1814. }
  1815. function normalizeURL(url) {
  1816. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1817. return url;
  1818. } else {
  1819. return url.split("#")[0];
  1820. }
  1821. }
  1822. function getCSSValue(value) {
  1823. let result = "";
  1824. try {
  1825. result = cssTree.generate(value);
  1826. } catch (error) {
  1827. // ignored
  1828. }
  1829. return result;
  1830. }
  1831. function getOnEventAttributeNames(doc) {
  1832. const element = doc.createElement("div");
  1833. const attributeNames = [];
  1834. for (const propertyName in element) {
  1835. if (propertyName.startsWith("on")) {
  1836. attributeNames.push(propertyName);
  1837. }
  1838. }
  1839. attributeNames.push("onunload");
  1840. return attributeNames;
  1841. }
  1842. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1843. let maxLength;
  1844. if (template) {
  1845. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1846. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1847. if (template.match(replaceRegExp)) {
  1848. const matchedLength = template.match(replaceRegExp)[0];
  1849. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1850. if (isNaN(maxLength) || maxLength <= 0) {
  1851. maxLength = null;
  1852. }
  1853. } else {
  1854. replaceRegExp = new RegExp(regExpVariable, "g");
  1855. }
  1856. if (template.match(replaceRegExp)) {
  1857. let value = await valueGetter();
  1858. if (!dontReplaceSlash) {
  1859. value = value.replace(/\/+/g, replacementCharacter);
  1860. }
  1861. if (maxLength) {
  1862. value = await util.truncateText(value, maxLength);
  1863. }
  1864. return template.replace(replaceRegExp, value);
  1865. }
  1866. }
  1867. return template;
  1868. }
  1869. function getLastSegment(url, replacementCharacter) {
  1870. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1871. if (!lastSegment) {
  1872. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1873. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1874. }
  1875. if (!lastSegment) {
  1876. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1877. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1878. }
  1879. if (!lastSegment) {
  1880. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1881. }
  1882. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1883. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1884. lastSegment = lastSegmentMatch[1];
  1885. }
  1886. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1887. return lastSegment;
  1888. }
  1889. function getRegExp(string) {
  1890. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1891. }
  1892. function getUrlFunctions(stylesheetContent, unique) {
  1893. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1894. if (unique) {
  1895. return [...new Set(result)];
  1896. } else {
  1897. return result;
  1898. }
  1899. }
  1900. function getImportFunctions(stylesheetContent) {
  1901. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1902. }
  1903. function findShortcutIcon(shortcutIcons) {
  1904. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1905. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1906. return shortcutIcons[0];
  1907. }
  1908. function matchURL(stylesheetContent) {
  1909. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1910. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1911. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1912. return match && match[1];
  1913. }
  1914. function testIgnoredPath(resourceURL) {
  1915. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1916. }
  1917. function testValidPath(resourceURL) {
  1918. return resourceURL && !resourceURL.match(EMPTY_URL);
  1919. }
  1920. function testValidURL(resourceURL) {
  1921. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1922. }
  1923. function matchImport(stylesheetContent) {
  1924. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1925. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1926. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1927. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1928. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1929. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1930. if (match) {
  1931. const [, resourceURL, media] = match;
  1932. return { resourceURL, media };
  1933. }
  1934. }
  1935. function removeCssComments(stylesheetContent) {
  1936. try {
  1937. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  1938. } catch (error) {
  1939. let start, end;
  1940. do {
  1941. start = stylesheetContent.indexOf("/*");
  1942. end = stylesheetContent.indexOf("*/", start + 2);
  1943. if (start != -1 && end != -1) {
  1944. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1945. }
  1946. } while (start != -1 && end != -1);
  1947. return stylesheetContent;
  1948. }
  1949. }
  1950. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  1951. if (mediaQuery) {
  1952. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1953. } else {
  1954. return stylesheetContent;
  1955. }
  1956. }
  1957. function log(...args) {
  1958. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1959. }
  1960. // -----
  1961. // Stats
  1962. // -----
  1963. const STATS_DEFAULT_VALUES = {
  1964. discarded: {
  1965. "HTML bytes": 0,
  1966. "hidden elements": 0,
  1967. "HTML imports": 0,
  1968. scripts: 0,
  1969. objects: 0,
  1970. "audio sources": 0,
  1971. "video sources": 0,
  1972. frames: 0,
  1973. "CSS rules": 0,
  1974. canvas: 0,
  1975. stylesheets: 0,
  1976. resources: 0,
  1977. medias: 0
  1978. },
  1979. processed: {
  1980. "HTML bytes": 0,
  1981. "hidden elements": 0,
  1982. "HTML imports": 0,
  1983. scripts: 0,
  1984. objects: 0,
  1985. "audio sources": 0,
  1986. "video sources": 0,
  1987. frames: 0,
  1988. "CSS rules": 0,
  1989. canvas: 0,
  1990. stylesheets: 0,
  1991. resources: 0,
  1992. medias: 0
  1993. }
  1994. };
  1995. class Stats {
  1996. constructor(options) {
  1997. this.options = options;
  1998. if (options.displayStats) {
  1999. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2000. }
  2001. }
  2002. set(type, subType, value) {
  2003. if (this.options.displayStats) {
  2004. this.data[type][subType] = value;
  2005. }
  2006. }
  2007. add(type, subType, value) {
  2008. if (this.options.displayStats) {
  2009. this.data[type][subType] += value;
  2010. }
  2011. }
  2012. addAll(pageData) {
  2013. if (this.options.displayStats) {
  2014. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2015. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2016. }
  2017. }
  2018. }
  2019. return { getClass };
  2020. })();