single-file-core.js 86 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115
  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. { action: "replaceStyleContents" },
  89. { action: "resetCharsetMeta" },
  90. { option: "saveFavicon", action: "saveFavicon" },
  91. { action: "replaceCanvasElements" },
  92. { action: "insertFonts" },
  93. { action: "insertShadowRootContents" },
  94. { action: "setInputValues" },
  95. { option: "removeScripts", action: "removeScripts" },
  96. { option: "selected", action: "removeUnselectedElements" },
  97. { option: "removeVideoSrc", action: "insertVideoPosters" },
  98. { option: "removeFrames", action: "removeFrames" },
  99. { option: "removeVideoSrc", action: "removeVideoSources" },
  100. { option: "removeAudioSrc", action: "removeAudioSources" },
  101. { action: "removeDiscardedResources" },
  102. { option: "removeHiddenElements", action: "removeHiddenElements" },
  103. { action: "resolveHrefs" },
  104. { action: "resolveStyleAttributeURLs" }
  105. ],
  106. parallel: [
  107. { action: "resolveStylesheetURLs" },
  108. { option: "!removeFrames", action: "resolveFrameURLs" },
  109. { action: "resolveHtmlImportURLs" }
  110. ]
  111. }, {
  112. sequential: [
  113. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  114. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  115. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  116. ],
  117. parallel: [
  118. { action: "processStylesheets" },
  119. { action: "processStyleAttributes" },
  120. { action: "processPageResources" },
  121. { option: "!removeScripts", action: "processScripts" }
  122. ]
  123. }, {
  124. sequential: [
  125. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  126. ],
  127. parallel: [
  128. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  129. { option: "!removeFrames", action: "processFrames" },
  130. { option: "!removeImports", action: "processHtmlImports" },
  131. ]
  132. }, {
  133. sequential: [
  134. { action: "replaceStylesheets" },
  135. { action: "replaceStyleAttributes" },
  136. { action: "insertVariables" },
  137. { option: "compressHTML", action: "compressHTML" },
  138. { action: "cleanupPage" }
  139. ],
  140. parallel: [
  141. { option: "enableMaff", action: "insertMAFFMetaData" },
  142. { action: "setDocInfo" }
  143. ]
  144. }];
  145. class Runner {
  146. constructor(options, root) {
  147. const rootDocDefined = root && options.doc;
  148. this.root = root;
  149. this.options = options;
  150. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  151. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  152. this.options.rootDocument = root;
  153. this.options.updatedResources = this.options.updatedResources || {};
  154. this.batchRequest = new BatchRequest();
  155. this.processor = new Processor(options, this.batchRequest);
  156. if (rootDocDefined) {
  157. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  158. this.options.canvases = docData.canvases;
  159. this.options.fonts = docData.fonts;
  160. this.options.stylesheets = docData.stylesheets;
  161. this.options.images = docData.images;
  162. this.options.posters = docData.posters;
  163. this.options.usedFonts = docData.usedFonts;
  164. this.options.shadowRoots = docData.shadowRoots;
  165. this.options.imports = docData.imports;
  166. this.options.referrer = docData.referrer;
  167. this.markedElements = docData.markedElements;
  168. }
  169. if (this.options.saveRawPage) {
  170. this.options.removeFrames = true;
  171. }
  172. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  173. this.onprogress = options.onprogress || (() => { });
  174. }
  175. async loadPage() {
  176. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  177. await this.processor.loadPage(this.options.content);
  178. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  179. }
  180. async initialize() {
  181. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  182. await this.executeStage(RESOLVE_URLS_STAGE);
  183. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  184. if (this.root && this.options.doc) {
  185. util.postProcessDoc(this.options.doc, this.markedElements);
  186. }
  187. }
  188. cancel() {
  189. this.cancelled = true;
  190. this.batchRequest.cancel();
  191. if (this.root) {
  192. if (this.options.frames) {
  193. this.options.frames.forEach(cancelRunner);
  194. }
  195. if (this.options.imports) {
  196. this.options.imports.forEach(cancelRunner);
  197. }
  198. }
  199. function cancelRunner(resourceData) {
  200. if (resourceData.runner) {
  201. resourceData.runner.cancel();
  202. }
  203. }
  204. }
  205. async run() {
  206. if (this.root) {
  207. this.processor.initialize(this.batchRequest);
  208. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  209. }
  210. await this.batchRequest.run(detail => {
  211. detail.pageURL = this.options.url;
  212. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  213. }, this.options);
  214. await this.pendingPromises;
  215. this.options.doc = null;
  216. this.options.win = null;
  217. await this.executeStage(REPLACE_DOCS_STAGE);
  218. await this.executeStage(POST_PROCESS_STAGE);
  219. this.processor.finalize();
  220. }
  221. getDocument() {
  222. return this.processor.doc;
  223. }
  224. getStyleSheets() {
  225. return this.processor.stylesheets;
  226. }
  227. getPageData() {
  228. if (this.root) {
  229. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  230. }
  231. return this.processor.getPageData();
  232. }
  233. async executeStage(step) {
  234. if (DEBUG) {
  235. log("**** STARTED STAGE", step, "****");
  236. }
  237. const frame = !this.root;
  238. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  239. STAGES[step].sequential.forEach(task => {
  240. let startTime;
  241. if (DEBUG) {
  242. startTime = Date.now();
  243. log(" -- STARTED task =", task.action);
  244. }
  245. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  246. if (!this.cancelled) {
  247. this.executeTask(task);
  248. }
  249. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  250. if (DEBUG) {
  251. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  252. }
  253. });
  254. let parallelTasksPromise;
  255. if (STAGES[step].parallel) {
  256. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  257. let startTime;
  258. if (DEBUG) {
  259. startTime = Date.now();
  260. log(" // STARTED task =", task.action);
  261. }
  262. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  263. if (!this.cancelled) {
  264. await this.executeTask(task);
  265. }
  266. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  267. if (DEBUG) {
  268. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  269. }
  270. }));
  271. } else {
  272. parallelTasksPromise = Promise.resolve();
  273. }
  274. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  275. if (DEBUG) {
  276. log("**** ENDED STAGE", step, "****");
  277. }
  278. return parallelTasksPromise;
  279. }
  280. executeTask(task) {
  281. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  282. return this.processor[task.action]();
  283. }
  284. }
  285. }
  286. // ------------
  287. // BatchRequest
  288. // ------------
  289. class BatchRequest {
  290. constructor() {
  291. this.requests = new Map();
  292. this.duplicates = new Map();
  293. }
  294. addURL(resourceURL, asBinary, groupDuplicates) {
  295. return new Promise((resolve, reject) => {
  296. const requestKey = JSON.stringify([resourceURL, asBinary]);
  297. let resourceRequests = this.requests.get(requestKey);
  298. if (!resourceRequests) {
  299. resourceRequests = [];
  300. this.requests.set(requestKey, resourceRequests);
  301. }
  302. const callbacks = { resolve, reject };
  303. resourceRequests.push(callbacks);
  304. if (groupDuplicates) {
  305. let duplicateRequests = this.duplicates.get(requestKey);
  306. if (!duplicateRequests) {
  307. duplicateRequests = [];
  308. this.duplicates.set(requestKey, duplicateRequests);
  309. }
  310. duplicateRequests.push(callbacks);
  311. }
  312. });
  313. }
  314. getMaxResources() {
  315. return this.requests.size;
  316. }
  317. run(onloadListener, options) {
  318. const resourceURLs = [...this.requests.keys()];
  319. let indexResource = 0;
  320. return Promise.all(resourceURLs.map(async requestKey => {
  321. const [resourceURL, asBinary] = JSON.parse(requestKey);
  322. const resourceRequests = this.requests.get(requestKey);
  323. try {
  324. const content = await util.getContent(resourceURL, {
  325. asBinary,
  326. maxResourceSize: options.maxResourceSize,
  327. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  328. frameId: options.windowId
  329. });
  330. indexResource = indexResource + 1;
  331. onloadListener({ url: resourceURL });
  332. if (!this.cancelled) {
  333. resourceRequests.forEach(callbacks => {
  334. const duplicateCallbacks = this.duplicates.get(requestKey);
  335. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  336. callbacks.resolve({ content: content.data, indexResource, duplicate });
  337. });
  338. }
  339. } catch (error) {
  340. indexResource = indexResource + 1;
  341. onloadListener({ url: resourceURL });
  342. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  343. }
  344. this.requests.delete(requestKey);
  345. }));
  346. }
  347. cancel() {
  348. this.cancelled = true;
  349. const resourceURLs = [...this.requests.keys()];
  350. resourceURLs.forEach(requestKey => {
  351. const resourceRequests = this.requests.get(requestKey);
  352. resourceRequests.forEach(callbacks => callbacks.reject());
  353. this.requests.delete(requestKey);
  354. });
  355. }
  356. }
  357. // ---------
  358. // Processor
  359. // ---------
  360. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  361. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  362. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  363. const SCRIPT_TAG_FOUND = /<script/gi;
  364. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  365. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  366. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  367. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  368. const UTF8_CHARSET = "utf-8";
  369. class Processor {
  370. constructor(options, batchRequest) {
  371. this.options = options;
  372. this.stats = new Stats(options);
  373. this.baseURI = normalizeURL(options.baseURI || options.url);
  374. this.batchRequest = batchRequest;
  375. this.stylesheets = new Map();
  376. this.styles = new Map();
  377. this.cssVariables = new Map();
  378. }
  379. initialize() {
  380. this.options.saveDate = new Date();
  381. this.options.saveUrl = this.options.url;
  382. if (this.options.enableMaff) {
  383. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  384. }
  385. this.maxResources = this.batchRequest.getMaxResources();
  386. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  387. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  388. }
  389. if (!this.options.removeImports && this.options.imports) {
  390. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  391. }
  392. this.stats.set("processed", "resources", this.maxResources);
  393. }
  394. async loadPage(pageContent, charset) {
  395. let content;
  396. if (!pageContent || this.options.saveRawPage) {
  397. content = await util.getContent(this.baseURI, {
  398. maxResourceSize: this.options.maxResourceSize,
  399. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  400. charset,
  401. frameId: this.options.windowId
  402. });
  403. pageContent = content.data;
  404. }
  405. this.doc = util.parseDocContent(pageContent, this.baseURI);
  406. if (this.options.saveRawPage) {
  407. let charset;
  408. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  409. const charsetDeclaration = element.content.split(";")[1];
  410. if (charsetDeclaration && !charset) {
  411. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  412. }
  413. });
  414. if (charset && charset != content.charset) {
  415. return this.loadPage(pageContent, charset);
  416. }
  417. }
  418. this.workStyleElement = this.doc.createElement("style");
  419. this.doc.body.appendChild(this.workStyleElement);
  420. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  421. }
  422. finalize() {
  423. if (this.workStyleElement.parentNode) {
  424. this.workStyleElement.remove();
  425. }
  426. }
  427. async getPageData() {
  428. util.postProcessDoc(this.doc);
  429. const url = util.parseURL(this.baseURI);
  430. if (this.options.insertSingleFileComment) {
  431. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  432. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  433. " \n url: " + this.options.saveUrl +
  434. " \n saved date: " + this.options.saveDate +
  435. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  436. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  437. }
  438. if (this.options.insertCanonicalLink) {
  439. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  440. if (!canonicalLink) {
  441. canonicalLink = this.doc.createElement("link");
  442. canonicalLink.setAttribute("rel", "canonical");
  443. this.doc.head.appendChild(canonicalLink);
  444. }
  445. if (canonicalLink && !canonicalLink.href) {
  446. canonicalLink.href = this.options.saveUrl;
  447. }
  448. }
  449. let size;
  450. if (this.options.displayStats) {
  451. size = util.getContentSize(this.doc.documentElement.outerHTML);
  452. }
  453. const content = util.serialize(this.doc, this.options.compressHTML);
  454. if (this.options.displayStats) {
  455. const contentSize = util.getContentSize(content);
  456. this.stats.set("processed", "HTML bytes", contentSize);
  457. this.stats.add("discarded", "HTML bytes", size - contentSize);
  458. }
  459. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  460. const replacementCharacter = this.options.filenameReplacementCharacter;
  461. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  462. if (!this.options.backgroundSave) {
  463. filename = filename.replace(/\//g, replacementCharacter);
  464. }
  465. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  466. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  467. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  468. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  469. filename = filename + "…" + extension;
  470. }
  471. if (!filename) {
  472. filename = "Unnamed page";
  473. }
  474. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  475. const pageData = {
  476. stats: this.stats.data,
  477. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  478. filename,
  479. content
  480. };
  481. if (this.options.addProof) {
  482. pageData.hash = await util.digest("SHA-256", content);
  483. }
  484. if (this.options.retrieveLinks) {
  485. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  486. }
  487. return pageData;
  488. }
  489. preProcessPage() {
  490. if (this.options.win) {
  491. 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));
  492. }
  493. if (this.options.images && !this.options.saveRawPage) {
  494. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  495. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  496. if (attributeValue) {
  497. const imageData = this.options.images[Number(attributeValue)];
  498. if (imageData) {
  499. if (this.options.removeHiddenElements && (
  500. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  501. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  502. )) {
  503. imgElement.setAttribute("src", EMPTY_IMAGE);
  504. } else {
  505. if (imageData.currentSrc) {
  506. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  507. imgElement.setAttribute("src", imageData.currentSrc);
  508. }
  509. if (this.options.loadDeferredImages) {
  510. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  511. imageData.src = imgElement.dataset.src;
  512. imgElement.setAttribute("src", imgElement.dataset.src);
  513. imgElement.removeAttribute("data-src");
  514. }
  515. }
  516. }
  517. }
  518. }
  519. });
  520. if (this.options.loadDeferredImages) {
  521. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  522. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  523. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  524. imgElement.removeAttribute("data-srcset");
  525. }
  526. });
  527. }
  528. }
  529. }
  530. replaceStyleContents() {
  531. if (this.options.stylesheets) {
  532. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  533. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  534. if (attributeValue) {
  535. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  536. if (stylesheetContent) {
  537. styleElement.textContent = stylesheetContent;
  538. }
  539. }
  540. });
  541. }
  542. }
  543. removeUnselectedElements() {
  544. removeUnmarkedElements(this.doc.body);
  545. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  546. function removeUnmarkedElements(element) {
  547. let selectedElementFound = false;
  548. Array.from(element.childNodes).forEach(node => {
  549. if (node.nodeType == 1) {
  550. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  551. selectedElementFound = selectedElementFound || isSelectedElement;
  552. if (isSelectedElement) {
  553. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  554. removeUnmarkedElements(node);
  555. } else if (selectedElementFound) {
  556. removeNode(node);
  557. } else {
  558. hideNode(node);
  559. }
  560. }
  561. });
  562. }
  563. function removeNode(node) {
  564. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  565. node.remove();
  566. } else {
  567. hideNode(node);
  568. }
  569. }
  570. function hideNode(node) {
  571. if (canHideNode(node)) {
  572. node.style.setProperty("display", "none", "important");
  573. Array.from(node.childNodes).forEach(removeNode);
  574. }
  575. }
  576. function canHideNode(node) {
  577. if (node.nodeType == 1) {
  578. const tagName = node.tagName && node.tagName.toLowerCase();
  579. return (tagName != "svg" && tagName != "style" && tagName != "link");
  580. }
  581. }
  582. }
  583. insertVideoPosters() {
  584. if (this.options.posters) {
  585. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  586. let videoElement;
  587. if (element.tagName == "VIDEO") {
  588. videoElement = element;
  589. } else {
  590. videoElement = element.parentElement;
  591. }
  592. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  593. if (attributeValue) {
  594. const posterURL = this.options.posters[Number(attributeValue)];
  595. if (!videoElement.poster && posterURL) {
  596. videoElement.setAttribute("poster", posterURL);
  597. }
  598. }
  599. });
  600. }
  601. }
  602. removeFrames() {
  603. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  604. this.stats.set("discarded", "frames", frameElements.length);
  605. this.stats.set("processed", "frames", frameElements.length);
  606. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  607. }
  608. removeScripts() {
  609. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  610. this.doc.querySelectorAll("[href]").forEach(element => {
  611. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  612. element.setAttribute("href", "");
  613. }
  614. });
  615. this.doc.querySelectorAll("[src]").forEach(element => {
  616. if (element.src && element.src.match(/^\s*javascript:/)) {
  617. element.removeAttribute("src");
  618. }
  619. });
  620. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  621. this.stats.set("discarded", "scripts", scriptElements.length);
  622. this.stats.set("processed", "scripts", scriptElements.length);
  623. scriptElements.forEach(element => element.remove());
  624. }
  625. removeVideoSources() {
  626. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  627. this.stats.set("discarded", "video sources", videoSourceElements.length);
  628. this.stats.set("processed", "video sources", videoSourceElements.length);
  629. videoSourceElements.forEach(element => {
  630. if (element.tagName == "SOURCE") {
  631. element.remove();
  632. } else {
  633. videoSourceElements.forEach(element => element.removeAttribute("src"));
  634. }
  635. });
  636. }
  637. removeAudioSources() {
  638. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  639. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  640. this.stats.set("processed", "audio sources", audioSourceElements.length);
  641. audioSourceElements.forEach(element => {
  642. if (element.tagName == "SOURCE") {
  643. element.remove();
  644. } else {
  645. audioSourceElements.forEach(element => element.removeAttribute("src"));
  646. }
  647. });
  648. }
  649. removeDiscardedResources() {
  650. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  651. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  652. 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\"])");
  653. this.stats.set("discarded", "objects", objectElements.length);
  654. this.stats.set("processed", "objects", objectElements.length);
  655. objectElements.forEach(element => element.remove());
  656. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  657. replacedAttributeValue.forEach(element => {
  658. let regExp;
  659. if (this.options.removeScripts) {
  660. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  661. } else {
  662. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  663. }
  664. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  665. if (relValue.length) {
  666. element.setAttribute("rel", relValue);
  667. } else {
  668. element.remove();
  669. }
  670. });
  671. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  672. if (this.options.compressHTML) {
  673. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  674. }
  675. if (!this.options.saveFavicon) {
  676. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  677. }
  678. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  679. }
  680. resetCharsetMeta() {
  681. let charset;
  682. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  683. const charsetDeclaration = element.content.split(";")[1];
  684. if (charsetDeclaration && !charset) {
  685. charset = charsetDeclaration.split("=")[1];
  686. if (charset) {
  687. this.charset = charset.trim().toLowerCase();
  688. }
  689. }
  690. element.remove();
  691. });
  692. const metaElement = this.doc.createElement("meta");
  693. metaElement.setAttribute("charset", UTF8_CHARSET);
  694. if (this.doc.head.firstChild) {
  695. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  696. } else {
  697. this.doc.head.appendChild(metaElement);
  698. }
  699. }
  700. setInputValues() {
  701. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  702. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  703. input.setAttribute("value", value || "");
  704. });
  705. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  706. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  707. if (value == "true") {
  708. input.setAttribute("checked", "");
  709. }
  710. });
  711. this.doc.querySelectorAll("textarea").forEach(textarea => {
  712. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  713. textarea.textContent = value || "";
  714. });
  715. this.doc.querySelectorAll("select").forEach(select => {
  716. select.querySelectorAll("option").forEach(option => {
  717. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  718. if (selected) {
  719. option.setAttribute("selected", "");
  720. }
  721. });
  722. });
  723. }
  724. saveFavicon() {
  725. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  726. if (!faviconElement) {
  727. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  728. }
  729. if (!faviconElement) {
  730. faviconElement = this.doc.createElement("link");
  731. faviconElement.setAttribute("type", "image/x-icon");
  732. faviconElement.setAttribute("rel", "shortcut icon");
  733. faviconElement.setAttribute("href", "/favicon.ico");
  734. }
  735. this.doc.head.appendChild(faviconElement);
  736. }
  737. replaceCanvasElements() {
  738. if (this.options.canvases) {
  739. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  740. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  741. if (attributeValue) {
  742. const canvasData = this.options.canvases[Number(attributeValue)];
  743. if (canvasData) {
  744. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  745. this.stats.add("processed", "canvas", 1);
  746. }
  747. }
  748. });
  749. }
  750. }
  751. insertFonts() {
  752. if (this.options.fonts && this.options.fonts.length) {
  753. let stylesheetContent = "";
  754. this.options.fonts.forEach(fontData => {
  755. if (fontData["font-family"] && fontData.src) {
  756. stylesheetContent += "@font-face{";
  757. let stylesContent = "";
  758. Object.keys(fontData).forEach(fontStyle => {
  759. if (stylesContent) {
  760. stylesContent += ";";
  761. }
  762. stylesContent += fontStyle + ":" + fontData[fontStyle];
  763. });
  764. stylesheetContent += stylesContent + "}";
  765. }
  766. });
  767. if (stylesheetContent) {
  768. const styleElement = this.doc.createElement("style");
  769. styleElement.textContent = stylesheetContent;
  770. const existingStyleElement = this.doc.querySelector("style");
  771. if (existingStyleElement) {
  772. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  773. } else {
  774. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  775. }
  776. }
  777. }
  778. }
  779. removeHiddenElements() {
  780. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  781. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  782. this.stats.set("discarded", "hidden elements", removedElements.length);
  783. this.stats.set("processed", "hidden elements", removedElements.length);
  784. if (hiddenElements.length) {
  785. const styleElement = this.doc.createElement("style");
  786. styleElement.textContent = ".sf-hidden{display:none!important;}";
  787. this.doc.head.appendChild(styleElement);
  788. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  789. }
  790. removedElements.forEach(element => element.remove());
  791. }
  792. resolveHrefs() {
  793. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  794. const href = element.getAttribute("href").trim();
  795. if (!testIgnoredPath(href)) {
  796. let resolvedURL;
  797. try {
  798. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  799. } catch (error) {
  800. // ignored
  801. }
  802. if (resolvedURL) {
  803. const url = normalizeURL(this.options.url);
  804. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  805. resolvedURL = resolvedURL.substring(url.length);
  806. }
  807. try {
  808. element.setAttribute("href", resolvedURL);
  809. } catch (error) {
  810. // ignored
  811. }
  812. }
  813. }
  814. });
  815. }
  816. resolveStyleAttributeURLs() {
  817. this.doc.querySelectorAll("[style]").forEach(element => {
  818. let styleContent = element.getAttribute("style");
  819. if (this.options.compressCSS) {
  820. styleContent = util.compressCSS(styleContent);
  821. }
  822. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  823. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  824. this.styles.set(element, declarationList);
  825. });
  826. }
  827. async resolveStylesheetURLs() {
  828. const options = {
  829. maxResourceSize: this.options.maxResourceSize,
  830. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  831. url: this.options.url,
  832. charset: this.charset,
  833. compressCSS: this.options.compressCSS,
  834. updatedResources: this.options.updatedResources,
  835. rootDocument: this.options.rootDocument,
  836. frameId: this.options.windowId
  837. };
  838. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  839. .map(async element => {
  840. let mediaText;
  841. if (element.media) {
  842. mediaText = element.media.toLowerCase();
  843. }
  844. const stylesheetInfo = { mediaText };
  845. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  846. stylesheetInfo.scoped = true;
  847. }
  848. if (element.tagName.toLowerCase() == "link") {
  849. if (element.charset) {
  850. options.charset = element.charset;
  851. }
  852. }
  853. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  854. }));
  855. if (options.rootDocument) {
  856. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  857. await Promise.all(newResources.map(async resource => {
  858. resource.retrieved = true;
  859. const stylesheetInfo = {};
  860. const element = this.doc.createElement("style");
  861. this.doc.body.appendChild(element);
  862. element.textContent = resource.content;
  863. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  864. }));
  865. }
  866. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  867. stylesheets.set(element, stylesheetInfo);
  868. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  869. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  870. if (match && match[1] && match[1] != options.charset) {
  871. options.charset = match[1];
  872. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  873. }
  874. let stylesheet;
  875. try {
  876. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  877. } catch (error) {
  878. // ignored
  879. }
  880. if (stylesheet && stylesheet.children) {
  881. if (options.compressCSS) {
  882. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  883. }
  884. stylesheetInfo.stylesheet = stylesheet;
  885. } else {
  886. stylesheets.delete(element);
  887. }
  888. }
  889. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  890. let content;
  891. if (element.tagName.toLowerCase() == "link") {
  892. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  893. } else {
  894. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  895. }
  896. return content || "";
  897. }
  898. }
  899. async resolveFrameURLs() {
  900. if (!this.options.saveRawPage) {
  901. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  902. await Promise.all(frameElements.map(async frameElement => {
  903. if (frameElement.tagName == "OBJECT") {
  904. frameElement.setAttribute("data", "data:text/html,");
  905. } else {
  906. frameElement.removeAttribute("src");
  907. frameElement.removeAttribute("srcdoc");
  908. }
  909. Array.from(frameElement.childNodes).forEach(node => node.remove());
  910. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  911. if (this.options.frames && frameWindowId) {
  912. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  913. if (frameData) {
  914. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  915. }
  916. }
  917. }));
  918. }
  919. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  920. options.insertSingleFileComment = false;
  921. options.insertCanonicalLink = false;
  922. options.saveFavicon = false;
  923. options.url = frameData.baseURI;
  924. options.windowId = frameWindowId;
  925. if (frameData.content) {
  926. options.content = frameData.content;
  927. options.canvases = frameData.canvases;
  928. options.fonts = frameData.fonts;
  929. options.stylesheets = frameData.stylesheets;
  930. options.images = frameData.images;
  931. options.posters = frameData.posters;
  932. options.usedFonts = frameData.usedFonts;
  933. options.shadowRoots = frameData.shadowRoots;
  934. options.imports = frameData.imports;
  935. frameData.runner = new Runner(options);
  936. frameData.frameElement = frameElement;
  937. await frameData.runner.loadPage();
  938. await frameData.runner.initialize();
  939. frameData.maxResources = batchRequest.getMaxResources();
  940. }
  941. }
  942. }
  943. insertShadowRootContents() {
  944. const doc = this.doc;
  945. const options = this.options;
  946. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  947. processElement(this.doc);
  948. if (this.options.removeScripts) {
  949. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  950. }
  951. const scriptElement = doc.createElement("script");
  952. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  953. 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)}})}})()`;
  954. doc.body.appendChild(scriptElement);
  955. }
  956. function processElement(element) {
  957. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  958. shadowRootElements.forEach(element => {
  959. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  960. if (attributeValue) {
  961. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  962. if (shadowRootData) {
  963. const templateElement = doc.createElement("template");
  964. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  965. if (shadowRootData.delegatesFocus) {
  966. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  967. }
  968. if (shadowRootData.adoptedStyleSheets) {
  969. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  970. const styleElement = doc.createElement("style");
  971. styleElement.textContent = stylesheetContent;
  972. templateElement.appendChild(styleElement);
  973. });
  974. }
  975. const shadowDoc = util.parseDocContent(shadowRootData.content);
  976. if (shadowDoc.head) {
  977. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  978. if (metaCharset) {
  979. metaCharset.remove();
  980. }
  981. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  982. }
  983. if (shadowDoc.body) {
  984. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  985. }
  986. processElement(templateElement);
  987. if (element.firstChild) {
  988. element.insertBefore(templateElement, element.firstChild);
  989. } else {
  990. element.appendChild(templateElement);
  991. }
  992. }
  993. }
  994. });
  995. }
  996. }
  997. async resolveHtmlImportURLs() {
  998. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  999. await Promise.all(linkElements.map(async linkElement => {
  1000. const resourceURL = linkElement.href;
  1001. linkElement.removeAttribute("href");
  1002. const options = Object.create(this.options);
  1003. options.insertSingleFileComment = false;
  1004. options.insertCanonicalLink = false;
  1005. options.saveFavicon = false;
  1006. options.removeUnusedStyles = false;
  1007. options.removeAlternativeMedias = false;
  1008. options.removeUnusedFonts = false;
  1009. options.removeHiddenElements = false;
  1010. options.url = resourceURL;
  1011. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1012. if (attributeValue) {
  1013. const importData = options.imports[Number(attributeValue)];
  1014. if (importData) {
  1015. options.content = importData.content;
  1016. importData.runner = new Runner(options);
  1017. await importData.runner.loadPage();
  1018. await importData.runner.initialize();
  1019. if (!options.removeImports) {
  1020. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1021. }
  1022. importData.runner.getStyleSheets().forEach(stylesheet => {
  1023. const importedStyleElement = this.doc.createElement("style");
  1024. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1025. this.stylesheets.set(importedStyleElement, stylesheet);
  1026. });
  1027. }
  1028. }
  1029. if (options.removeImports) {
  1030. linkElement.remove();
  1031. this.stats.add("discarded", "HTML imports", 1);
  1032. }
  1033. }));
  1034. }
  1035. removeUnusedStyles() {
  1036. if (!this.mediaAllInfo) {
  1037. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1038. }
  1039. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1040. this.stats.set("processed", "CSS rules", stats.processed);
  1041. this.stats.set("discarded", "CSS rules", stats.discarded);
  1042. }
  1043. removeUnusedFonts() {
  1044. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1045. }
  1046. removeAlternativeMedias() {
  1047. const stats = util.minifyMedias(this.stylesheets);
  1048. this.stats.set("processed", "medias", stats.processed);
  1049. this.stats.set("discarded", "medias", stats.discarded);
  1050. }
  1051. async processStylesheets() {
  1052. this.options.fontURLs = new Map();
  1053. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1054. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1055. ));
  1056. }
  1057. async processStyleAttributes() {
  1058. return Promise.all([...this.styles].map(([, declarationList]) =>
  1059. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1060. ));
  1061. }
  1062. async processPageResources() {
  1063. const processAttributeArgs = [
  1064. ["link[href][rel*=\"icon\"]", "href", false, true],
  1065. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1066. ["img[src], input[src][type=image]", "src", true],
  1067. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1068. ["video[poster]", "poster"],
  1069. ["*[background]", "background"],
  1070. ["image", "xlink:href"],
  1071. ["image", "href"]
  1072. ];
  1073. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1074. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1075. );
  1076. resourcePromises = resourcePromises.concat([
  1077. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1078. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1079. ]);
  1080. if (!this.options.removeAudioSrc) {
  1081. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1082. }
  1083. if (!this.options.removeVideoSrc) {
  1084. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1085. }
  1086. await Promise.all(resourcePromises);
  1087. if (this.options.saveFavicon) {
  1088. ProcessorHelper.processShortcutIcons(this.doc);
  1089. }
  1090. }
  1091. async processScripts() {
  1092. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1093. let resourceURL;
  1094. let scriptSrc;
  1095. if (element.tagName == "SCRIPT") {
  1096. scriptSrc = element.getAttribute("src");
  1097. element.removeAttribute("src");
  1098. } else {
  1099. scriptSrc = element.getAttribute("href");
  1100. element.removeAttribute("href");
  1101. }
  1102. element.removeAttribute("integrity");
  1103. element.textContent = "";
  1104. try {
  1105. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1106. } catch (error) {
  1107. // ignored
  1108. }
  1109. if (testValidURL(resourceURL)) {
  1110. this.stats.add("processed", "scripts", 1);
  1111. const content = await util.getContent(resourceURL, {
  1112. asBinary: true,
  1113. charset: this.charset != UTF8_CHARSET && this.charset,
  1114. maxResourceSize: this.options.maxResourceSize,
  1115. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1116. frameId: this.options.windowId
  1117. });
  1118. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1119. if (element.tagName == "SCRIPT") {
  1120. element.setAttribute("src", content.data);
  1121. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1122. element.setAttribute("async", "");
  1123. }
  1124. } else {
  1125. const scriptElement = this.doc.createElement("script");
  1126. scriptElement.setAttribute("src", content.data);
  1127. scriptElement.setAttribute("async", "");
  1128. element.parentElement.replaceChild(scriptElement, element);
  1129. }
  1130. }
  1131. }));
  1132. }
  1133. removeAlternativeImages() {
  1134. util.removeAlternativeImages(this.doc);
  1135. }
  1136. async removeAlternativeFonts() {
  1137. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs);
  1138. }
  1139. async processFrames() {
  1140. if (this.options.frames) {
  1141. const frameElements = Array.from(this.doc.querySelectorAll("iframe:not([" + util.HIDDEN_FRAME_ATTRIBUTE_NAME + "]), frame, object[type=\"text/html\"][data]"));
  1142. await Promise.all(frameElements.map(async frameElement => {
  1143. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1144. if (frameWindowId) {
  1145. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1146. if (frameData) {
  1147. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1148. if (frameData.runner) {
  1149. this.stats.add("processed", "frames", 1);
  1150. await frameData.runner.run();
  1151. const pageData = await frameData.runner.getPageData();
  1152. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1153. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1154. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1155. sandbox += " allow-scripts allow-same-origin";
  1156. }
  1157. frameElement.setAttribute("sandbox", sandbox);
  1158. if (frameElement.tagName == "OBJECT") {
  1159. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1160. } else {
  1161. if (frameElement.tagName == "FRAME") {
  1162. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1163. } else {
  1164. frameElement.setAttribute("srcdoc", pageData.content);
  1165. frameElement.removeAttribute("src");
  1166. }
  1167. }
  1168. this.stats.addAll(pageData);
  1169. } else {
  1170. this.stats.add("discarded", "frames", 1);
  1171. }
  1172. }
  1173. }
  1174. }));
  1175. }
  1176. }
  1177. async processHtmlImports() {
  1178. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1179. await Promise.all(linkElements.map(async linkElement => {
  1180. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1181. if (attributeValue) {
  1182. const importData = this.options.imports[Number(attributeValue)];
  1183. if (importData.runner) {
  1184. this.stats.add("processed", "HTML imports", 1);
  1185. await importData.runner.run();
  1186. const pageData = await importData.runner.getPageData();
  1187. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1188. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1189. this.stats.addAll(pageData);
  1190. } else {
  1191. this.stats.add("discarded", "HTML imports", 1);
  1192. }
  1193. }
  1194. }));
  1195. }
  1196. replaceStylesheets() {
  1197. this.doc.querySelectorAll("style").forEach(styleElement => {
  1198. const stylesheetInfo = this.stylesheets.get(styleElement);
  1199. if (stylesheetInfo) {
  1200. this.stylesheets.delete(styleElement);
  1201. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1202. styleElement.textContent = stylesheetContent;
  1203. if (stylesheetInfo.mediaText) {
  1204. styleElement.media = stylesheetInfo.mediaText;
  1205. }
  1206. } else {
  1207. styleElement.remove();
  1208. }
  1209. });
  1210. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1211. const stylesheetInfo = this.stylesheets.get(linkElement);
  1212. if (stylesheetInfo) {
  1213. this.stylesheets.delete(linkElement);
  1214. const styleElement = this.doc.createElement("style");
  1215. if (stylesheetInfo.mediaText) {
  1216. styleElement.media = stylesheetInfo.mediaText;
  1217. }
  1218. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1219. styleElement.textContent = stylesheetContent;
  1220. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1221. } else {
  1222. linkElement.remove();
  1223. }
  1224. });
  1225. }
  1226. replaceStyleAttributes() {
  1227. this.doc.querySelectorAll("[style]").forEach(element => {
  1228. const declarations = this.styles.get(element);
  1229. if (declarations) {
  1230. this.styles.delete(element);
  1231. let styleContent = cssTree.generate(declarations);
  1232. element.setAttribute("style", styleContent);
  1233. } else {
  1234. element.setAttribute("style", "");
  1235. }
  1236. });
  1237. }
  1238. insertVariables() {
  1239. if (this.cssVariables.size) {
  1240. const styleElement = this.doc.createElement("style");
  1241. const firstStyleElement = this.doc.head.querySelector("style");
  1242. if (firstStyleElement) {
  1243. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1244. } else {
  1245. this.doc.head.appendChild(styleElement);
  1246. }
  1247. let stylesheetContent = "";
  1248. this.cssVariables.forEach((content, indexResource) => {
  1249. this.cssVariables.delete(indexResource);
  1250. if (stylesheetContent) {
  1251. stylesheetContent += ";";
  1252. }
  1253. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1254. });
  1255. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1256. }
  1257. }
  1258. compressHTML() {
  1259. let size;
  1260. if (this.options.displayStats) {
  1261. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1262. }
  1263. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1264. if (this.options.displayStats) {
  1265. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1266. }
  1267. }
  1268. cleanupPage() {
  1269. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1270. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1271. if (metaCharset) {
  1272. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1273. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1274. this.doc.head.querySelector("meta[charset]").remove();
  1275. }
  1276. }
  1277. }
  1278. async insertMAFFMetaData() {
  1279. const maffMetaData = await this.maffMetaDataPromise;
  1280. if (maffMetaData && maffMetaData.content) {
  1281. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1282. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1283. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1284. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1285. if (originalURLElement) {
  1286. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1287. }
  1288. if (archiveTimeElement) {
  1289. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1290. if (value) {
  1291. const date = new Date(value);
  1292. if (!isNaN(date.getTime())) {
  1293. this.options.saveDate = new Date(value);
  1294. }
  1295. }
  1296. }
  1297. }
  1298. }
  1299. async setDocInfo() {
  1300. const titleElement = this.doc.querySelector("title");
  1301. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1302. const authorElement = this.doc.querySelector("meta[name=author]");
  1303. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1304. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1305. const headingElement = this.doc.querySelector("h1");
  1306. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1307. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1308. this.options.info = {
  1309. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1310. lang: this.doc.documentElement.lang,
  1311. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1312. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1313. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1314. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1315. };
  1316. }
  1317. }
  1318. // ---------------
  1319. // ProcessorHelper
  1320. // ---------------
  1321. const DATA_URI_PREFIX = "data:";
  1322. const ABOUT_BLANK_URI = "about:blank";
  1323. const EMPTY_DATA_URI = "data:null;base64,";
  1324. const REGEXP_URL_HASH = /(#.+?)$/;
  1325. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1326. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1327. class ProcessorHelper {
  1328. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1329. const date = options.saveDate;
  1330. const url = util.parseURL(options.saveUrl);
  1331. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1332. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1333. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1334. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1335. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1336. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1337. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1338. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1339. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1340. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1341. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1342. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1343. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1344. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1345. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1346. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1347. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1348. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1349. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1350. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1351. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1352. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1353. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1354. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1357. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1358. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1359. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1360. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1361. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1362. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1363. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1364. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1365. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1366. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1367. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1368. for (const [name, value] of params) {
  1369. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1370. }
  1371. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1372. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1376. if (content) {
  1377. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1380. }
  1381. return template.trim();
  1382. function decode(value) {
  1383. try {
  1384. return decodeURI(value);
  1385. } catch (error) {
  1386. return value;
  1387. }
  1388. }
  1389. }
  1390. static setBackgroundImage(element, url, style) {
  1391. element.style.setProperty("background-blend-mode", "normal", "important");
  1392. element.style.setProperty("background-clip", "content-box", "important");
  1393. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1394. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1395. element.style.setProperty("background-image", url, "important");
  1396. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1397. element.style.setProperty("background-origin", "content-box", "important");
  1398. element.style.setProperty("background-repeat", "no-repeat", "important");
  1399. }
  1400. static processShortcutIcons(doc) {
  1401. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1402. if (!shortcutIcon) {
  1403. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1404. if (shortcutIcon) {
  1405. shortcutIcon.rel = "icon";
  1406. }
  1407. }
  1408. if (shortcutIcon) {
  1409. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1410. if (linkElement != shortcutIcon) {
  1411. linkElement.remove();
  1412. }
  1413. });
  1414. }
  1415. }
  1416. static removeSingleLineCssComments(stylesheet) {
  1417. const removedRules = [];
  1418. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1419. const ruleData = cssRule.data;
  1420. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1421. removedRules.push(cssRule);
  1422. }
  1423. }
  1424. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1425. }
  1426. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1427. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1428. const imports = getImportFunctions(stylesheetContent);
  1429. await Promise.all(imports.map(async cssImport => {
  1430. const match = matchImport(cssImport);
  1431. if (match) {
  1432. const regExpCssImport = getRegExp(cssImport);
  1433. let resourceURL = normalizeURL(match.resourceURL);
  1434. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1435. try {
  1436. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1437. } catch (error) {
  1438. // ignored
  1439. }
  1440. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1441. const content = await util.getContent(resourceURL, {
  1442. maxResourceSize: options.maxResourceSize,
  1443. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1444. validateTextContentType: true,
  1445. frameId: options.windowId
  1446. });
  1447. resourceURL = content.resourceURL;
  1448. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1449. let importedStylesheetContent = removeCssComments(content.data);
  1450. if (options.compressCSS) {
  1451. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1452. }
  1453. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1454. if (stylesheetContent.includes(cssImport)) {
  1455. const ancestorStyleSheets = new Set(importedStyleSheets);
  1456. ancestorStyleSheets.add(resourceURL);
  1457. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1458. workStylesheet.textContent = importedStylesheetContent;
  1459. if (workStylesheet.sheet.cssRules.length) {
  1460. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1461. } else {
  1462. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1463. }
  1464. }
  1465. } else {
  1466. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1467. }
  1468. } else {
  1469. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1470. }
  1471. }
  1472. }));
  1473. return stylesheetContent;
  1474. }
  1475. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1476. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1477. urlFunctions.map(urlFunction => {
  1478. const originalResourceURL = matchURL(urlFunction);
  1479. const resourceURL = normalizeURL(originalResourceURL);
  1480. if (!testIgnoredPath(resourceURL)) {
  1481. if (!resourceURL || testValidPath(resourceURL)) {
  1482. let resolvedURL;
  1483. if (!originalResourceURL.startsWith("#")) {
  1484. try {
  1485. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1486. } catch (error) {
  1487. // ignored
  1488. }
  1489. }
  1490. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1491. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1492. }
  1493. } else {
  1494. let newUrlFunction;
  1495. if (originalResourceURL) {
  1496. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1497. } else {
  1498. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1499. }
  1500. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1501. }
  1502. }
  1503. });
  1504. return stylesheetContent;
  1505. }
  1506. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1507. resourceURL = normalizeURL(resourceURL);
  1508. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1509. const content = await util.getContent(resourceURL, {
  1510. maxResourceSize: options.maxResourceSize,
  1511. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1512. charset: options.charset,
  1513. frameId: options.windowId,
  1514. validateTextContentType: true
  1515. });
  1516. resourceURL = content.resourceURL;
  1517. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1518. let stylesheetContent = removeCssComments(content.data);
  1519. if (options.compressCSS) {
  1520. stylesheetContent = util.compressCSS(stylesheetContent);
  1521. }
  1522. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1523. return stylesheetContent;
  1524. }
  1525. }
  1526. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1527. const promises = [];
  1528. const removedRules = [];
  1529. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1530. const ruleData = cssRule.data;
  1531. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1532. removedRules.push(cssRule);
  1533. } else if (ruleData.block && ruleData.block.children) {
  1534. if (ruleData.type == "Rule") {
  1535. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1536. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1537. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1538. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1539. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1540. }
  1541. }
  1542. }
  1543. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1544. await Promise.all(promises);
  1545. async function processFontFaceRule(ruleData, fontURLs) {
  1546. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1547. if (declaration.type == "Declaration" && declaration.value.children) {
  1548. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1549. await Promise.all(urlFunctions.map(async urlFunction => {
  1550. const originalResourceURL = matchURL(urlFunction);
  1551. const resourceURL = normalizeURL(originalResourceURL);
  1552. if (!testIgnoredPath(resourceURL)) {
  1553. if (testValidURL(resourceURL)) {
  1554. let { content } = await batchRequest.addURL(resourceURL, true);
  1555. let resourceURLs = fontURLs.get(declaration);
  1556. if (!resourceURLs) {
  1557. resourceURLs = [];
  1558. fontURLs.set(declaration, resourceURLs);
  1559. }
  1560. resourceURLs.push(resourceURL);
  1561. replaceURLs(declaration, originalResourceURL, content);
  1562. }
  1563. }
  1564. }));
  1565. }
  1566. }));
  1567. function replaceURLs(declaration, oldURL, newURL) {
  1568. declaration.value.children.forEach(token => {
  1569. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1570. token.value.value = newURL;
  1571. }
  1572. });
  1573. }
  1574. }
  1575. }
  1576. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1577. await Promise.all(declarations.map(async declaration => {
  1578. if (declaration.type == "Declaration" && declaration.value.children) {
  1579. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1580. await Promise.all(urlFunctions.map(async urlFunction => {
  1581. const originalResourceURL = matchURL(urlFunction);
  1582. const resourceURL = normalizeURL(originalResourceURL);
  1583. if (!testIgnoredPath(resourceURL)) {
  1584. if (testValidURL(resourceURL)) {
  1585. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1586. let variableDefined;
  1587. const tokens = [];
  1588. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1589. if (!originalResourceURL.startsWith("#")) {
  1590. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1591. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1592. tokens.push({ parent, token, value });
  1593. variableDefined = true;
  1594. } else {
  1595. token.data.value.value = content;
  1596. }
  1597. }
  1598. });
  1599. if (variableDefined) {
  1600. cssVariables.set(indexResource, content);
  1601. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1602. }
  1603. }
  1604. }
  1605. }));
  1606. }
  1607. }));
  1608. function findURLToken(url, children, callback, depth = 0) {
  1609. for (let token = children.head; token; token = token.next) {
  1610. if (token.data.children) {
  1611. findURLToken(url, token.data.children, callback, depth + 1);
  1612. }
  1613. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1614. callback(token, children, depth == 0);
  1615. }
  1616. }
  1617. }
  1618. }
  1619. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1620. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1621. let resourceURL = resourceElement.getAttribute(attributeName);
  1622. if (resourceURL != null) {
  1623. resourceURL = normalizeURL(resourceURL);
  1624. if (!testIgnoredPath(resourceURL)) {
  1625. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1626. if (testValidPath(resourceURL)) {
  1627. try {
  1628. resourceURL = util.resolveURL(resourceURL, baseURI);
  1629. } catch (error) {
  1630. // ignored
  1631. }
  1632. if (testValidURL(resourceURL)) {
  1633. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1634. if (resourceElement.dataset.singleFileOriginURL) {
  1635. let originURL = resourceElement.dataset.singleFileOriginURL;
  1636. delete resourceElement.dataset.singleFileOriginURL;
  1637. if (content == EMPTY_DATA_URI) {
  1638. try {
  1639. originURL = util.resolveURL(originURL, baseURI);
  1640. } catch (error) {
  1641. // ignored
  1642. }
  1643. try {
  1644. resourceURL = originURL;
  1645. content = (await util.getContent(resourceURL, {
  1646. asBinary: true,
  1647. maxResourceSize: options.maxResourceSize,
  1648. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1649. frameId: options.windowId
  1650. })).data;
  1651. } catch (error) {
  1652. // ignored
  1653. }
  1654. }
  1655. }
  1656. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1657. resourceElement.remove();
  1658. } else {
  1659. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1660. if (!forbiddenPrefixFound) {
  1661. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1662. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1663. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1664. cssVariables.set(indexResource, content);
  1665. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1666. styles.set(resourceElement, declarationList);
  1667. } else {
  1668. resourceElement.setAttribute(attributeName, content);
  1669. }
  1670. } else {
  1671. resourceElement.setAttribute(attributeName, content);
  1672. }
  1673. }
  1674. }
  1675. }
  1676. }
  1677. }
  1678. }
  1679. }));
  1680. }
  1681. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1682. let attributeName = "xlink:href";
  1683. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1684. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1685. if (originalResourceURL == null) {
  1686. attributeName = "href";
  1687. originalResourceURL = resourceElement.getAttribute(attributeName);
  1688. }
  1689. let resourceURL = normalizeURL(originalResourceURL);
  1690. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1691. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1692. try {
  1693. resourceURL = util.resolveURL(resourceURL, baseURI);
  1694. } catch (error) {
  1695. // ignored
  1696. }
  1697. if (testValidURL(resourceURL)) {
  1698. const { content } = await batchRequest.addURL(resourceURL);
  1699. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1700. if (hashMatch && hashMatch[0]) {
  1701. let symbolElement;
  1702. try {
  1703. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1704. } catch (error) {
  1705. // ignored
  1706. }
  1707. if (symbolElement) {
  1708. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1709. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1710. }
  1711. } else {
  1712. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1713. }
  1714. }
  1715. } else if (resourceURL == options.url) {
  1716. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1717. }
  1718. }));
  1719. }
  1720. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1721. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1722. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1723. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1724. let resourceURL = normalizeURL(srcsetValue.url);
  1725. if (!testIgnoredPath(resourceURL)) {
  1726. if (testValidPath(resourceURL)) {
  1727. try {
  1728. resourceURL = util.resolveURL(resourceURL, baseURI);
  1729. } catch (error) {
  1730. // ignored
  1731. }
  1732. if (testValidURL(resourceURL)) {
  1733. const { content } = await batchRequest.addURL(resourceURL, true);
  1734. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1735. if (forbiddenPrefixFound) {
  1736. return "";
  1737. }
  1738. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1739. } else {
  1740. return "";
  1741. }
  1742. } else {
  1743. return "";
  1744. }
  1745. } else {
  1746. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1747. }
  1748. }));
  1749. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1750. }));
  1751. }
  1752. static replaceImageSource(imgElement, variableName, options) {
  1753. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1754. if (attributeValue) {
  1755. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1756. if (imageData && imageData.replaceable) {
  1757. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1758. const backgroundStyle = {};
  1759. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1760. if (backgroundSize) {
  1761. backgroundStyle["background-size"] = imageData.objectFit;
  1762. }
  1763. if (imageData.objectPosition) {
  1764. backgroundStyle["background-position"] = imageData.objectPosition;
  1765. }
  1766. if (imageData.backgroundColor) {
  1767. backgroundStyle["background-color"] = imageData.backgroundColor;
  1768. }
  1769. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1770. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1771. return true;
  1772. }
  1773. }
  1774. }
  1775. }
  1776. // ----
  1777. // Util
  1778. // ----
  1779. const BLOB_URI_PREFIX = "blob:";
  1780. const HTTP_URI_PREFIX = /^https?:\/\//;
  1781. const FILE_URI_PREFIX = /^file:\/\//;
  1782. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1783. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1784. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1785. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1786. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1787. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1788. 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;
  1789. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1790. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1791. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1792. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1793. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1794. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1795. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1796. function getUpdatedResourceContent(resourceURL, content, options) {
  1797. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1798. options.updatedResources[resourceURL].retrieved = true;
  1799. return options.updatedResources[resourceURL].content;
  1800. } else {
  1801. return content.data || "";
  1802. }
  1803. }
  1804. function normalizeURL(url) {
  1805. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1806. return url;
  1807. } else {
  1808. return url.split("#")[0];
  1809. }
  1810. }
  1811. function getCSSValue(value) {
  1812. let result = "";
  1813. try {
  1814. result = cssTree.generate(value);
  1815. } catch (error) {
  1816. // ignored
  1817. }
  1818. return result;
  1819. }
  1820. function getOnEventAttributeNames(doc) {
  1821. const element = doc.createElement("div");
  1822. const attributeNames = [];
  1823. for (const propertyName in element) {
  1824. if (propertyName.startsWith("on")) {
  1825. attributeNames.push(propertyName);
  1826. }
  1827. }
  1828. attributeNames.push("onunload");
  1829. return attributeNames;
  1830. }
  1831. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1832. let maxLength;
  1833. if (template) {
  1834. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1835. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1836. if (template.match(replaceRegExp)) {
  1837. const matchedLength = template.match(replaceRegExp)[0];
  1838. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1839. if (isNaN(maxLength) || maxLength <= 0) {
  1840. maxLength = null;
  1841. }
  1842. } else {
  1843. replaceRegExp = new RegExp(regExpVariable, "g");
  1844. }
  1845. if (template.match(replaceRegExp)) {
  1846. let value = await valueGetter();
  1847. if (!dontReplaceSlash) {
  1848. value = value.replace(/\/+/g, replacementCharacter);
  1849. }
  1850. if (maxLength) {
  1851. value = await util.truncateText(value, maxLength);
  1852. }
  1853. return template.replace(replaceRegExp, value);
  1854. }
  1855. }
  1856. return template;
  1857. }
  1858. function getLastSegment(url, replacementCharacter) {
  1859. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1860. if (!lastSegment) {
  1861. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1862. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1863. }
  1864. if (!lastSegment) {
  1865. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1866. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1867. }
  1868. if (!lastSegment) {
  1869. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1870. }
  1871. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1872. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1873. lastSegment = lastSegmentMatch[1];
  1874. }
  1875. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1876. return lastSegment;
  1877. }
  1878. function getRegExp(string) {
  1879. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1880. }
  1881. function getUrlFunctions(stylesheetContent, unique) {
  1882. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1883. if (unique) {
  1884. return [...new Set(result)];
  1885. } else {
  1886. return result;
  1887. }
  1888. }
  1889. function getImportFunctions(stylesheetContent) {
  1890. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1891. }
  1892. function findShortcutIcon(shortcutIcons) {
  1893. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1894. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1895. return shortcutIcons[0];
  1896. }
  1897. function matchURL(stylesheetContent) {
  1898. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1899. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1900. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1901. return match && match[1];
  1902. }
  1903. function testIgnoredPath(resourceURL) {
  1904. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1905. }
  1906. function testValidPath(resourceURL) {
  1907. return resourceURL && !resourceURL.match(EMPTY_URL);
  1908. }
  1909. function testValidURL(resourceURL) {
  1910. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1911. }
  1912. function matchImport(stylesheetContent) {
  1913. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1914. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1915. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1916. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1917. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1918. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1919. if (match) {
  1920. const [, resourceURL, media] = match;
  1921. return { resourceURL, media };
  1922. }
  1923. }
  1924. function removeCssComments(stylesheetContent) {
  1925. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  1926. }
  1927. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  1928. if (mediaQuery) {
  1929. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1930. } else {
  1931. return stylesheetContent;
  1932. }
  1933. }
  1934. function log(...args) {
  1935. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1936. }
  1937. // -----
  1938. // Stats
  1939. // -----
  1940. const STATS_DEFAULT_VALUES = {
  1941. discarded: {
  1942. "HTML bytes": 0,
  1943. "hidden elements": 0,
  1944. "HTML imports": 0,
  1945. scripts: 0,
  1946. objects: 0,
  1947. "audio sources": 0,
  1948. "video sources": 0,
  1949. frames: 0,
  1950. "CSS rules": 0,
  1951. canvas: 0,
  1952. stylesheets: 0,
  1953. resources: 0,
  1954. medias: 0
  1955. },
  1956. processed: {
  1957. "HTML bytes": 0,
  1958. "hidden elements": 0,
  1959. "HTML imports": 0,
  1960. scripts: 0,
  1961. objects: 0,
  1962. "audio sources": 0,
  1963. "video sources": 0,
  1964. frames: 0,
  1965. "CSS rules": 0,
  1966. canvas: 0,
  1967. stylesheets: 0,
  1968. resources: 0,
  1969. medias: 0
  1970. }
  1971. };
  1972. class Stats {
  1973. constructor(options) {
  1974. this.options = options;
  1975. if (options.displayStats) {
  1976. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1977. }
  1978. }
  1979. set(type, subType, value) {
  1980. if (this.options.displayStats) {
  1981. this.data[type][subType] = value;
  1982. }
  1983. }
  1984. add(type, subType, value) {
  1985. if (this.options.displayStats) {
  1986. this.data[type][subType] += value;
  1987. }
  1988. }
  1989. addAll(pageData) {
  1990. if (this.options.displayStats) {
  1991. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1992. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1993. }
  1994. }
  1995. }
  1996. return { getClass };
  1997. })();