single-file-core.js 86 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109
  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. this.root = root;
  148. this.options = options;
  149. this.options.url = this.options.url || (this.options.doc && this.options.doc.location.href);
  150. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  151. this.options.rootDocument = root;
  152. this.options.updatedResources = this.options.updatedResources || {};
  153. this.batchRequest = new BatchRequest();
  154. this.processor = new Processor(options, this.batchRequest);
  155. if (this.options.doc) {
  156. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  157. this.options.canvases = docData.canvases;
  158. this.options.fonts = docData.fonts;
  159. this.options.stylesheets = docData.stylesheets;
  160. this.options.images = docData.images;
  161. this.options.posters = docData.posters;
  162. this.options.usedFonts = docData.usedFonts;
  163. this.options.shadowRoots = docData.shadowRoots;
  164. this.options.imports = docData.imports;
  165. this.options.referrer = docData.referrer;
  166. this.markedElements = docData.markedElements;
  167. }
  168. if (this.options.saveRawPage) {
  169. this.options.removeFrames = true;
  170. }
  171. this.options.content = this.options.content || (this.options.doc ? util.serialize(this.options.doc) : null);
  172. this.onprogress = options.onprogress || (() => { });
  173. }
  174. async loadPage() {
  175. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  176. await this.processor.loadPage(this.options.content);
  177. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  178. }
  179. async initialize() {
  180. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  181. await this.executeStage(RESOLVE_URLS_STAGE);
  182. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  183. if (this.options.doc) {
  184. util.postProcessDoc(this.options.doc, this.markedElements);
  185. }
  186. this.options.doc = null;
  187. this.options.win = null;
  188. }
  189. cancel() {
  190. this.cancelled = true;
  191. this.batchRequest.cancel();
  192. if (this.root) {
  193. if (this.options.frames) {
  194. this.options.frames.forEach(cancelRunner);
  195. }
  196. if (this.options.imports) {
  197. this.options.imports.forEach(cancelRunner);
  198. }
  199. }
  200. function cancelRunner(resourceData) {
  201. if (resourceData.runner) {
  202. resourceData.runner.cancel();
  203. }
  204. }
  205. }
  206. async run() {
  207. if (this.root) {
  208. this.processor.initialize(this.batchRequest);
  209. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  210. }
  211. await this.batchRequest.run(detail => {
  212. detail.pageURL = this.options.url;
  213. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  214. }, this.options);
  215. await this.pendingPromises;
  216. await this.executeStage(REPLACE_DOCS_STAGE);
  217. await this.executeStage(POST_PROCESS_STAGE);
  218. this.processor.finalize();
  219. }
  220. getDocument() {
  221. return this.processor.doc;
  222. }
  223. getStyleSheets() {
  224. return this.processor.stylesheets;
  225. }
  226. getPageData() {
  227. if (this.root) {
  228. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  229. }
  230. return this.processor.getPageData();
  231. }
  232. async executeStage(step) {
  233. if (DEBUG) {
  234. log("**** STARTED STAGE", step, "****");
  235. }
  236. const frame = !this.root;
  237. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  238. STAGES[step].sequential.forEach(task => {
  239. let startTime;
  240. if (DEBUG) {
  241. startTime = Date.now();
  242. log(" -- STARTED task =", task.action);
  243. }
  244. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  245. if (!this.cancelled) {
  246. this.executeTask(task);
  247. }
  248. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  249. if (DEBUG) {
  250. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  251. }
  252. });
  253. let parallelTasksPromise;
  254. if (STAGES[step].parallel) {
  255. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  256. let startTime;
  257. if (DEBUG) {
  258. startTime = Date.now();
  259. log(" // STARTED task =", task.action);
  260. }
  261. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  262. if (!this.cancelled) {
  263. await this.executeTask(task);
  264. }
  265. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  266. if (DEBUG) {
  267. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  268. }
  269. }));
  270. } else {
  271. parallelTasksPromise = Promise.resolve();
  272. }
  273. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  274. if (DEBUG) {
  275. log("**** ENDED STAGE", step, "****");
  276. }
  277. return parallelTasksPromise;
  278. }
  279. executeTask(task) {
  280. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  281. return this.processor[task.action]();
  282. }
  283. }
  284. }
  285. // ------------
  286. // BatchRequest
  287. // ------------
  288. class BatchRequest {
  289. constructor() {
  290. this.requests = new Map();
  291. this.duplicates = new Map();
  292. }
  293. addURL(resourceURL, asBinary, groupDuplicates) {
  294. return new Promise((resolve, reject) => {
  295. const requestKey = JSON.stringify([resourceURL, asBinary]);
  296. let resourceRequests = this.requests.get(requestKey);
  297. if (!resourceRequests) {
  298. resourceRequests = [];
  299. this.requests.set(requestKey, resourceRequests);
  300. }
  301. const callbacks = { resolve, reject };
  302. resourceRequests.push(callbacks);
  303. if (groupDuplicates) {
  304. let duplicateRequests = this.duplicates.get(requestKey);
  305. if (!duplicateRequests) {
  306. duplicateRequests = [];
  307. this.duplicates.set(requestKey, duplicateRequests);
  308. }
  309. duplicateRequests.push(callbacks);
  310. }
  311. });
  312. }
  313. getMaxResources() {
  314. return this.requests.size;
  315. }
  316. run(onloadListener, options) {
  317. const resourceURLs = [...this.requests.keys()];
  318. let indexResource = 0;
  319. return Promise.all(resourceURLs.map(async requestKey => {
  320. const [resourceURL, asBinary] = JSON.parse(requestKey);
  321. const resourceRequests = this.requests.get(requestKey);
  322. try {
  323. const content = await util.getContent(resourceURL, {
  324. asBinary,
  325. maxResourceSize: options.maxResourceSize,
  326. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  327. frameId: options.windowId
  328. });
  329. indexResource = indexResource + 1;
  330. onloadListener({ url: resourceURL });
  331. if (!this.cancelled) {
  332. resourceRequests.forEach(callbacks => {
  333. const duplicateCallbacks = this.duplicates.get(requestKey);
  334. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  335. callbacks.resolve({ content: content.data, indexResource, duplicate });
  336. });
  337. }
  338. } catch (error) {
  339. indexResource = indexResource + 1;
  340. onloadListener({ url: resourceURL });
  341. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  342. }
  343. this.requests.delete(requestKey);
  344. }));
  345. }
  346. cancel() {
  347. this.cancelled = true;
  348. const resourceURLs = [...this.requests.keys()];
  349. resourceURLs.forEach(requestKey => {
  350. const resourceRequests = this.requests.get(requestKey);
  351. resourceRequests.forEach(callbacks => callbacks.reject());
  352. this.requests.delete(requestKey);
  353. });
  354. }
  355. }
  356. // ---------
  357. // Processor
  358. // ---------
  359. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  360. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  361. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  362. const SCRIPT_TAG_FOUND = /<script/gi;
  363. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  364. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  365. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  366. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  367. class Processor {
  368. constructor(options, batchRequest) {
  369. this.options = options;
  370. this.stats = new Stats(options);
  371. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  372. this.batchRequest = batchRequest;
  373. this.stylesheets = new Map();
  374. this.styles = new Map();
  375. this.cssVariables = new Map();
  376. }
  377. initialize() {
  378. this.options.saveDate = new Date();
  379. this.options.saveUrl = this.options.url;
  380. if (this.options.enableMaff) {
  381. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  382. }
  383. this.maxResources = this.batchRequest.getMaxResources();
  384. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  385. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  386. }
  387. if (!this.options.removeImports && this.options.imports) {
  388. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  389. }
  390. this.stats.set("processed", "resources", this.maxResources);
  391. }
  392. async loadPage(pageContent, charset) {
  393. let content;
  394. if (!pageContent || this.options.saveRawPage) {
  395. content = await util.getContent(this.baseURI, {
  396. maxResourceSize: this.options.maxResourceSize,
  397. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  398. charset,
  399. frameId: this.options.windowId
  400. });
  401. pageContent = content.data;
  402. }
  403. this.doc = util.parseDocContent(pageContent, this.baseURI);
  404. if (this.options.saveRawPage) {
  405. let charset;
  406. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  407. const charsetDeclaration = element.content.split(";")[1];
  408. if (charsetDeclaration && !charset) {
  409. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  410. }
  411. });
  412. if (charset && charset != content.charset) {
  413. return this.loadPage(pageContent, charset);
  414. }
  415. }
  416. this.workStyleElement = this.doc.createElement("style");
  417. this.doc.body.appendChild(this.workStyleElement);
  418. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  419. }
  420. finalize() {
  421. if (this.workStyleElement.parentNode) {
  422. this.workStyleElement.remove();
  423. }
  424. }
  425. async getPageData() {
  426. util.postProcessDoc(this.doc);
  427. const url = util.parseURL(this.baseURI);
  428. if (this.options.insertSingleFileComment) {
  429. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  430. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  431. " \n url: " + this.options.saveUrl +
  432. " \n saved date: " + this.options.saveDate +
  433. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  434. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  435. }
  436. if (this.options.insertCanonicalLink) {
  437. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  438. if (!canonicalLink) {
  439. canonicalLink = this.doc.createElement("link");
  440. canonicalLink.setAttribute("rel", "canonical");
  441. this.doc.head.appendChild(canonicalLink);
  442. }
  443. if (canonicalLink && !canonicalLink.href) {
  444. canonicalLink.href = this.options.saveUrl;
  445. }
  446. }
  447. let size;
  448. if (this.options.displayStats) {
  449. size = util.getContentSize(this.doc.documentElement.outerHTML);
  450. }
  451. const content = util.serialize(this.doc, this.options.compressHTML);
  452. if (this.options.displayStats) {
  453. const contentSize = util.getContentSize(content);
  454. this.stats.set("processed", "HTML bytes", contentSize);
  455. this.stats.add("discarded", "HTML bytes", size - contentSize);
  456. }
  457. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  458. const replacementCharacter = this.options.filenameReplacementCharacter;
  459. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  460. if (!this.options.backgroundSave) {
  461. filename = filename.replace(/\//g, replacementCharacter);
  462. }
  463. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  464. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  465. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  466. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  467. filename = filename + "…" + extension;
  468. }
  469. if (!filename) {
  470. filename = "Unnamed page";
  471. }
  472. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  473. const pageData = {
  474. stats: this.stats.data,
  475. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  476. filename,
  477. content
  478. };
  479. if (this.options.addProof) {
  480. pageData.hash = await util.digest("SHA-256", content);
  481. }
  482. return pageData;
  483. }
  484. preProcessPage() {
  485. if (this.options.win) {
  486. 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));
  487. }
  488. if (this.options.images && !this.options.saveRawPage) {
  489. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  490. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  491. if (attributeValue) {
  492. const imageData = this.options.images[Number(attributeValue)];
  493. if (imageData) {
  494. if (this.options.removeHiddenElements && (
  495. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  496. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  497. )) {
  498. imgElement.setAttribute("src", EMPTY_IMAGE);
  499. } else {
  500. if (imageData.currentSrc) {
  501. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  502. imgElement.setAttribute("src", imageData.currentSrc);
  503. }
  504. if (this.options.loadDeferredImages) {
  505. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  506. imageData.src = imgElement.dataset.src;
  507. imgElement.setAttribute("src", imgElement.dataset.src);
  508. imgElement.removeAttribute("data-src");
  509. }
  510. }
  511. }
  512. }
  513. }
  514. });
  515. if (this.options.loadDeferredImages) {
  516. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  517. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  518. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  519. imgElement.removeAttribute("data-srcset");
  520. }
  521. });
  522. }
  523. }
  524. }
  525. replaceStyleContents() {
  526. if (this.options.stylesheets) {
  527. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  528. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  529. if (attributeValue) {
  530. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  531. if (stylesheetContent) {
  532. styleElement.textContent = stylesheetContent;
  533. }
  534. }
  535. });
  536. }
  537. }
  538. removeUnselectedElements() {
  539. removeUnmarkedElements(this.doc.body);
  540. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  541. function removeUnmarkedElements(element) {
  542. let selectedElementFound = false;
  543. Array.from(element.childNodes).forEach(node => {
  544. if (node.nodeType == 1) {
  545. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  546. selectedElementFound = selectedElementFound || isSelectedElement;
  547. if (isSelectedElement) {
  548. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  549. removeUnmarkedElements(node);
  550. } else if (selectedElementFound) {
  551. removeNode(node);
  552. } else {
  553. hideNode(node);
  554. }
  555. }
  556. });
  557. }
  558. function removeNode(node) {
  559. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  560. node.remove();
  561. } else {
  562. hideNode(node);
  563. }
  564. }
  565. function hideNode(node) {
  566. if (canHideNode(node)) {
  567. node.style.setProperty("display", "none", "important");
  568. Array.from(node.childNodes).forEach(removeNode);
  569. }
  570. }
  571. function canHideNode(node) {
  572. if (node.nodeType == 1) {
  573. const tagName = node.tagName && node.tagName.toLowerCase();
  574. return (tagName != "svg" && tagName != "style" && tagName != "link");
  575. }
  576. }
  577. }
  578. insertVideoPosters() {
  579. if (this.options.posters) {
  580. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  581. let videoElement;
  582. if (element.tagName == "VIDEO") {
  583. videoElement = element;
  584. } else {
  585. videoElement = element.parentElement;
  586. }
  587. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  588. if (attributeValue) {
  589. const posterURL = this.options.posters[Number(attributeValue)];
  590. if (!videoElement.poster && posterURL) {
  591. videoElement.setAttribute("poster", posterURL);
  592. }
  593. }
  594. });
  595. }
  596. }
  597. removeFrames() {
  598. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  599. this.stats.set("discarded", "frames", frameElements.length);
  600. this.stats.set("processed", "frames", frameElements.length);
  601. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  602. }
  603. removeScripts() {
  604. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  605. this.doc.querySelectorAll("[href]").forEach(element => {
  606. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  607. element.setAttribute("href", "");
  608. }
  609. });
  610. this.doc.querySelectorAll("[src]").forEach(element => {
  611. if (element.src && element.src.match(/^\s*javascript:/)) {
  612. element.removeAttribute("src");
  613. }
  614. });
  615. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  616. this.stats.set("discarded", "scripts", scriptElements.length);
  617. this.stats.set("processed", "scripts", scriptElements.length);
  618. scriptElements.forEach(element => element.remove());
  619. }
  620. removeVideoSources() {
  621. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  622. this.stats.set("discarded", "video sources", videoSourceElements.length);
  623. this.stats.set("processed", "video sources", videoSourceElements.length);
  624. videoSourceElements.forEach(element => {
  625. if (element.tagName == "SOURCE") {
  626. element.remove();
  627. } else {
  628. videoSourceElements.forEach(element => element.removeAttribute("src"));
  629. }
  630. });
  631. }
  632. removeAudioSources() {
  633. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  634. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  635. this.stats.set("processed", "audio sources", audioSourceElements.length);
  636. audioSourceElements.forEach(element => {
  637. if (element.tagName == "SOURCE") {
  638. element.remove();
  639. } else {
  640. audioSourceElements.forEach(element => element.removeAttribute("src"));
  641. }
  642. });
  643. }
  644. removeDiscardedResources() {
  645. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  646. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  647. 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\"])");
  648. this.stats.set("discarded", "objects", objectElements.length);
  649. this.stats.set("processed", "objects", objectElements.length);
  650. objectElements.forEach(element => element.remove());
  651. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  652. replacedAttributeValue.forEach(element => {
  653. let regExp;
  654. if (this.options.removeScripts) {
  655. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  656. } else {
  657. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  658. }
  659. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  660. if (relValue.length) {
  661. element.setAttribute("rel", relValue);
  662. } else {
  663. element.remove();
  664. }
  665. });
  666. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  667. if (this.options.compressHTML) {
  668. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  669. }
  670. if (!this.options.saveFavicon) {
  671. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  672. }
  673. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  674. }
  675. resetCharsetMeta() {
  676. let charset;
  677. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  678. const charsetDeclaration = element.content.split(";")[1];
  679. if (charsetDeclaration && !charset) {
  680. charset = charsetDeclaration.split("=")[1];
  681. if (charset) {
  682. this.charset = charset.trim().toLowerCase();
  683. }
  684. }
  685. element.remove();
  686. });
  687. const metaElement = this.doc.createElement("meta");
  688. metaElement.setAttribute("charset", "utf-8");
  689. if (this.doc.head.firstChild) {
  690. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  691. } else {
  692. this.doc.head.appendChild(metaElement);
  693. }
  694. }
  695. setInputValues() {
  696. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  697. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  698. input.setAttribute("value", value || "");
  699. });
  700. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  701. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  702. if (value == "true") {
  703. input.setAttribute("checked", "");
  704. }
  705. });
  706. this.doc.querySelectorAll("textarea").forEach(textarea => {
  707. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  708. textarea.textContent = value || "";
  709. });
  710. this.doc.querySelectorAll("select").forEach(select => {
  711. select.querySelectorAll("option").forEach(option => {
  712. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  713. if (selected) {
  714. option.setAttribute("selected", "");
  715. }
  716. });
  717. });
  718. }
  719. saveFavicon() {
  720. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  721. if (!faviconElement) {
  722. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  723. }
  724. if (!faviconElement) {
  725. faviconElement = this.doc.createElement("link");
  726. faviconElement.setAttribute("type", "image/x-icon");
  727. faviconElement.setAttribute("rel", "shortcut icon");
  728. faviconElement.setAttribute("href", "/favicon.ico");
  729. }
  730. this.doc.head.appendChild(faviconElement);
  731. }
  732. replaceCanvasElements() {
  733. if (this.options.canvases) {
  734. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  735. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  736. if (attributeValue) {
  737. const canvasData = this.options.canvases[Number(attributeValue)];
  738. if (canvasData) {
  739. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  740. this.stats.add("processed", "canvas", 1);
  741. }
  742. }
  743. });
  744. }
  745. }
  746. insertFonts() {
  747. if (this.options.fonts && this.options.fonts.length) {
  748. let stylesheetContent = "";
  749. this.options.fonts.forEach(fontData => {
  750. if (fontData["font-family"] && fontData.src) {
  751. stylesheetContent += "@font-face{";
  752. let stylesContent = "";
  753. Object.keys(fontData).forEach(fontStyle => {
  754. if (stylesContent) {
  755. stylesContent += ";";
  756. }
  757. stylesContent += fontStyle + ":" + fontData[fontStyle];
  758. });
  759. stylesheetContent += stylesContent + "}";
  760. }
  761. });
  762. if (stylesheetContent) {
  763. const styleElement = this.doc.createElement("style");
  764. styleElement.textContent = stylesheetContent;
  765. const existingStyleElement = this.doc.querySelector("style");
  766. if (existingStyleElement) {
  767. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  768. } else {
  769. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  770. }
  771. }
  772. }
  773. }
  774. removeHiddenElements() {
  775. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  776. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  777. this.stats.set("discarded", "hidden elements", removedElements.length);
  778. this.stats.set("processed", "hidden elements", removedElements.length);
  779. if (hiddenElements.length) {
  780. const styleElement = this.doc.createElement("style");
  781. styleElement.textContent = ".sf-hidden{display:none!important;}";
  782. this.doc.head.appendChild(styleElement);
  783. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  784. }
  785. removedElements.forEach(element => element.remove());
  786. }
  787. resolveHrefs() {
  788. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  789. const href = element.getAttribute("href").trim();
  790. if (!Util.testIgnoredPath(href)) {
  791. let resolvedURL;
  792. try {
  793. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  794. } catch (error) {
  795. // ignored
  796. }
  797. if (resolvedURL) {
  798. const url = Util.normalizeURL(this.options.url);
  799. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  800. resolvedURL = resolvedURL.substring(url.length);
  801. }
  802. try {
  803. element.setAttribute("href", resolvedURL);
  804. } catch (error) {
  805. // ignored
  806. }
  807. }
  808. }
  809. });
  810. }
  811. resolveStyleAttributeURLs() {
  812. this.doc.querySelectorAll("[style]").forEach(element => {
  813. let styleContent = element.getAttribute("style");
  814. if (this.options.compressCSS) {
  815. styleContent = util.compressCSS(styleContent);
  816. }
  817. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  818. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  819. this.styles.set(element, declarationList);
  820. });
  821. }
  822. async resolveStylesheetURLs() {
  823. const options = {
  824. maxResourceSize: this.options.maxResourceSize,
  825. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  826. url: this.options.url,
  827. charset: this.charset,
  828. compressCSS: this.options.compressCSS,
  829. updatedResources: this.options.updatedResources,
  830. rootDocument: this.options.rootDocument,
  831. frameId: this.options.windowId
  832. };
  833. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  834. .map(async element => {
  835. let mediaText;
  836. if (element.media) {
  837. mediaText = element.media.toLowerCase();
  838. }
  839. const stylesheetInfo = { mediaText };
  840. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  841. stylesheetInfo.scoped = true;
  842. }
  843. if (element.tagName.toLowerCase() == "link") {
  844. if (element.charset) {
  845. options.charset = element.charset;
  846. }
  847. }
  848. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  849. }));
  850. if (options.rootDocument) {
  851. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  852. await Promise.all(newResources.map(async resource => {
  853. resource.retrieved = true;
  854. const stylesheetInfo = {};
  855. const element = this.doc.createElement("style");
  856. this.doc.body.appendChild(element);
  857. element.textContent = resource.content;
  858. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  859. }));
  860. }
  861. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  862. stylesheets.set(element, stylesheetInfo);
  863. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  864. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  865. if (match && match[1] && match[1] != options.charset) {
  866. options.charset = match[1];
  867. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  868. }
  869. let stylesheet;
  870. try {
  871. stylesheet = cssTree.parse(Util.removeCssComments(stylesheetContent));
  872. } catch (error) {
  873. // ignored
  874. }
  875. if (stylesheet && stylesheet.children) {
  876. if (options.compressCSS) {
  877. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  878. }
  879. stylesheetInfo.stylesheet = stylesheet;
  880. } else {
  881. stylesheets.delete(element);
  882. }
  883. }
  884. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  885. let content;
  886. if (element.tagName.toLowerCase() == "link") {
  887. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  888. } else {
  889. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  890. }
  891. return content || "";
  892. }
  893. }
  894. async resolveFrameURLs() {
  895. if (!this.options.saveRawPage && this.options.frames) {
  896. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  897. await Promise.all(frameElements.map(async frameElement => {
  898. if (frameElement.tagName == "OBJECT") {
  899. frameElement.setAttribute("data", "data:text/html,");
  900. } else {
  901. frameElement.removeAttribute("src");
  902. frameElement.removeAttribute("srcdoc");
  903. }
  904. Array.from(frameElement.childNodes).forEach(node => node.remove());
  905. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  906. if (frameWindowId) {
  907. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  908. if (frameData) {
  909. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  910. }
  911. }
  912. }));
  913. }
  914. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  915. options.insertSingleFileComment = false;
  916. options.insertCanonicalLink = false;
  917. options.saveFavicon = false;
  918. options.url = frameData.baseURI;
  919. options.windowId = frameWindowId;
  920. if (frameData.content) {
  921. options.content = frameData.content;
  922. options.canvases = frameData.canvases;
  923. options.fonts = frameData.fonts;
  924. options.stylesheets = frameData.stylesheets;
  925. options.images = frameData.images;
  926. options.posters = frameData.posters;
  927. options.usedFonts = frameData.usedFonts;
  928. options.shadowRoots = frameData.shadowRoots;
  929. options.imports = frameData.imports;
  930. frameData.runner = new Runner(options);
  931. frameData.frameElement = frameElement;
  932. await frameData.runner.loadPage();
  933. await frameData.runner.initialize();
  934. frameData.maxResources = batchRequest.getMaxResources();
  935. }
  936. }
  937. }
  938. insertShadowRootContents() {
  939. const doc = this.doc;
  940. const options = this.options;
  941. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  942. processElement(this.doc);
  943. if (this.options.removeScripts) {
  944. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  945. }
  946. const scriptElement = doc.createElement("script");
  947. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  948. 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)}})}})()`;
  949. doc.body.appendChild(scriptElement);
  950. }
  951. function processElement(element) {
  952. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  953. shadowRootElements.forEach(element => {
  954. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  955. if (attributeValue) {
  956. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  957. if (shadowRootData) {
  958. const templateElement = doc.createElement("template");
  959. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  960. if (shadowRootData.delegatesFocus) {
  961. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  962. }
  963. if (shadowRootData.adoptedStyleSheets) {
  964. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  965. const styleElement = doc.createElement("style");
  966. styleElement.textContent = stylesheetContent;
  967. templateElement.appendChild(styleElement);
  968. });
  969. }
  970. const shadowDoc = util.parseDocContent(shadowRootData.content);
  971. if (shadowDoc.head) {
  972. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  973. if (metaCharset) {
  974. metaCharset.remove();
  975. }
  976. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  977. }
  978. if (shadowDoc.body) {
  979. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  980. }
  981. processElement(templateElement);
  982. if (element.firstChild) {
  983. element.insertBefore(templateElement, element.firstChild);
  984. } else {
  985. element.appendChild(templateElement);
  986. }
  987. }
  988. }
  989. });
  990. }
  991. }
  992. async resolveHtmlImportURLs() {
  993. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  994. await Promise.all(linkElements.map(async linkElement => {
  995. const resourceURL = linkElement.href;
  996. linkElement.removeAttribute("href");
  997. const options = Object.create(this.options);
  998. options.insertSingleFileComment = false;
  999. options.insertCanonicalLink = false;
  1000. options.saveFavicon = false;
  1001. options.removeUnusedStyles = false;
  1002. options.removeAlternativeMedias = false;
  1003. options.removeUnusedFonts = false;
  1004. options.removeHiddenElements = false;
  1005. options.url = resourceURL;
  1006. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1007. if (attributeValue) {
  1008. const importData = options.imports[Number(attributeValue)];
  1009. if (importData) {
  1010. options.content = importData.content;
  1011. importData.runner = new Runner(options);
  1012. await importData.runner.loadPage();
  1013. await importData.runner.initialize();
  1014. if (!options.removeImports) {
  1015. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1016. }
  1017. importData.runner.getStyleSheets().forEach(stylesheet => {
  1018. const importedStyleElement = this.doc.createElement("style");
  1019. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1020. this.stylesheets.set(importedStyleElement, stylesheet);
  1021. });
  1022. }
  1023. }
  1024. if (options.removeImports) {
  1025. linkElement.remove();
  1026. this.stats.add("discarded", "HTML imports", 1);
  1027. }
  1028. }));
  1029. }
  1030. removeUnusedStyles() {
  1031. if (!this.mediaAllInfo) {
  1032. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1033. }
  1034. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1035. this.stats.set("processed", "CSS rules", stats.processed);
  1036. this.stats.set("discarded", "CSS rules", stats.discarded);
  1037. }
  1038. removeUnusedFonts() {
  1039. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1040. }
  1041. removeAlternativeMedias() {
  1042. const stats = util.minifyMedias(this.stylesheets);
  1043. this.stats.set("processed", "medias", stats.processed);
  1044. this.stats.set("discarded", "medias", stats.discarded);
  1045. }
  1046. async processStylesheets() {
  1047. this.options.fontURLs = new Map();
  1048. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1049. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1050. ));
  1051. }
  1052. async processStyleAttributes() {
  1053. return Promise.all([...this.styles].map(([, declarationList]) =>
  1054. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1055. ));
  1056. }
  1057. async processPageResources() {
  1058. const processAttributeArgs = [
  1059. ["link[href][rel*=\"icon\"]", "href", false, true],
  1060. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1061. ["img[src], input[src][type=image]", "src", true],
  1062. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1063. ["video[poster]", "poster"],
  1064. ["*[background]", "background"],
  1065. ["image", "xlink:href"],
  1066. ["image", "href"]
  1067. ];
  1068. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1069. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1070. );
  1071. resourcePromises = resourcePromises.concat([
  1072. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1073. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1074. ]);
  1075. if (!this.options.removeAudioSrc) {
  1076. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1077. }
  1078. if (!this.options.removeVideoSrc) {
  1079. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1080. }
  1081. await Promise.all(resourcePromises);
  1082. if (this.options.saveFavicon) {
  1083. ProcessorHelper.processShortcutIcons(this.doc);
  1084. }
  1085. }
  1086. async processScripts() {
  1087. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1088. let resourceURL;
  1089. let scriptSrc;
  1090. if (element.tagName == "SCRIPT") {
  1091. scriptSrc = element.getAttribute("src");
  1092. element.removeAttribute("src");
  1093. } else {
  1094. scriptSrc = element.getAttribute("href");
  1095. element.removeAttribute("href");
  1096. }
  1097. element.removeAttribute("integrity");
  1098. element.textContent = "";
  1099. try {
  1100. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1101. } catch (error) {
  1102. // ignored
  1103. }
  1104. if (Util.testValidURL(resourceURL)) {
  1105. this.stats.add("processed", "scripts", 1);
  1106. const content = await util.getContent(resourceURL, {
  1107. asBinary: true,
  1108. maxResourceSize: this.options.maxResourceSize,
  1109. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1110. frameId: this.options.windowId
  1111. });
  1112. content.data = Util.getUpdatedResourceContent(resourceURL, content, this.options);
  1113. if (element.tagName == "SCRIPT") {
  1114. element.setAttribute("src", content.data);
  1115. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1116. element.setAttribute("async", "");
  1117. }
  1118. } else {
  1119. const scriptElement = this.doc.createElement("script");
  1120. scriptElement.setAttribute("src", content.data);
  1121. scriptElement.setAttribute("async", "");
  1122. element.parentElement.replaceChild(scriptElement, element);
  1123. }
  1124. }
  1125. }));
  1126. }
  1127. removeAlternativeImages() {
  1128. util.removeAlternativeImages(this.doc);
  1129. }
  1130. async removeAlternativeFonts() {
  1131. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs);
  1132. }
  1133. async processFrames() {
  1134. if (this.options.frames) {
  1135. const frameElements = Array.from(this.doc.querySelectorAll("iframe:not([" + util.HIDDEN_FRAME_ATTRIBUTE_NAME + "]), frame, object[type=\"text/html\"][data]"));
  1136. await Promise.all(frameElements.map(async frameElement => {
  1137. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1138. if (frameWindowId) {
  1139. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1140. if (frameData) {
  1141. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1142. if (frameData.runner) {
  1143. this.stats.add("processed", "frames", 1);
  1144. await frameData.runner.run();
  1145. const pageData = await frameData.runner.getPageData();
  1146. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1147. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1148. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1149. sandbox += " allow-scripts allow-same-origin";
  1150. }
  1151. frameElement.setAttribute("sandbox", sandbox);
  1152. if (frameElement.tagName == "OBJECT") {
  1153. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1154. } else {
  1155. if (frameElement.tagName == "FRAME") {
  1156. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1157. } else {
  1158. frameElement.setAttribute("srcdoc", pageData.content);
  1159. frameElement.removeAttribute("src");
  1160. }
  1161. }
  1162. this.stats.addAll(pageData);
  1163. } else {
  1164. this.stats.add("discarded", "frames", 1);
  1165. }
  1166. }
  1167. }
  1168. }));
  1169. }
  1170. }
  1171. async processHtmlImports() {
  1172. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1173. await Promise.all(linkElements.map(async linkElement => {
  1174. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1175. if (attributeValue) {
  1176. const importData = this.options.imports[Number(attributeValue)];
  1177. if (importData.runner) {
  1178. this.stats.add("processed", "HTML imports", 1);
  1179. await importData.runner.run();
  1180. const pageData = await importData.runner.getPageData();
  1181. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1182. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1183. this.stats.addAll(pageData);
  1184. } else {
  1185. this.stats.add("discarded", "HTML imports", 1);
  1186. }
  1187. }
  1188. }));
  1189. }
  1190. replaceStylesheets() {
  1191. this.doc.querySelectorAll("style").forEach(styleElement => {
  1192. const stylesheetInfo = this.stylesheets.get(styleElement);
  1193. if (stylesheetInfo) {
  1194. this.stylesheets.delete(styleElement);
  1195. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1196. styleElement.textContent = stylesheetContent;
  1197. if (stylesheetInfo.mediaText) {
  1198. styleElement.media = stylesheetInfo.mediaText;
  1199. }
  1200. } else {
  1201. styleElement.remove();
  1202. }
  1203. });
  1204. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1205. const stylesheetInfo = this.stylesheets.get(linkElement);
  1206. if (stylesheetInfo) {
  1207. this.stylesheets.delete(linkElement);
  1208. const styleElement = this.doc.createElement("style");
  1209. if (stylesheetInfo.mediaText) {
  1210. styleElement.media = stylesheetInfo.mediaText;
  1211. }
  1212. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1213. styleElement.textContent = stylesheetContent;
  1214. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1215. } else {
  1216. linkElement.remove();
  1217. }
  1218. });
  1219. }
  1220. replaceStyleAttributes() {
  1221. this.doc.querySelectorAll("[style]").forEach(element => {
  1222. const declarations = this.styles.get(element);
  1223. if (declarations) {
  1224. this.styles.delete(element);
  1225. let styleContent = cssTree.generate(declarations);
  1226. element.setAttribute("style", styleContent);
  1227. } else {
  1228. element.setAttribute("style", "");
  1229. }
  1230. });
  1231. }
  1232. insertVariables() {
  1233. if (this.cssVariables.size) {
  1234. const styleElement = this.doc.createElement("style");
  1235. const firstStyleElement = this.doc.head.querySelector("style");
  1236. if (firstStyleElement) {
  1237. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1238. } else {
  1239. this.doc.head.appendChild(styleElement);
  1240. }
  1241. let stylesheetContent = "";
  1242. this.cssVariables.forEach((content, indexResource) => {
  1243. this.cssVariables.delete(indexResource);
  1244. if (stylesheetContent) {
  1245. stylesheetContent += ";";
  1246. }
  1247. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1248. });
  1249. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1250. }
  1251. }
  1252. compressHTML() {
  1253. let size;
  1254. if (this.options.displayStats) {
  1255. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1256. }
  1257. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1258. if (this.options.displayStats) {
  1259. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1260. }
  1261. }
  1262. cleanupPage() {
  1263. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1264. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1265. if (metaCharset) {
  1266. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1267. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1268. this.doc.head.querySelector("meta[charset]").remove();
  1269. }
  1270. }
  1271. }
  1272. async insertMAFFMetaData() {
  1273. const maffMetaData = await this.maffMetaDataPromise;
  1274. if (maffMetaData && maffMetaData.content) {
  1275. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1276. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1277. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1278. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1279. if (originalURLElement) {
  1280. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1281. }
  1282. if (archiveTimeElement) {
  1283. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1284. if (value) {
  1285. const date = new Date(value);
  1286. if (!isNaN(date.getTime())) {
  1287. this.options.saveDate = new Date(value);
  1288. }
  1289. }
  1290. }
  1291. }
  1292. }
  1293. async setDocInfo() {
  1294. const titleElement = this.doc.querySelector("title");
  1295. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1296. const authorElement = this.doc.querySelector("meta[name=author]");
  1297. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1298. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1299. const headingElement = this.doc.querySelector("h1");
  1300. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1301. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1302. this.options.info = {
  1303. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1304. lang: this.doc.documentElement.lang,
  1305. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1306. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1307. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1308. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1309. };
  1310. }
  1311. }
  1312. // ---------------
  1313. // ProcessorHelper
  1314. // ---------------
  1315. const DATA_URI_PREFIX = "data:";
  1316. const ABOUT_BLANK_URI = "about:blank";
  1317. const EMPTY_DATA_URI = "data:null;base64,";
  1318. const REGEXP_URL_HASH = /(#.+?)$/;
  1319. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1320. const VAR_MAX_SIZE = 1024 * 1024 * 2;
  1321. class ProcessorHelper {
  1322. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1323. const date = options.saveDate;
  1324. const url = util.parseURL(options.saveUrl);
  1325. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1326. template = await Util.evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1327. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1328. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1329. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1330. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1331. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1332. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1333. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1334. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1335. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1336. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1337. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1338. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1339. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1340. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1341. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1342. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1343. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1344. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1345. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1346. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1347. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1348. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1349. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1350. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1351. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1352. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1353. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1354. template = await Util.evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await Util.evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1357. template = await Util.evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1358. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1359. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1360. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1361. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1362. for (const [name, value] of params) {
  1363. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1364. }
  1365. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1366. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1367. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1368. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decode(Util.getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1370. if (content) {
  1371. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1374. }
  1375. return template.trim();
  1376. function decode(value) {
  1377. try {
  1378. return decodeURI(value);
  1379. } catch (error) {
  1380. return value;
  1381. }
  1382. }
  1383. }
  1384. static setBackgroundImage(element, url, style) {
  1385. element.style.setProperty("background-blend-mode", "normal", "important");
  1386. element.style.setProperty("background-clip", "content-box", "important");
  1387. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1388. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1389. element.style.setProperty("background-image", url, "important");
  1390. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1391. element.style.setProperty("background-origin", "content-box", "important");
  1392. element.style.setProperty("background-repeat", "no-repeat", "important");
  1393. }
  1394. static processShortcutIcons(doc) {
  1395. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1396. if (!shortcutIcon) {
  1397. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1398. if (shortcutIcon) {
  1399. shortcutIcon.rel = "icon";
  1400. }
  1401. }
  1402. if (shortcutIcon) {
  1403. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1404. if (linkElement != shortcutIcon) {
  1405. linkElement.remove();
  1406. }
  1407. });
  1408. }
  1409. }
  1410. static removeSingleLineCssComments(stylesheet) {
  1411. const removedRules = [];
  1412. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1413. const ruleData = cssRule.data;
  1414. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1415. removedRules.push(cssRule);
  1416. }
  1417. }
  1418. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1419. }
  1420. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1421. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1422. const imports = Util.getImportFunctions(stylesheetContent);
  1423. await Promise.all(imports.map(async cssImport => {
  1424. const match = Util.matchImport(cssImport);
  1425. if (match) {
  1426. const regExpCssImport = Util.getRegExp(cssImport);
  1427. let resourceURL = Util.normalizeURL(match.resourceURL);
  1428. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1429. try {
  1430. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1431. } catch (error) {
  1432. // ignored
  1433. }
  1434. if (Util.testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1435. const content = await util.getContent(resourceURL, {
  1436. maxResourceSize: options.maxResourceSize,
  1437. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1438. validateTextContentType: true,
  1439. frameId: options.windowId
  1440. });
  1441. resourceURL = content.resourceURL;
  1442. content.data = Util.getUpdatedResourceContent(resourceURL, content, options);
  1443. let importedStylesheetContent = Util.removeCssComments(content.data);
  1444. if (options.compressCSS) {
  1445. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1446. }
  1447. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1448. if (stylesheetContent.includes(cssImport)) {
  1449. const ancestorStyleSheets = new Set(importedStyleSheets);
  1450. ancestorStyleSheets.add(resourceURL);
  1451. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1452. workStylesheet.textContent = importedStylesheetContent;
  1453. if (workStylesheet.sheet.cssRules.length) {
  1454. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1455. } else {
  1456. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1457. }
  1458. }
  1459. } else {
  1460. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1461. }
  1462. } else {
  1463. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1464. }
  1465. }
  1466. }));
  1467. return stylesheetContent;
  1468. }
  1469. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1470. const urlFunctions = Util.getUrlFunctions(stylesheetContent, true);
  1471. urlFunctions.map(urlFunction => {
  1472. const originalResourceURL = Util.matchURL(urlFunction);
  1473. const resourceURL = Util.normalizeURL(originalResourceURL);
  1474. if (!Util.testIgnoredPath(resourceURL)) {
  1475. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1476. let resolvedURL;
  1477. if (!originalResourceURL.startsWith("#")) {
  1478. try {
  1479. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1480. } catch (error) {
  1481. // ignored
  1482. }
  1483. }
  1484. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1485. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1486. }
  1487. } else {
  1488. let newUrlFunction;
  1489. if (originalResourceURL) {
  1490. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1491. } else {
  1492. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1493. }
  1494. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1495. }
  1496. }
  1497. });
  1498. return stylesheetContent;
  1499. }
  1500. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1501. resourceURL = Util.normalizeURL(resourceURL);
  1502. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1503. const content = await util.getContent(resourceURL, {
  1504. maxResourceSize: options.maxResourceSize,
  1505. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1506. charset: options.charset,
  1507. frameId: options.windowId
  1508. });
  1509. resourceURL = content.resourceURL;
  1510. content.data = Util.getUpdatedResourceContent(content.resourceURL, content, options);
  1511. let stylesheetContent = Util.removeCssComments(content.data);
  1512. if (options.compressCSS) {
  1513. stylesheetContent = util.compressCSS(stylesheetContent);
  1514. }
  1515. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1516. return stylesheetContent;
  1517. }
  1518. }
  1519. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1520. const promises = [];
  1521. const removedRules = [];
  1522. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1523. const ruleData = cssRule.data;
  1524. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1525. removedRules.push(cssRule);
  1526. } else if (ruleData.block && ruleData.block.children) {
  1527. if (ruleData.type == "Rule") {
  1528. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1529. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1530. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1531. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1532. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1533. }
  1534. }
  1535. }
  1536. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1537. await Promise.all(promises);
  1538. async function processFontFaceRule(ruleData, fontURLs) {
  1539. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1540. if (declaration.type == "Declaration" && declaration.value.children) {
  1541. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value), true);
  1542. await Promise.all(urlFunctions.map(async urlFunction => {
  1543. const originalResourceURL = Util.matchURL(urlFunction);
  1544. const resourceURL = Util.normalizeURL(originalResourceURL);
  1545. if (!Util.testIgnoredPath(resourceURL)) {
  1546. if (Util.testValidURL(resourceURL)) {
  1547. let { content } = await batchRequest.addURL(resourceURL, true);
  1548. fontURLs.set(declaration, resourceURL);
  1549. replaceURLs(declaration, originalResourceURL, content);
  1550. }
  1551. }
  1552. }));
  1553. }
  1554. }));
  1555. function replaceURLs(declaration, oldURL, newURL) {
  1556. declaration.value.children.forEach(token => {
  1557. if (token.type == "Url" && util.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1558. token.value.value = newURL;
  1559. }
  1560. });
  1561. }
  1562. }
  1563. }
  1564. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1565. await Promise.all(declarations.map(async declaration => {
  1566. if (declaration.type == "Declaration" && declaration.value.children) {
  1567. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1568. await Promise.all(urlFunctions.map(async urlFunction => {
  1569. const originalResourceURL = Util.matchURL(urlFunction);
  1570. const resourceURL = Util.normalizeURL(originalResourceURL);
  1571. if (!Util.testIgnoredPath(resourceURL)) {
  1572. if (Util.testValidURL(resourceURL)) {
  1573. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1574. let variableDefined;
  1575. const tokens = [];
  1576. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1577. if (!originalResourceURL.startsWith("#")) {
  1578. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < VAR_MAX_SIZE) {
  1579. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1580. tokens.push({ parent, token, value });
  1581. variableDefined = true;
  1582. } else {
  1583. token.data.value.value = content;
  1584. }
  1585. }
  1586. });
  1587. if (variableDefined) {
  1588. cssVariables.set(indexResource, content);
  1589. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1590. }
  1591. }
  1592. }
  1593. }));
  1594. }
  1595. }));
  1596. function findURLToken(url, children, callback, depth = 0) {
  1597. for (let token = children.head; token; token = token.next) {
  1598. if (token.data.children) {
  1599. findURLToken(url, token.data.children, callback, depth + 1);
  1600. }
  1601. if (token.data.type == "Url" && util.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1602. callback(token, children, depth == 0);
  1603. }
  1604. }
  1605. }
  1606. }
  1607. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1608. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1609. let resourceURL = resourceElement.getAttribute(attributeName);
  1610. if (resourceURL != null) {
  1611. resourceURL = Util.normalizeURL(resourceURL);
  1612. if (!Util.testIgnoredPath(resourceURL)) {
  1613. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1614. if (Util.testValidPath(resourceURL)) {
  1615. try {
  1616. resourceURL = util.resolveURL(resourceURL, baseURI);
  1617. } catch (error) {
  1618. // ignored
  1619. }
  1620. if (Util.testValidURL(resourceURL)) {
  1621. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1622. if (resourceElement.dataset.singleFileOriginURL) {
  1623. let originURL = resourceElement.dataset.singleFileOriginURL;
  1624. delete resourceElement.dataset.singleFileOriginURL;
  1625. if (content == EMPTY_DATA_URI) {
  1626. try {
  1627. originURL = util.resolveURL(originURL, baseURI);
  1628. } catch (error) {
  1629. // ignored
  1630. }
  1631. try {
  1632. resourceURL = originURL;
  1633. content = (await util.getContent(resourceURL, {
  1634. asBinary: true,
  1635. maxResourceSize: options.maxResourceSize,
  1636. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1637. frameId: options.windowId
  1638. })).data;
  1639. } catch (error) {
  1640. // ignored
  1641. }
  1642. }
  1643. }
  1644. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1645. resourceElement.remove();
  1646. } else {
  1647. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1648. if (!forbiddenPrefixFound) {
  1649. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1650. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < VAR_MAX_SIZE) {
  1651. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1652. cssVariables.set(indexResource, content);
  1653. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1654. styles.set(resourceElement, declarationList);
  1655. } else {
  1656. resourceElement.setAttribute(attributeName, content);
  1657. }
  1658. } else {
  1659. resourceElement.setAttribute(attributeName, content);
  1660. }
  1661. }
  1662. }
  1663. }
  1664. }
  1665. }
  1666. }
  1667. }));
  1668. }
  1669. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1670. const attributeName = "xlink:href";
  1671. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1672. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1673. let resourceURL = Util.normalizeURL(originalResourceURL);
  1674. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1675. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1676. try {
  1677. resourceURL = util.resolveURL(resourceURL, baseURI);
  1678. } catch (error) {
  1679. // ignored
  1680. }
  1681. if (Util.testValidURL(resourceURL)) {
  1682. const { content } = await batchRequest.addURL(resourceURL);
  1683. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1684. if (hashMatch && hashMatch[0]) {
  1685. let symbolElement;
  1686. try {
  1687. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1688. } catch (error) {
  1689. // ignored
  1690. }
  1691. if (symbolElement) {
  1692. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1693. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1694. }
  1695. } else {
  1696. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1697. }
  1698. }
  1699. } else if (resourceURL == options.url) {
  1700. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1701. }
  1702. }));
  1703. }
  1704. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1705. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1706. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1707. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1708. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1709. if (!Util.testIgnoredPath(resourceURL)) {
  1710. if (Util.testValidPath(resourceURL)) {
  1711. try {
  1712. resourceURL = util.resolveURL(resourceURL, baseURI);
  1713. } catch (error) {
  1714. // ignored
  1715. }
  1716. if (Util.testValidURL(resourceURL)) {
  1717. const { content } = await batchRequest.addURL(resourceURL, true);
  1718. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1719. if (forbiddenPrefixFound) {
  1720. return "";
  1721. }
  1722. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1723. } else {
  1724. return "";
  1725. }
  1726. } else {
  1727. return "";
  1728. }
  1729. } else {
  1730. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1731. }
  1732. }));
  1733. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1734. }));
  1735. }
  1736. static replaceImageSource(imgElement, variableName, options) {
  1737. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1738. if (attributeValue) {
  1739. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1740. if (imageData && imageData.replaceable) {
  1741. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1742. const backgroundStyle = {};
  1743. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1744. if (backgroundSize) {
  1745. backgroundStyle["background-size"] = imageData.objectFit;
  1746. }
  1747. if (imageData.objectPosition) {
  1748. backgroundStyle["background-position"] = imageData.objectPosition;
  1749. }
  1750. if (imageData.backgroundColor) {
  1751. backgroundStyle["background-color"] = imageData.backgroundColor;
  1752. }
  1753. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1754. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1755. return true;
  1756. }
  1757. }
  1758. }
  1759. }
  1760. // ----
  1761. // Util
  1762. // ----
  1763. const BLOB_URI_PREFIX = "blob:";
  1764. const HTTP_URI_PREFIX = /^https?:\/\//;
  1765. const FILE_URI_PREFIX = /^file:\/\//;
  1766. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1767. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1768. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1769. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1770. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1771. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1772. 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;
  1773. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1774. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1775. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1776. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1777. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1778. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1779. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1780. class Util {
  1781. static getUpdatedResourceContent(resourceURL, content, options) {
  1782. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1783. options.updatedResources[resourceURL].retrieved = true;
  1784. return options.updatedResources[resourceURL].content;
  1785. } else {
  1786. return content.data || "";
  1787. }
  1788. }
  1789. static normalizeURL(url) {
  1790. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1791. return url;
  1792. } else {
  1793. return url.split("#")[0];
  1794. }
  1795. }
  1796. static getCSSValue(value) {
  1797. let result = "";
  1798. try {
  1799. result = cssTree.generate(value);
  1800. } catch (error) {
  1801. // ignored
  1802. }
  1803. return result;
  1804. }
  1805. static getOnEventAttributeNames(doc) {
  1806. const element = doc.createElement("div");
  1807. const attributeNames = [];
  1808. for (const propertyName in element) {
  1809. if (propertyName.startsWith("on")) {
  1810. attributeNames.push(propertyName);
  1811. }
  1812. }
  1813. attributeNames.push("onunload");
  1814. return attributeNames;
  1815. }
  1816. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1817. let maxLength;
  1818. if (template) {
  1819. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1820. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1821. if (template.match(replaceRegExp)) {
  1822. const matchedLength = template.match(replaceRegExp)[0];
  1823. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1824. if (isNaN(maxLength) || maxLength <= 0) {
  1825. maxLength = null;
  1826. }
  1827. } else {
  1828. replaceRegExp = new RegExp(regExpVariable, "g");
  1829. }
  1830. if (template.match(replaceRegExp)) {
  1831. let value = await valueGetter();
  1832. if (!dontReplaceSlash) {
  1833. value = value.replace(/\/+/g, replacementCharacter);
  1834. }
  1835. if (maxLength) {
  1836. value = await util.truncateText(value, maxLength);
  1837. }
  1838. return template.replace(replaceRegExp, value);
  1839. }
  1840. }
  1841. return template;
  1842. }
  1843. static getLastSegment(url, replacementCharacter) {
  1844. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1845. if (!lastSegment) {
  1846. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1847. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1848. }
  1849. if (!lastSegment) {
  1850. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1851. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1852. }
  1853. if (!lastSegment) {
  1854. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1855. }
  1856. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1857. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1858. lastSegment = lastSegmentMatch[1];
  1859. }
  1860. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1861. return lastSegment;
  1862. }
  1863. static getRegExp(string) {
  1864. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1865. }
  1866. static getUrlFunctions(stylesheetContent, unique) {
  1867. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1868. if (unique) {
  1869. return [...new Set(result)];
  1870. } else {
  1871. return result;
  1872. }
  1873. }
  1874. static getImportFunctions(stylesheetContent) {
  1875. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1876. }
  1877. static findShortcutIcon(shortcutIcons) {
  1878. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1879. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1880. return shortcutIcons[0];
  1881. }
  1882. static matchURL(stylesheetContent) {
  1883. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1884. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1885. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1886. return match && match[1];
  1887. }
  1888. static testIgnoredPath(resourceURL) {
  1889. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1890. }
  1891. static testValidPath(resourceURL) {
  1892. return resourceURL && !resourceURL.match(EMPTY_URL);
  1893. }
  1894. static testValidURL(resourceURL) {
  1895. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1896. }
  1897. static matchImport(stylesheetContent) {
  1898. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1899. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1900. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1901. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1902. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1903. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1904. if (match) {
  1905. const [, resourceURL, media] = match;
  1906. return { resourceURL, media };
  1907. }
  1908. }
  1909. static removeCssComments(stylesheetContent) {
  1910. let start, end;
  1911. do {
  1912. start = stylesheetContent.indexOf("/*");
  1913. end = stylesheetContent.indexOf("*/", start + 2);
  1914. if (start != -1 && end != -1) {
  1915. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1916. }
  1917. } while (start != -1 && end != -1);
  1918. return stylesheetContent;
  1919. }
  1920. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1921. if (mediaQuery) {
  1922. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1923. } else {
  1924. return stylesheetContent;
  1925. }
  1926. }
  1927. }
  1928. function log(...args) {
  1929. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1930. }
  1931. // -----
  1932. // Stats
  1933. // -----
  1934. const STATS_DEFAULT_VALUES = {
  1935. discarded: {
  1936. "HTML bytes": 0,
  1937. "hidden elements": 0,
  1938. "HTML imports": 0,
  1939. scripts: 0,
  1940. objects: 0,
  1941. "audio sources": 0,
  1942. "video sources": 0,
  1943. frames: 0,
  1944. "CSS rules": 0,
  1945. canvas: 0,
  1946. stylesheets: 0,
  1947. resources: 0,
  1948. medias: 0
  1949. },
  1950. processed: {
  1951. "HTML bytes": 0,
  1952. "hidden elements": 0,
  1953. "HTML imports": 0,
  1954. scripts: 0,
  1955. objects: 0,
  1956. "audio sources": 0,
  1957. "video sources": 0,
  1958. frames: 0,
  1959. "CSS rules": 0,
  1960. canvas: 0,
  1961. stylesheets: 0,
  1962. resources: 0,
  1963. medias: 0
  1964. }
  1965. };
  1966. class Stats {
  1967. constructor(options) {
  1968. this.options = options;
  1969. if (options.displayStats) {
  1970. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1971. }
  1972. }
  1973. set(type, subType, value) {
  1974. if (this.options.displayStats) {
  1975. this.data[type][subType] = value;
  1976. }
  1977. }
  1978. add(type, subType, value) {
  1979. if (this.options.displayStats) {
  1980. this.data[type][subType] += value;
  1981. }
  1982. }
  1983. addAll(pageData) {
  1984. if (this.options.displayStats) {
  1985. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1986. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1987. }
  1988. }
  1989. }
  1990. return { getClass };
  1991. })();