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