single-file-core.js 92 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232
  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. /* global window */
  24. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  25. const DEBUG = false;
  26. const Set = window.Set;
  27. const Map = window.Map;
  28. let util, cssTree;
  29. function getClass(...args) {
  30. [util, cssTree] = args;
  31. return SingleFileClass;
  32. }
  33. class SingleFileClass {
  34. constructor(options) {
  35. this.options = options;
  36. }
  37. async run() {
  38. if (this.options.userScriptEnabled) {
  39. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  40. }
  41. this.runner = new Runner(this.options, true);
  42. await this.runner.loadPage();
  43. await this.runner.initialize();
  44. if (this.options.userScriptEnabled) {
  45. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  46. }
  47. await this.runner.run();
  48. }
  49. cancel() {
  50. this.cancelled = true;
  51. if (this.runner) {
  52. this.runner.cancel();
  53. }
  54. }
  55. getPageData() {
  56. return this.runner.getPageData();
  57. }
  58. }
  59. // -------------
  60. // ProgressEvent
  61. // -------------
  62. const PAGE_LOADING = "page-loading";
  63. const PAGE_LOADED = "page-loaded";
  64. const RESOURCES_INITIALIZING = "resource-initializing";
  65. const RESOURCES_INITIALIZED = "resources-initialized";
  66. const RESOURCE_LOADED = "resource-loaded";
  67. const PAGE_ENDED = "page-ended";
  68. const STAGE_STARTED = "stage-started";
  69. const STAGE_ENDED = "stage-ended";
  70. const STAGE_TASK_STARTED = "stage-task-started";
  71. const STAGE_TASK_ENDED = "stage-task-ended";
  72. class ProgressEvent {
  73. constructor(type, detail) {
  74. 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 };
  75. }
  76. }
  77. // ------
  78. // Runner
  79. // ------
  80. const RESOLVE_URLS_STAGE = 0;
  81. const REPLACE_DATA_STAGE = 1;
  82. const REPLACE_DOCS_STAGE = 2;
  83. const POST_PROCESS_STAGE = 3;
  84. const STAGES = [{
  85. sequential: [
  86. { action: "preProcessPage" },
  87. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  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. const matchResourceReferrer = this.options.url.match(/^.*\//);
  152. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  153. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  154. this.options.rootDocument = root;
  155. this.options.updatedResources = this.options.updatedResources || {};
  156. this.options.fontTests = new Map();
  157. this.batchRequest = new BatchRequest();
  158. this.processor = new Processor(options, this.batchRequest);
  159. if (rootDocDefined) {
  160. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  161. this.options.canvases = docData.canvases;
  162. this.options.fonts = docData.fonts;
  163. this.options.stylesheets = docData.stylesheets;
  164. this.options.images = docData.images;
  165. this.options.posters = docData.posters;
  166. this.options.usedFonts = docData.usedFonts;
  167. this.options.shadowRoots = docData.shadowRoots;
  168. this.options.imports = docData.imports;
  169. this.options.referrer = docData.referrer;
  170. this.markedElements = docData.markedElements;
  171. }
  172. if (this.options.saveRawPage) {
  173. this.options.removeFrames = true;
  174. }
  175. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  176. this.onprogress = options.onprogress || (() => { });
  177. }
  178. async loadPage() {
  179. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  180. await this.processor.loadPage(this.options.content);
  181. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  182. }
  183. async initialize() {
  184. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  185. await this.executeStage(RESOLVE_URLS_STAGE);
  186. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  187. if (this.root && this.options.doc) {
  188. util.postProcessDoc(this.options.doc, this.markedElements);
  189. }
  190. }
  191. cancel() {
  192. this.cancelled = true;
  193. this.batchRequest.cancel();
  194. if (this.root) {
  195. if (this.options.frames) {
  196. this.options.frames.forEach(cancelRunner);
  197. }
  198. if (this.options.imports) {
  199. this.options.imports.forEach(cancelRunner);
  200. }
  201. }
  202. function cancelRunner(resourceData) {
  203. if (resourceData.runner) {
  204. resourceData.runner.cancel();
  205. }
  206. }
  207. }
  208. async run() {
  209. if (this.root) {
  210. this.processor.initialize(this.batchRequest);
  211. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  212. }
  213. await this.batchRequest.run(detail => {
  214. detail.pageURL = this.options.url;
  215. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  216. }, this.options);
  217. await this.pendingPromises;
  218. this.options.doc = null;
  219. this.options.win = null;
  220. await this.executeStage(REPLACE_DOCS_STAGE);
  221. await this.executeStage(POST_PROCESS_STAGE);
  222. this.processor.finalize();
  223. }
  224. getDocument() {
  225. return this.processor.doc;
  226. }
  227. getStyleSheets() {
  228. return this.processor.stylesheets;
  229. }
  230. getPageData() {
  231. if (this.root) {
  232. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  233. }
  234. return this.processor.getPageData();
  235. }
  236. async executeStage(step) {
  237. if (DEBUG) {
  238. log("**** STARTED STAGE", step, "****");
  239. }
  240. const frame = !this.root;
  241. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  242. STAGES[step].sequential.forEach(task => {
  243. let startTime;
  244. if (DEBUG) {
  245. startTime = Date.now();
  246. log(" -- STARTED task =", task.action);
  247. }
  248. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  249. if (!this.cancelled) {
  250. this.executeTask(task);
  251. }
  252. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  253. if (DEBUG) {
  254. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  255. }
  256. });
  257. let parallelTasksPromise;
  258. if (STAGES[step].parallel) {
  259. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  260. let startTime;
  261. if (DEBUG) {
  262. startTime = Date.now();
  263. log(" // STARTED task =", task.action);
  264. }
  265. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  266. if (!this.cancelled) {
  267. await this.executeTask(task);
  268. }
  269. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  270. if (DEBUG) {
  271. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  272. }
  273. }));
  274. } else {
  275. parallelTasksPromise = Promise.resolve();
  276. }
  277. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  278. if (DEBUG) {
  279. log("**** ENDED STAGE", step, "****");
  280. }
  281. return parallelTasksPromise;
  282. }
  283. executeTask(task) {
  284. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  285. return this.processor[task.action]();
  286. }
  287. }
  288. }
  289. // ------------
  290. // BatchRequest
  291. // ------------
  292. class BatchRequest {
  293. constructor() {
  294. this.requests = new Map();
  295. this.duplicates = new Map();
  296. }
  297. addURL(resourceURL, asBinary, expectedType, groupDuplicates) {
  298. return new Promise((resolve, reject) => {
  299. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType]);
  300. let resourceRequests = this.requests.get(requestKey);
  301. if (!resourceRequests) {
  302. resourceRequests = [];
  303. this.requests.set(requestKey, resourceRequests);
  304. }
  305. const callbacks = { resolve, reject };
  306. resourceRequests.push(callbacks);
  307. if (groupDuplicates) {
  308. let duplicateRequests = this.duplicates.get(requestKey);
  309. if (!duplicateRequests) {
  310. duplicateRequests = [];
  311. this.duplicates.set(requestKey, duplicateRequests);
  312. }
  313. duplicateRequests.push(callbacks);
  314. }
  315. });
  316. }
  317. getMaxResources() {
  318. return this.requests.size;
  319. }
  320. run(onloadListener, options) {
  321. const resourceURLs = [...this.requests.keys()];
  322. let indexResource = 0;
  323. return Promise.all(resourceURLs.map(async requestKey => {
  324. const [resourceURL, asBinary, expectedType] = JSON.parse(requestKey);
  325. const resourceRequests = this.requests.get(requestKey);
  326. try {
  327. const currentIndexResource = indexResource;
  328. indexResource = indexResource + 1;
  329. const content = await util.getContent(resourceURL, {
  330. asBinary,
  331. expectedType,
  332. maxResourceSize: options.maxResourceSize,
  333. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  334. frameId: options.windowId,
  335. resourceReferrer: options.resourceReferrer
  336. });
  337. onloadListener({ url: resourceURL });
  338. if (!this.cancelled) {
  339. resourceRequests.forEach(callbacks => {
  340. const duplicateCallbacks = this.duplicates.get(requestKey);
  341. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  342. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  343. });
  344. }
  345. } catch (error) {
  346. indexResource = indexResource + 1;
  347. onloadListener({ url: resourceURL });
  348. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  349. }
  350. this.requests.delete(requestKey);
  351. }));
  352. }
  353. cancel() {
  354. this.cancelled = true;
  355. const resourceURLs = [...this.requests.keys()];
  356. resourceURLs.forEach(requestKey => {
  357. const resourceRequests = this.requests.get(requestKey);
  358. resourceRequests.forEach(callbacks => callbacks.reject());
  359. this.requests.delete(requestKey);
  360. });
  361. }
  362. }
  363. // ---------
  364. // Processor
  365. // ---------
  366. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  367. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  368. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  369. const SCRIPT_TAG_FOUND = /<script/gi;
  370. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  371. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  372. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  373. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  374. const UTF8_CHARSET = "utf-8";
  375. class Processor {
  376. constructor(options, batchRequest) {
  377. this.options = options;
  378. this.stats = new Stats(options);
  379. this.baseURI = normalizeURL(options.baseURI || options.url);
  380. this.batchRequest = batchRequest;
  381. this.stylesheets = new Map();
  382. this.styles = new Map();
  383. this.cssVariables = new Map();
  384. this.fontTests = options.fontTests;
  385. }
  386. initialize() {
  387. this.options.saveDate = new Date();
  388. this.options.saveUrl = this.options.url;
  389. if (this.options.enableMaff) {
  390. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url));
  391. }
  392. this.maxResources = this.batchRequest.getMaxResources();
  393. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  394. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  395. }
  396. if (!this.options.removeImports && this.options.imports) {
  397. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  398. }
  399. this.stats.set("processed", "resources", this.maxResources);
  400. }
  401. async loadPage(pageContent, charset) {
  402. let content;
  403. if (!pageContent || this.options.saveRawPage) {
  404. content = await util.getContent(this.baseURI, {
  405. maxResourceSize: this.options.maxResourceSize,
  406. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  407. charset,
  408. frameId: this.options.windowId,
  409. resourceReferrer: this.options.resourceReferrer
  410. });
  411. pageContent = content.data;
  412. }
  413. this.doc = util.parseDocContent(pageContent, this.baseURI);
  414. if (this.options.saveRawPage) {
  415. let charset;
  416. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  417. const charsetDeclaration = element.content.split(";")[1];
  418. if (charsetDeclaration && !charset) {
  419. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  420. }
  421. });
  422. if (charset && charset != content.charset) {
  423. return this.loadPage(pageContent, charset);
  424. }
  425. }
  426. this.workStyleElement = this.doc.createElement("style");
  427. this.doc.body.appendChild(this.workStyleElement);
  428. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  429. }
  430. finalize() {
  431. if (this.workStyleElement.parentNode) {
  432. this.workStyleElement.remove();
  433. }
  434. }
  435. async getPageData() {
  436. util.postProcessDoc(this.doc);
  437. const url = util.parseURL(this.baseURI);
  438. if (this.options.insertSingleFileComment) {
  439. const firstComment = this.doc.documentElement.firstChild;
  440. let infobarURL = this.options.saveUrl, infobarSaveDate = this.options.saveDate;
  441. if (firstComment.nodeType == 8 && (firstComment.textContent.includes(util.COMMENT_HEADER_LEGACY) || firstComment.textContent.includes(util.COMMENT_HEADER))) {
  442. const info = this.doc.documentElement.firstChild.textContent.split("\n");
  443. const [, , url, saveDate] = info;
  444. infobarURL = url.split("url: ")[1];
  445. infobarSaveDate = saveDate.split("saved date: ")[1];
  446. firstComment.remove();
  447. }
  448. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  449. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  450. " \n url: " + infobarURL +
  451. " \n saved date: " + infobarSaveDate +
  452. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  453. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  454. }
  455. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  456. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  457. if (!canonicalLink) {
  458. canonicalLink = this.doc.createElement("link");
  459. canonicalLink.setAttribute("rel", "canonical");
  460. this.doc.head.appendChild(canonicalLink);
  461. }
  462. if (canonicalLink && !canonicalLink.href) {
  463. canonicalLink.href = this.options.saveUrl;
  464. }
  465. }
  466. if (this.options.insertMetaNoIndex) {
  467. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  468. if (!metaElement) {
  469. metaElement = this.doc.createElement("meta");
  470. metaElement.setAttribute("name", "robots");
  471. metaElement.setAttribute("content", "noindex");
  472. this.doc.head.appendChild(metaElement);
  473. }
  474. }
  475. let size;
  476. if (this.options.displayStats) {
  477. size = util.getContentSize(this.doc.documentElement.outerHTML);
  478. }
  479. const content = util.serialize(this.doc, this.options.compressHTML);
  480. if (this.options.displayStats) {
  481. const contentSize = util.getContentSize(content);
  482. this.stats.set("processed", "HTML bytes", contentSize);
  483. this.stats.add("discarded", "HTML bytes", size - contentSize);
  484. }
  485. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  486. const replacementCharacter = this.options.filenameReplacementCharacter;
  487. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  488. if (!this.options.backgroundSave) {
  489. filename = filename.replace(/\//g, replacementCharacter);
  490. }
  491. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  492. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  493. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  494. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  495. filename = filename + "…" + extension;
  496. }
  497. if (!filename) {
  498. filename = "Unnamed page";
  499. }
  500. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  501. const pageData = {
  502. stats: this.stats.data,
  503. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  504. filename,
  505. content
  506. };
  507. if (this.options.addProof) {
  508. pageData.hash = await util.digest("SHA-256", content);
  509. }
  510. if (this.options.retrieveLinks) {
  511. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  512. }
  513. return pageData;
  514. }
  515. preProcessPage() {
  516. if (this.options.win) {
  517. 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));
  518. }
  519. if (this.options.images && !this.options.saveRawPage) {
  520. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  521. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  522. if (attributeValue) {
  523. const imageData = this.options.images[Number(attributeValue)];
  524. if (imageData) {
  525. if (this.options.removeHiddenElements && (
  526. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  527. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  528. )) {
  529. imgElement.setAttribute("src", EMPTY_IMAGE);
  530. } else {
  531. if (imageData.currentSrc) {
  532. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  533. imgElement.setAttribute("src", imageData.currentSrc);
  534. }
  535. if (this.options.loadDeferredImages) {
  536. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  537. imageData.src = imgElement.dataset.src;
  538. imgElement.setAttribute("src", imgElement.dataset.src);
  539. imgElement.removeAttribute("data-src");
  540. }
  541. }
  542. }
  543. }
  544. }
  545. });
  546. if (this.options.loadDeferredImages) {
  547. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  548. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  549. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  550. imgElement.removeAttribute("data-srcset");
  551. }
  552. });
  553. }
  554. }
  555. }
  556. replaceStyleContents() {
  557. if (this.options.stylesheets) {
  558. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  559. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  560. if (attributeValue) {
  561. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  562. if (stylesheetContent) {
  563. styleElement.textContent = stylesheetContent;
  564. }
  565. }
  566. });
  567. }
  568. }
  569. removeUnselectedElements() {
  570. removeUnmarkedElements(this.doc.body);
  571. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  572. function removeUnmarkedElements(element) {
  573. let selectedElementFound = false;
  574. Array.from(element.childNodes).forEach(node => {
  575. if (node.nodeType == 1) {
  576. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  577. selectedElementFound = selectedElementFound || isSelectedElement;
  578. if (isSelectedElement) {
  579. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  580. removeUnmarkedElements(node);
  581. } else if (selectedElementFound) {
  582. removeNode(node);
  583. } else {
  584. hideNode(node);
  585. }
  586. }
  587. });
  588. }
  589. function removeNode(node) {
  590. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  591. node.remove();
  592. } else {
  593. hideNode(node);
  594. }
  595. }
  596. function hideNode(node) {
  597. if (canHideNode(node)) {
  598. node.style.setProperty("display", "none", "important");
  599. Array.from(node.childNodes).forEach(removeNode);
  600. }
  601. }
  602. function canHideNode(node) {
  603. if (node.nodeType == 1) {
  604. const tagName = node.tagName && node.tagName.toLowerCase();
  605. return (tagName != "svg" && tagName != "style" && tagName != "link");
  606. }
  607. }
  608. }
  609. insertVideoPosters() {
  610. if (this.options.posters) {
  611. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  612. let videoElement;
  613. if (element.tagName == "VIDEO") {
  614. videoElement = element;
  615. } else {
  616. videoElement = element.parentElement;
  617. }
  618. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  619. if (attributeValue) {
  620. const posterURL = this.options.posters[Number(attributeValue)];
  621. if (!videoElement.poster && posterURL) {
  622. videoElement.setAttribute("poster", posterURL);
  623. }
  624. }
  625. });
  626. }
  627. }
  628. removeFrames() {
  629. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  630. this.stats.set("discarded", "frames", frameElements.length);
  631. this.stats.set("processed", "frames", frameElements.length);
  632. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  633. }
  634. removeScripts() {
  635. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  636. this.doc.querySelectorAll("[href]").forEach(element => {
  637. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  638. element.setAttribute("href", "");
  639. }
  640. });
  641. this.doc.querySelectorAll("[src]").forEach(element => {
  642. if (element.src && element.src.match(/^\s*javascript:/)) {
  643. element.removeAttribute("src");
  644. }
  645. });
  646. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  647. this.stats.set("discarded", "scripts", scriptElements.length);
  648. this.stats.set("processed", "scripts", scriptElements.length);
  649. scriptElements.forEach(element => element.remove());
  650. }
  651. removeVideoSources() {
  652. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  653. this.stats.set("discarded", "video sources", videoSourceElements.length);
  654. this.stats.set("processed", "video sources", videoSourceElements.length);
  655. videoSourceElements.forEach(element => {
  656. if (element.tagName == "SOURCE") {
  657. element.remove();
  658. } else {
  659. videoSourceElements.forEach(element => element.removeAttribute("src"));
  660. }
  661. });
  662. }
  663. removeAudioSources() {
  664. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  665. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  666. this.stats.set("processed", "audio sources", audioSourceElements.length);
  667. audioSourceElements.forEach(element => {
  668. if (element.tagName == "SOURCE") {
  669. element.remove();
  670. } else {
  671. audioSourceElements.forEach(element => element.removeAttribute("src"));
  672. }
  673. });
  674. }
  675. removeDiscardedResources() {
  676. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  677. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  678. 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\"])");
  679. this.stats.set("discarded", "objects", objectElements.length);
  680. this.stats.set("processed", "objects", objectElements.length);
  681. objectElements.forEach(element => element.remove());
  682. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  683. replacedAttributeValue.forEach(element => {
  684. let regExp;
  685. if (this.options.removeScripts) {
  686. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  687. } else {
  688. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  689. }
  690. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  691. if (relValue.length) {
  692. element.setAttribute("rel", relValue);
  693. } else {
  694. element.remove();
  695. }
  696. });
  697. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  698. if (this.options.removeHiddenElements) {
  699. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  700. }
  701. if (!this.options.saveFavicon) {
  702. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  703. }
  704. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  705. }
  706. resetCharsetMeta() {
  707. let charset;
  708. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  709. const charsetDeclaration = element.content.split(";")[1];
  710. if (charsetDeclaration && !charset) {
  711. charset = charsetDeclaration.split("=")[1];
  712. if (charset) {
  713. this.charset = charset.trim().toLowerCase();
  714. }
  715. }
  716. element.remove();
  717. });
  718. const metaElement = this.doc.createElement("meta");
  719. metaElement.setAttribute("charset", UTF8_CHARSET);
  720. if (this.doc.head.firstChild) {
  721. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  722. } else {
  723. this.doc.head.appendChild(metaElement);
  724. }
  725. }
  726. setInputValues() {
  727. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  728. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  729. input.setAttribute("value", value || "");
  730. });
  731. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  732. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  733. if (value == "true") {
  734. input.setAttribute("checked", "");
  735. }
  736. });
  737. this.doc.querySelectorAll("textarea").forEach(textarea => {
  738. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  739. textarea.textContent = value || "";
  740. });
  741. this.doc.querySelectorAll("select").forEach(select => {
  742. select.querySelectorAll("option").forEach(option => {
  743. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  744. if (selected) {
  745. option.setAttribute("selected", "");
  746. }
  747. });
  748. });
  749. }
  750. saveFavicon() {
  751. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  752. if (!faviconElement) {
  753. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  754. }
  755. if (!faviconElement) {
  756. faviconElement = this.doc.createElement("link");
  757. faviconElement.setAttribute("type", "image/x-icon");
  758. faviconElement.setAttribute("rel", "shortcut icon");
  759. faviconElement.setAttribute("href", "/favicon.ico");
  760. }
  761. this.doc.head.appendChild(faviconElement);
  762. }
  763. replaceCanvasElements() {
  764. if (this.options.canvases) {
  765. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  766. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  767. if (attributeValue) {
  768. const canvasData = this.options.canvases[Number(attributeValue)];
  769. if (canvasData) {
  770. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  771. this.stats.add("processed", "canvas", 1);
  772. }
  773. }
  774. });
  775. }
  776. }
  777. insertFonts() {
  778. if (this.options.fonts && this.options.fonts.length) {
  779. let stylesheetContent = "";
  780. this.options.fonts.forEach(fontData => {
  781. if (fontData["font-family"] && fontData.src) {
  782. stylesheetContent += "@font-face{";
  783. let stylesContent = "";
  784. Object.keys(fontData).forEach(fontStyle => {
  785. if (stylesContent) {
  786. stylesContent += ";";
  787. }
  788. stylesContent += fontStyle + ":" + fontData[fontStyle];
  789. });
  790. stylesheetContent += stylesContent + "}";
  791. }
  792. });
  793. if (stylesheetContent) {
  794. const styleElement = this.doc.createElement("style");
  795. styleElement.textContent = stylesheetContent;
  796. const existingStyleElement = this.doc.querySelector("style");
  797. if (existingStyleElement) {
  798. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  799. } else {
  800. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  801. }
  802. }
  803. }
  804. }
  805. removeHiddenElements() {
  806. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  807. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  808. this.stats.set("discarded", "hidden elements", removedElements.length);
  809. this.stats.set("processed", "hidden elements", removedElements.length);
  810. if (hiddenElements.length) {
  811. const styleElement = this.doc.createElement("style");
  812. styleElement.textContent = ".sf-hidden{display:none!important;}";
  813. this.doc.head.appendChild(styleElement);
  814. hiddenElements.forEach(element => {
  815. if (element.style.getPropertyValue("display") != "none") {
  816. if (element.style.getPropertyPriority("display") == "important") {
  817. element.style.setProperty("display", "none", "important");
  818. } else {
  819. element.classList.add("sf-hidden");
  820. }
  821. }
  822. });
  823. }
  824. removedElements.forEach(element => element.remove());
  825. }
  826. resolveHrefs() {
  827. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  828. const href = element.getAttribute("href").trim();
  829. if (!testIgnoredPath(href)) {
  830. let resolvedURL;
  831. try {
  832. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  833. } catch (error) {
  834. // ignored
  835. }
  836. if (resolvedURL) {
  837. const url = normalizeURL(this.options.url);
  838. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  839. resolvedURL = resolvedURL.substring(url.length);
  840. }
  841. try {
  842. element.setAttribute("href", resolvedURL);
  843. } catch (error) {
  844. // ignored
  845. }
  846. }
  847. }
  848. });
  849. }
  850. resolveStyleAttributeURLs() {
  851. this.doc.querySelectorAll("[style]").forEach(element => {
  852. let styleContent = element.getAttribute("style");
  853. if (this.options.compressCSS) {
  854. styleContent = util.compressCSS(styleContent);
  855. }
  856. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement);
  857. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  858. this.styles.set(element, declarationList);
  859. });
  860. }
  861. async resolveStylesheetURLs() {
  862. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]")).map(async element => {
  863. const options = {
  864. maxResourceSize: this.options.maxResourceSize,
  865. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  866. url: this.options.url,
  867. charset: this.charset,
  868. compressCSS: this.options.compressCSS,
  869. updatedResources: this.options.updatedResources,
  870. rootDocument: this.options.rootDocument,
  871. frameId: this.options.windowId,
  872. resourceReferrer: this.options.resourceReferrer
  873. };
  874. let mediaText;
  875. if (element.media) {
  876. mediaText = element.media.toLowerCase();
  877. }
  878. const stylesheetInfo = { mediaText };
  879. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  880. stylesheetInfo.scoped = true;
  881. }
  882. if (element.tagName == "LINK" && element.charset) {
  883. options.charset = element.charset;
  884. }
  885. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  886. }));
  887. if (this.options.rootDocument) {
  888. const newResources = Object.keys(this.options.updatedResources).filter(url => this.options.updatedResources[url].type == "stylesheet" && !this.options.updatedResources[url].retrieved).map(url => this.options.updatedResources[url]);
  889. await Promise.all(newResources.map(async resource => {
  890. resource.retrieved = true;
  891. const stylesheetInfo = {};
  892. const element = this.doc.createElement("style");
  893. this.doc.body.appendChild(element);
  894. element.textContent = resource.content;
  895. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, this.options, this.workStyleElement);
  896. }));
  897. }
  898. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  899. stylesheets.set(element, stylesheetInfo);
  900. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  901. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  902. options = Object.assign({}, options, { charset: getCharset(stylesheetContent) });
  903. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  904. }
  905. let stylesheet;
  906. try {
  907. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  908. } catch (error) {
  909. // ignored
  910. }
  911. if (stylesheet && stylesheet.children) {
  912. if (options.compressCSS) {
  913. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  914. }
  915. stylesheetInfo.stylesheet = stylesheet;
  916. } else {
  917. stylesheets.delete(element);
  918. }
  919. }
  920. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  921. let content;
  922. if (element.tagName == "LINK") {
  923. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  924. } else {
  925. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  926. }
  927. return content || "";
  928. }
  929. }
  930. async resolveFrameURLs() {
  931. if (!this.options.saveRawPage) {
  932. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  933. await Promise.all(frameElements.map(async frameElement => {
  934. if (frameElement.tagName == "OBJECT") {
  935. frameElement.setAttribute("data", "data:text/html,");
  936. } else {
  937. frameElement.removeAttribute("src");
  938. frameElement.removeAttribute("srcdoc");
  939. }
  940. Array.from(frameElement.childNodes).forEach(node => node.remove());
  941. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  942. if (this.options.frames && frameWindowId) {
  943. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  944. if (frameData) {
  945. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  946. }
  947. }
  948. }));
  949. }
  950. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  951. options.insertSingleFileComment = false;
  952. options.insertCanonicalLink = false;
  953. options.insertMetaNoIndex = false;
  954. options.saveFavicon = false;
  955. options.url = frameData.baseURI;
  956. options.windowId = frameWindowId;
  957. if (frameData.content) {
  958. options.content = frameData.content;
  959. options.canvases = frameData.canvases;
  960. options.fonts = frameData.fonts;
  961. options.stylesheets = frameData.stylesheets;
  962. options.images = frameData.images;
  963. options.posters = frameData.posters;
  964. options.usedFonts = frameData.usedFonts;
  965. options.shadowRoots = frameData.shadowRoots;
  966. options.imports = frameData.imports;
  967. frameData.runner = new Runner(options);
  968. frameData.frameElement = frameElement;
  969. await frameData.runner.loadPage();
  970. await frameData.runner.initialize();
  971. frameData.maxResources = batchRequest.getMaxResources();
  972. }
  973. }
  974. }
  975. insertShadowRootContents() {
  976. const doc = this.doc;
  977. const options = this.options;
  978. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  979. processElement(this.doc);
  980. if (this.options.removeScripts) {
  981. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  982. }
  983. const scriptElement = doc.createElement("script");
  984. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  985. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{let shadowRoot = element.parentElement.shadowRoot;if (!shadowRoot) {try {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()} catch (error) {} if (shadowRoot) {processNode(shadowRoot)}}})}})()`;
  986. doc.body.appendChild(scriptElement);
  987. }
  988. function processElement(element) {
  989. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  990. shadowRootElements.forEach(element => {
  991. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  992. if (attributeValue) {
  993. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  994. if (shadowRootData) {
  995. const templateElement = doc.createElement("template");
  996. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  997. if (shadowRootData.delegatesFocus) {
  998. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  999. }
  1000. if (shadowRootData.adoptedStyleSheets) {
  1001. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  1002. const styleElement = doc.createElement("style");
  1003. styleElement.textContent = stylesheetContent;
  1004. templateElement.appendChild(styleElement);
  1005. });
  1006. }
  1007. const shadowDoc = util.parseDocContent(shadowRootData.content);
  1008. if (shadowDoc.head) {
  1009. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  1010. if (metaCharset) {
  1011. metaCharset.remove();
  1012. }
  1013. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1014. }
  1015. if (shadowDoc.body) {
  1016. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1017. }
  1018. processElement(templateElement);
  1019. if (element.firstChild) {
  1020. element.insertBefore(templateElement, element.firstChild);
  1021. } else {
  1022. element.appendChild(templateElement);
  1023. }
  1024. }
  1025. }
  1026. });
  1027. }
  1028. }
  1029. async resolveHtmlImportURLs() {
  1030. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1031. await Promise.all(linkElements.map(async linkElement => {
  1032. const resourceURL = linkElement.href;
  1033. linkElement.removeAttribute("href");
  1034. const options = Object.create(this.options);
  1035. options.insertSingleFileComment = false;
  1036. options.insertCanonicalLink = false;
  1037. options.insertMetaNoIndex = false;
  1038. options.saveFavicon = false;
  1039. options.removeUnusedStyles = false;
  1040. options.removeAlternativeMedias = false;
  1041. options.removeUnusedFonts = false;
  1042. options.removeHiddenElements = false;
  1043. options.url = resourceURL;
  1044. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1045. if (attributeValue) {
  1046. const importData = options.imports[Number(attributeValue)];
  1047. if (importData) {
  1048. options.content = importData.content;
  1049. importData.runner = new Runner(options);
  1050. await importData.runner.loadPage();
  1051. await importData.runner.initialize();
  1052. if (!options.removeImports) {
  1053. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1054. }
  1055. importData.runner.getStyleSheets().forEach(stylesheet => {
  1056. const importedStyleElement = this.doc.createElement("style");
  1057. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1058. this.stylesheets.set(importedStyleElement, stylesheet);
  1059. });
  1060. }
  1061. }
  1062. if (options.removeImports) {
  1063. linkElement.remove();
  1064. this.stats.add("discarded", "HTML imports", 1);
  1065. }
  1066. }));
  1067. }
  1068. removeUnusedStyles() {
  1069. if (!this.mediaAllInfo) {
  1070. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1071. }
  1072. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1073. this.stats.set("processed", "CSS rules", stats.processed);
  1074. this.stats.set("discarded", "CSS rules", stats.discarded);
  1075. }
  1076. removeUnusedFonts() {
  1077. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1078. }
  1079. removeAlternativeMedias() {
  1080. const stats = util.minifyMedias(this.stylesheets);
  1081. this.stats.set("processed", "medias", stats.processed);
  1082. this.stats.set("discarded", "medias", stats.discarded);
  1083. }
  1084. async processStylesheets() {
  1085. this.options.fontURLs = new Map();
  1086. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1087. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1088. ));
  1089. }
  1090. async processStyleAttributes() {
  1091. return Promise.all([...this.styles].map(([, declarationList]) =>
  1092. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1093. ));
  1094. }
  1095. async processPageResources() {
  1096. const processAttributeArgs = [
  1097. ["link[href][rel*=\"icon\"]", "href", false, true],
  1098. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1099. ["img[src], input[src][type=image]", "src", true],
  1100. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1101. ["video[poster]", "poster"],
  1102. ["*[background]", "background"],
  1103. ["image", "xlink:href"],
  1104. ["image", "href"]
  1105. ];
  1106. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1107. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1108. );
  1109. resourcePromises = resourcePromises.concat([
  1110. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1111. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1112. ]);
  1113. if (!this.options.removeAudioSrc) {
  1114. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1115. }
  1116. if (!this.options.removeVideoSrc) {
  1117. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1118. }
  1119. await Promise.all(resourcePromises);
  1120. if (this.options.saveFavicon) {
  1121. ProcessorHelper.processShortcutIcons(this.doc);
  1122. }
  1123. }
  1124. async processScripts() {
  1125. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1126. let resourceURL;
  1127. let scriptSrc;
  1128. if (element.tagName == "SCRIPT") {
  1129. scriptSrc = element.getAttribute("src");
  1130. element.removeAttribute("src");
  1131. } else {
  1132. scriptSrc = element.getAttribute("href");
  1133. element.removeAttribute("href");
  1134. }
  1135. element.removeAttribute("integrity");
  1136. element.textContent = "";
  1137. try {
  1138. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1139. } catch (error) {
  1140. // ignored
  1141. }
  1142. if (testValidURL(resourceURL)) {
  1143. this.stats.add("processed", "scripts", 1);
  1144. const content = await util.getContent(resourceURL, {
  1145. asBinary: true,
  1146. charset: this.charset != UTF8_CHARSET && this.charset,
  1147. maxResourceSize: this.options.maxResourceSize,
  1148. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1149. frameId: this.options.windowId,
  1150. resourceReferrer: this.options.resourceReferrer
  1151. });
  1152. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1153. if (element.tagName == "SCRIPT") {
  1154. element.setAttribute("src", content.data);
  1155. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1156. element.setAttribute("async", "");
  1157. }
  1158. } else {
  1159. const scriptElement = this.doc.createElement("script");
  1160. scriptElement.setAttribute("src", content.data);
  1161. scriptElement.setAttribute("async", "");
  1162. element.parentElement.replaceChild(scriptElement, element);
  1163. }
  1164. }
  1165. }));
  1166. }
  1167. removeAlternativeImages() {
  1168. util.removeAlternativeImages(this.doc);
  1169. }
  1170. async removeAlternativeFonts() {
  1171. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1172. }
  1173. async processFrames() {
  1174. if (this.options.frames) {
  1175. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1176. await Promise.all(frameElements.map(async frameElement => {
  1177. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1178. if (frameWindowId) {
  1179. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1180. if (frameData) {
  1181. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1182. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1183. this.stats.add("processed", "frames", 1);
  1184. await frameData.runner.run();
  1185. const pageData = await frameData.runner.getPageData();
  1186. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1187. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1188. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1189. sandbox += " allow-scripts allow-same-origin";
  1190. }
  1191. frameElement.setAttribute("sandbox", sandbox);
  1192. if (frameElement.tagName == "OBJECT") {
  1193. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1194. } else {
  1195. if (frameElement.tagName == "FRAME") {
  1196. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1197. } else {
  1198. frameElement.setAttribute("srcdoc", pageData.content);
  1199. frameElement.removeAttribute("src");
  1200. }
  1201. }
  1202. this.stats.addAll(pageData);
  1203. } else {
  1204. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1205. this.stats.add("discarded", "frames", 1);
  1206. }
  1207. }
  1208. }
  1209. }));
  1210. }
  1211. }
  1212. async processHtmlImports() {
  1213. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1214. await Promise.all(linkElements.map(async linkElement => {
  1215. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1216. if (attributeValue) {
  1217. const importData = this.options.imports[Number(attributeValue)];
  1218. if (importData.runner) {
  1219. this.stats.add("processed", "HTML imports", 1);
  1220. await importData.runner.run();
  1221. const pageData = await importData.runner.getPageData();
  1222. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1223. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1224. this.stats.addAll(pageData);
  1225. } else {
  1226. this.stats.add("discarded", "HTML imports", 1);
  1227. }
  1228. }
  1229. }));
  1230. }
  1231. replaceStylesheets() {
  1232. this.doc.querySelectorAll("style").forEach(styleElement => {
  1233. const stylesheetInfo = this.stylesheets.get(styleElement);
  1234. if (stylesheetInfo) {
  1235. this.stylesheets.delete(styleElement);
  1236. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1237. styleElement.textContent = stylesheetContent;
  1238. if (stylesheetInfo.mediaText) {
  1239. styleElement.media = stylesheetInfo.mediaText;
  1240. }
  1241. } else {
  1242. styleElement.remove();
  1243. }
  1244. });
  1245. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1246. const stylesheetInfo = this.stylesheets.get(linkElement);
  1247. if (stylesheetInfo) {
  1248. this.stylesheets.delete(linkElement);
  1249. const styleElement = this.doc.createElement("style");
  1250. if (stylesheetInfo.mediaText) {
  1251. styleElement.media = stylesheetInfo.mediaText;
  1252. }
  1253. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1254. styleElement.textContent = stylesheetContent;
  1255. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1256. } else {
  1257. linkElement.remove();
  1258. }
  1259. });
  1260. }
  1261. replaceStyleAttributes() {
  1262. this.doc.querySelectorAll("[style]").forEach(element => {
  1263. const declarations = this.styles.get(element);
  1264. if (declarations) {
  1265. this.styles.delete(element);
  1266. let styleContent = cssTree.generate(declarations);
  1267. element.setAttribute("style", styleContent);
  1268. } else {
  1269. element.setAttribute("style", "");
  1270. }
  1271. });
  1272. }
  1273. insertVariables() {
  1274. if (this.cssVariables.size) {
  1275. const styleElement = this.doc.createElement("style");
  1276. const firstStyleElement = this.doc.head.querySelector("style");
  1277. if (firstStyleElement) {
  1278. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1279. } else {
  1280. this.doc.head.appendChild(styleElement);
  1281. }
  1282. let stylesheetContent = "";
  1283. this.cssVariables.forEach((content, indexResource) => {
  1284. this.cssVariables.delete(indexResource);
  1285. if (stylesheetContent) {
  1286. stylesheetContent += ";";
  1287. }
  1288. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1289. });
  1290. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1291. }
  1292. }
  1293. compressHTML() {
  1294. let size;
  1295. if (this.options.displayStats) {
  1296. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1297. }
  1298. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1299. if (this.options.displayStats) {
  1300. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1301. }
  1302. }
  1303. cleanupPage() {
  1304. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1305. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1306. if (metaCharset) {
  1307. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1308. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1309. this.doc.head.querySelector("meta[charset]").remove();
  1310. }
  1311. }
  1312. }
  1313. resetZoomLevel() {
  1314. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1315. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1316. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1317. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1318. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1319. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1320. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1321. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1322. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1323. this.doc.documentElement.style.removeProperty("-sf-transform");
  1324. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1325. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1326. }
  1327. async insertMAFFMetaData() {
  1328. const maffMetaData = await this.maffMetaDataPromise;
  1329. if (maffMetaData && maffMetaData.content) {
  1330. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1331. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1332. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1333. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1334. if (originalURLElement) {
  1335. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1336. }
  1337. if (archiveTimeElement) {
  1338. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1339. if (value) {
  1340. const date = new Date(value);
  1341. if (!isNaN(date.getTime())) {
  1342. this.options.saveDate = new Date(value);
  1343. }
  1344. }
  1345. }
  1346. }
  1347. }
  1348. async setDocInfo() {
  1349. const titleElement = this.doc.querySelector("title");
  1350. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1351. const authorElement = this.doc.querySelector("meta[name=author]");
  1352. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1353. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1354. const headingElement = this.doc.querySelector("h1");
  1355. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1356. this.options.info = {
  1357. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1358. lang: this.doc.documentElement.lang,
  1359. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1360. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1361. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1362. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1363. };
  1364. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1365. }
  1366. }
  1367. // ---------------
  1368. // ProcessorHelper
  1369. // ---------------
  1370. const DATA_URI_PREFIX = "data:";
  1371. const ABOUT_BLANK_URI = "about:blank";
  1372. const EMPTY_DATA_URI = "data:null;base64,";
  1373. const REGEXP_URL_HASH = /(#.+?)$/;
  1374. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1375. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1376. class ProcessorHelper {
  1377. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1378. const date = options.saveDate;
  1379. const url = util.parseURL(options.saveUrl);
  1380. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1387. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1388. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1397. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1398. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1401. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1402. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1403. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1404. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1405. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1406. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1407. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1408. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1409. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1410. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1411. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1412. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1413. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1414. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1415. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1416. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1417. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1418. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1419. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1420. for (const [name, value] of params) {
  1421. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1422. }
  1423. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1424. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1425. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1426. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1427. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1428. if (content) {
  1429. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1430. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1431. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1432. }
  1433. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1434. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1435. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1436. template = await evalTemplateVariable(template, "profile-name", () => options.profileName, dontReplaceSlash, options.filenameReplacementCharacter);
  1437. return template.trim();
  1438. function decode(value) {
  1439. try {
  1440. return decodeURI(value);
  1441. } catch (error) {
  1442. return value;
  1443. }
  1444. }
  1445. }
  1446. static setBackgroundImage(element, url, style) {
  1447. element.style.setProperty("background-blend-mode", "normal", "important");
  1448. element.style.setProperty("background-clip", "content-box", "important");
  1449. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1450. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1451. element.style.setProperty("background-image", url, "important");
  1452. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1453. element.style.setProperty("background-origin", "content-box", "important");
  1454. element.style.setProperty("background-repeat", "no-repeat", "important");
  1455. }
  1456. static processShortcutIcons(doc) {
  1457. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1458. if (!shortcutIcon) {
  1459. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1460. if (shortcutIcon) {
  1461. shortcutIcon.rel = "icon";
  1462. }
  1463. }
  1464. if (shortcutIcon) {
  1465. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1466. if (linkElement != shortcutIcon) {
  1467. linkElement.remove();
  1468. }
  1469. });
  1470. }
  1471. }
  1472. static removeSingleLineCssComments(stylesheet) {
  1473. const removedRules = [];
  1474. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1475. const ruleData = cssRule.data;
  1476. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1477. removedRules.push(cssRule);
  1478. }
  1479. }
  1480. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1481. }
  1482. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1483. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet);
  1484. const imports = getImportFunctions(stylesheetContent);
  1485. await Promise.all(imports.map(async cssImport => {
  1486. const match = matchImport(cssImport);
  1487. if (match) {
  1488. const regExpCssImport = getRegExp(cssImport);
  1489. let resourceURL = normalizeURL(match.resourceURL);
  1490. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1491. try {
  1492. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1493. } catch (error) {
  1494. // ignored
  1495. }
  1496. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1497. const content = await getStylesheetContent(resourceURL);
  1498. resourceURL = content.resourceURL;
  1499. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1500. let importedStylesheetContent = removeCssComments(content.data);
  1501. if (options.compressCSS) {
  1502. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1503. }
  1504. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1505. if (stylesheetContent.includes(cssImport)) {
  1506. const ancestorStyleSheets = new Set(importedStyleSheets);
  1507. ancestorStyleSheets.add(resourceURL);
  1508. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1509. workStylesheet.textContent = importedStylesheetContent;
  1510. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1511. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1512. } else {
  1513. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1514. }
  1515. }
  1516. } else {
  1517. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1518. }
  1519. } else {
  1520. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1521. }
  1522. }
  1523. }));
  1524. return stylesheetContent;
  1525. async function getStylesheetContent(resourceURL) {
  1526. const content = await util.getContent(resourceURL, {
  1527. maxResourceSize: options.maxResourceSize,
  1528. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1529. validateTextContentType: true,
  1530. frameId: options.frameId,
  1531. charset: options.charset,
  1532. resourceReferrer: options.resourceReferrer
  1533. });
  1534. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1535. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1536. return getStylesheetContent(resourceURL);
  1537. } else {
  1538. return content;
  1539. }
  1540. }
  1541. }
  1542. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet) {
  1543. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1544. urlFunctions.map(urlFunction => {
  1545. const originalResourceURL = matchURL(urlFunction);
  1546. let resourceURL = normalizeURL(originalResourceURL);
  1547. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1548. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1549. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1550. }
  1551. if (!testIgnoredPath(resourceURL)) {
  1552. if (!resourceURL || testValidPath(resourceURL)) {
  1553. let resolvedURL;
  1554. if (!originalResourceURL.startsWith("#")) {
  1555. try {
  1556. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1557. } catch (error) {
  1558. // ignored
  1559. }
  1560. }
  1561. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1562. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1563. }
  1564. } else {
  1565. let newUrlFunction;
  1566. if (originalResourceURL) {
  1567. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1568. } else {
  1569. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1570. }
  1571. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1572. }
  1573. }
  1574. });
  1575. return stylesheetContent;
  1576. }
  1577. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1578. resourceURL = normalizeURL(resourceURL);
  1579. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1580. const content = await util.getContent(resourceURL, {
  1581. maxResourceSize: options.maxResourceSize,
  1582. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1583. charset: options.charset,
  1584. frameId: options.frameId,
  1585. resourceReferrer: options.resourceReferrer,
  1586. validateTextContentType: true
  1587. });
  1588. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1589. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1590. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1591. }
  1592. resourceURL = content.resourceURL;
  1593. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1594. let stylesheetContent = removeCssComments(content.data);
  1595. if (options.compressCSS) {
  1596. stylesheetContent = util.compressCSS(stylesheetContent);
  1597. }
  1598. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1599. return stylesheetContent;
  1600. }
  1601. }
  1602. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1603. const promises = [];
  1604. const removedRules = [];
  1605. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1606. const ruleData = cssRule.data;
  1607. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1608. removedRules.push(cssRule);
  1609. } else if (ruleData.block && ruleData.block.children) {
  1610. if (ruleData.type == "Rule") {
  1611. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1612. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1613. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1614. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1615. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1616. }
  1617. }
  1618. }
  1619. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1620. await Promise.all(promises);
  1621. async function processFontFaceRule(ruleData, fontURLs) {
  1622. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1623. if (declaration.type == "Declaration" && declaration.value.children) {
  1624. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1625. await Promise.all(urlFunctions.map(async urlFunction => {
  1626. const originalResourceURL = matchURL(urlFunction);
  1627. const resourceURL = normalizeURL(originalResourceURL);
  1628. if (!testIgnoredPath(resourceURL)) {
  1629. if (testValidURL(resourceURL)) {
  1630. let { content } = await batchRequest.addURL(resourceURL, true, "font");
  1631. let resourceURLs = fontURLs.get(declaration);
  1632. if (!resourceURLs) {
  1633. resourceURLs = [];
  1634. fontURLs.set(declaration, resourceURLs);
  1635. }
  1636. resourceURLs.push(resourceURL);
  1637. replaceURLs(declaration, originalResourceURL, content);
  1638. }
  1639. }
  1640. }));
  1641. }
  1642. }));
  1643. function replaceURLs(declaration, oldURL, newURL) {
  1644. declaration.value.children.forEach(token => {
  1645. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1646. token.value.value = newURL;
  1647. }
  1648. });
  1649. }
  1650. }
  1651. }
  1652. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1653. await Promise.all(declarations.map(async declaration => {
  1654. if (declaration.value && !declaration.value.children && declaration.value.type == "Raw") {
  1655. try {
  1656. declaration.value = cssTree.parse(declaration.value.value, { context: "value" });
  1657. } catch (error) {
  1658. // ignored
  1659. }
  1660. }
  1661. if (declaration.type == "Declaration" && declaration.value.children) {
  1662. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1663. await Promise.all(urlFunctions.map(async urlFunction => {
  1664. const originalResourceURL = matchURL(urlFunction);
  1665. const resourceURL = normalizeURL(originalResourceURL);
  1666. if (!testIgnoredPath(resourceURL)) {
  1667. if (testValidURL(resourceURL)) {
  1668. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", true);
  1669. let variableDefined;
  1670. const tokens = [];
  1671. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1672. if (!originalResourceURL.startsWith("#")) {
  1673. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1674. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1675. tokens.push({ parent, token, value });
  1676. variableDefined = true;
  1677. } else {
  1678. token.data.value.value = content;
  1679. }
  1680. }
  1681. });
  1682. if (variableDefined) {
  1683. cssVariables.set(indexResource, content);
  1684. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1685. }
  1686. }
  1687. }
  1688. }));
  1689. }
  1690. }));
  1691. function findURLToken(url, children, callback, depth = 0) {
  1692. for (let token = children.head; token; token = token.next) {
  1693. if (token.data.children) {
  1694. findURLToken(url, token.data.children, callback, depth + 1);
  1695. }
  1696. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1697. callback(token, children, depth == 0);
  1698. }
  1699. }
  1700. }
  1701. }
  1702. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1703. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1704. let resourceURL = resourceElement.getAttribute(attributeName);
  1705. if (resourceURL != null) {
  1706. resourceURL = normalizeURL(resourceURL);
  1707. let originURL = resourceElement.dataset.singleFileOriginURL;
  1708. delete resourceElement.dataset.singleFileOriginURL;
  1709. if (!testIgnoredPath(resourceURL)) {
  1710. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1711. if (testValidPath(resourceURL)) {
  1712. try {
  1713. resourceURL = util.resolveURL(resourceURL, baseURI);
  1714. } catch (error) {
  1715. // ignored
  1716. }
  1717. if (testValidURL(resourceURL)) {
  1718. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", resourceElement.tagName == "IMG" && attributeName == "src");
  1719. if (originURL) {
  1720. if (content == EMPTY_DATA_URI) {
  1721. try {
  1722. originURL = util.resolveURL(originURL, baseURI);
  1723. } catch (error) {
  1724. // ignored
  1725. }
  1726. try {
  1727. resourceURL = originURL;
  1728. content = (await util.getContent(resourceURL, {
  1729. asBinary: true,
  1730. expectedType: "image",
  1731. maxResourceSize: options.maxResourceSize,
  1732. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1733. frameId: options.windowId,
  1734. resourceReferrer: options.resourceReferrer
  1735. })).data;
  1736. } catch (error) {
  1737. // ignored
  1738. }
  1739. }
  1740. }
  1741. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1742. resourceElement.remove();
  1743. } else {
  1744. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1745. if (!forbiddenPrefixFound) {
  1746. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1747. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1748. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1749. cssVariables.set(indexResource, content);
  1750. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1751. styles.set(resourceElement, declarationList);
  1752. } else {
  1753. resourceElement.setAttribute(attributeName, content);
  1754. }
  1755. } else {
  1756. resourceElement.setAttribute(attributeName, content);
  1757. }
  1758. }
  1759. }
  1760. }
  1761. }
  1762. }
  1763. }
  1764. }));
  1765. }
  1766. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1767. let attributeName = "xlink:href";
  1768. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1769. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1770. if (originalResourceURL == null) {
  1771. attributeName = "href";
  1772. originalResourceURL = resourceElement.getAttribute(attributeName);
  1773. }
  1774. let resourceURL = normalizeURL(originalResourceURL);
  1775. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1776. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1777. try {
  1778. resourceURL = util.resolveURL(resourceURL, baseURI);
  1779. } catch (error) {
  1780. // ignored
  1781. }
  1782. if (testValidURL(resourceURL)) {
  1783. const { content } = await batchRequest.addURL(resourceURL);
  1784. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1785. if (hashMatch && hashMatch[0]) {
  1786. let symbolElement;
  1787. try {
  1788. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1789. } catch (error) {
  1790. // ignored
  1791. }
  1792. if (symbolElement) {
  1793. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1794. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1795. }
  1796. } else {
  1797. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1798. }
  1799. }
  1800. } else if (resourceURL == options.url) {
  1801. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1802. }
  1803. }));
  1804. }
  1805. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1806. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1807. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1808. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1809. let resourceURL = normalizeURL(srcsetValue.url);
  1810. if (!testIgnoredPath(resourceURL)) {
  1811. if (testValidPath(resourceURL)) {
  1812. try {
  1813. resourceURL = util.resolveURL(resourceURL, baseURI);
  1814. } catch (error) {
  1815. // ignored
  1816. }
  1817. if (testValidURL(resourceURL)) {
  1818. const { content } = await batchRequest.addURL(resourceURL, true, "image");
  1819. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1820. if (forbiddenPrefixFound) {
  1821. return "";
  1822. }
  1823. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1824. } else {
  1825. return "";
  1826. }
  1827. } else {
  1828. return "";
  1829. }
  1830. } else {
  1831. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1832. }
  1833. }));
  1834. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1835. }));
  1836. }
  1837. static replaceImageSource(imgElement, variableName, options) {
  1838. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1839. if (attributeValue) {
  1840. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1841. if (imageData && imageData.replaceable) {
  1842. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1843. const backgroundStyle = {};
  1844. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1845. if (backgroundSize) {
  1846. backgroundStyle["background-size"] = imageData.objectFit;
  1847. }
  1848. if (imageData.objectPosition) {
  1849. backgroundStyle["background-position"] = imageData.objectPosition;
  1850. }
  1851. if (imageData.backgroundColor) {
  1852. backgroundStyle["background-color"] = imageData.backgroundColor;
  1853. }
  1854. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1855. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1856. return true;
  1857. }
  1858. }
  1859. }
  1860. }
  1861. // ----
  1862. // Util
  1863. // ----
  1864. const BLOB_URI_PREFIX = "blob:";
  1865. const HTTP_URI_PREFIX = /^https?:\/\//;
  1866. const FILE_URI_PREFIX = /^file:\/\//;
  1867. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1868. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1869. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1870. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1871. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1872. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1873. 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;
  1874. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1875. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1876. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1877. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1878. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1879. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1880. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1881. function getUpdatedResourceContent(resourceURL, content, options) {
  1882. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1883. options.updatedResources[resourceURL].retrieved = true;
  1884. return options.updatedResources[resourceURL].content;
  1885. } else {
  1886. return content.data || "";
  1887. }
  1888. }
  1889. function normalizeURL(url) {
  1890. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1891. return url;
  1892. } else {
  1893. return url.split("#")[0];
  1894. }
  1895. }
  1896. function getCSSValue(value) {
  1897. let result = "";
  1898. try {
  1899. result = cssTree.generate(value);
  1900. } catch (error) {
  1901. // ignored
  1902. }
  1903. return result;
  1904. }
  1905. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  1906. const stylesheetCharset = getCharset(stylesheetContent);
  1907. if (stylesheetCharset) {
  1908. return stylesheetCharset == charset.toLowerCase();
  1909. } else {
  1910. return true;
  1911. }
  1912. }
  1913. function getCharset(stylesheetContent) {
  1914. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  1915. if (match && match[1]) {
  1916. return match[1].toLowerCase().trim();
  1917. }
  1918. }
  1919. function getOnEventAttributeNames(doc) {
  1920. const element = doc.createElement("div");
  1921. const attributeNames = [];
  1922. for (const propertyName in element) {
  1923. if (propertyName.startsWith("on")) {
  1924. attributeNames.push(propertyName);
  1925. }
  1926. }
  1927. attributeNames.push("onunload");
  1928. return attributeNames;
  1929. }
  1930. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1931. let maxLength;
  1932. if (template) {
  1933. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1934. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1935. if (template.match(replaceRegExp)) {
  1936. const matchedLength = template.match(replaceRegExp)[0];
  1937. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1938. if (isNaN(maxLength) || maxLength <= 0) {
  1939. maxLength = null;
  1940. }
  1941. } else {
  1942. replaceRegExp = new RegExp(regExpVariable, "g");
  1943. }
  1944. if (template.match(replaceRegExp)) {
  1945. let value = await valueGetter();
  1946. if (!dontReplaceSlash) {
  1947. value = value.replace(/\/+/g, replacementCharacter);
  1948. }
  1949. if (maxLength) {
  1950. value = await util.truncateText(value, maxLength);
  1951. }
  1952. return template.replace(replaceRegExp, value);
  1953. }
  1954. }
  1955. return template;
  1956. }
  1957. function getLastSegment(url, replacementCharacter) {
  1958. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1959. if (!lastSegment) {
  1960. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1961. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1962. }
  1963. if (!lastSegment) {
  1964. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1965. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1966. }
  1967. if (!lastSegment) {
  1968. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1969. }
  1970. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1971. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1972. lastSegment = lastSegmentMatch[1];
  1973. }
  1974. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1975. return lastSegment;
  1976. }
  1977. function getRegExp(string) {
  1978. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1979. }
  1980. function getUrlFunctions(stylesheetContent, unique) {
  1981. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1982. if (unique) {
  1983. return [...new Set(result)];
  1984. } else {
  1985. return result;
  1986. }
  1987. }
  1988. function getImportFunctions(stylesheetContent) {
  1989. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1990. }
  1991. function findShortcutIcon(shortcutIcons) {
  1992. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1993. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1994. return shortcutIcons[0];
  1995. }
  1996. function matchURL(stylesheetContent) {
  1997. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1998. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1999. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  2000. return match && match[1];
  2001. }
  2002. function testIgnoredPath(resourceURL) {
  2003. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  2004. }
  2005. function testValidPath(resourceURL) {
  2006. return resourceURL && !resourceURL.match(EMPTY_URL);
  2007. }
  2008. function testValidURL(resourceURL) {
  2009. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  2010. }
  2011. function matchImport(stylesheetContent) {
  2012. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  2013. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  2014. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2015. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2016. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2017. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2018. if (match) {
  2019. const [, resourceURL, media] = match;
  2020. return { resourceURL, media };
  2021. }
  2022. }
  2023. function removeCssComments(stylesheetContent) {
  2024. try {
  2025. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2026. } catch (error) {
  2027. let start, end;
  2028. do {
  2029. start = stylesheetContent.indexOf("/*");
  2030. end = stylesheetContent.indexOf("*/", start + 2);
  2031. if (start != -1 && end != -1) {
  2032. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2033. }
  2034. } while (start != -1 && end != -1);
  2035. return stylesheetContent;
  2036. }
  2037. }
  2038. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2039. if (mediaQuery) {
  2040. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2041. } else {
  2042. return stylesheetContent;
  2043. }
  2044. }
  2045. function log(...args) {
  2046. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2047. }
  2048. // -----
  2049. // Stats
  2050. // -----
  2051. const STATS_DEFAULT_VALUES = {
  2052. discarded: {
  2053. "HTML bytes": 0,
  2054. "hidden elements": 0,
  2055. "HTML imports": 0,
  2056. scripts: 0,
  2057. objects: 0,
  2058. "audio sources": 0,
  2059. "video sources": 0,
  2060. frames: 0,
  2061. "CSS rules": 0,
  2062. canvas: 0,
  2063. stylesheets: 0,
  2064. resources: 0,
  2065. medias: 0
  2066. },
  2067. processed: {
  2068. "HTML bytes": 0,
  2069. "hidden elements": 0,
  2070. "HTML imports": 0,
  2071. scripts: 0,
  2072. objects: 0,
  2073. "audio sources": 0,
  2074. "video sources": 0,
  2075. frames: 0,
  2076. "CSS rules": 0,
  2077. canvas: 0,
  2078. stylesheets: 0,
  2079. resources: 0,
  2080. medias: 0
  2081. }
  2082. };
  2083. class Stats {
  2084. constructor(options) {
  2085. this.options = options;
  2086. if (options.displayStats) {
  2087. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2088. }
  2089. }
  2090. set(type, subType, value) {
  2091. if (this.options.displayStats) {
  2092. this.data[type][subType] = value;
  2093. }
  2094. }
  2095. add(type, subType, value) {
  2096. if (this.options.displayStats) {
  2097. this.data[type][subType] += value;
  2098. }
  2099. }
  2100. addAll(pageData) {
  2101. if (this.options.displayStats) {
  2102. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2103. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2104. }
  2105. }
  2106. }
  2107. return { getClass };
  2108. })();