single-file-core.js 91 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. }
  34. async run() {
  35. if (this.options.userScriptEnabled) {
  36. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  37. }
  38. this.runner = new Runner(this.options, true);
  39. await this.runner.loadPage();
  40. await this.runner.initialize();
  41. if (this.options.userScriptEnabled) {
  42. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  43. }
  44. await this.runner.run();
  45. }
  46. cancel() {
  47. this.cancelled = true;
  48. if (this.runner) {
  49. this.runner.cancel();
  50. }
  51. }
  52. getPageData() {
  53. return this.runner.getPageData();
  54. }
  55. }
  56. // -------------
  57. // ProgressEvent
  58. // -------------
  59. const PAGE_LOADING = "page-loading";
  60. const PAGE_LOADED = "page-loaded";
  61. const RESOURCES_INITIALIZING = "resource-initializing";
  62. const RESOURCES_INITIALIZED = "resources-initialized";
  63. const RESOURCE_LOADED = "resource-loaded";
  64. const PAGE_ENDED = "page-ended";
  65. const STAGE_STARTED = "stage-started";
  66. const STAGE_ENDED = "stage-ended";
  67. const STAGE_TASK_STARTED = "stage-task-started";
  68. const STAGE_TASK_ENDED = "stage-task-ended";
  69. class ProgressEvent {
  70. constructor(type, detail) {
  71. 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 };
  72. }
  73. }
  74. // ------
  75. // Runner
  76. // ------
  77. const RESOLVE_URLS_STAGE = 0;
  78. const REPLACE_DATA_STAGE = 1;
  79. const REPLACE_DOCS_STAGE = 2;
  80. const POST_PROCESS_STAGE = 3;
  81. const STAGES = [{
  82. sequential: [
  83. { action: "preProcessPage" },
  84. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  85. { action: "replaceStyleContents" },
  86. { action: "resetCharsetMeta" },
  87. { option: "saveFavicon", action: "saveFavicon" },
  88. { action: "replaceCanvasElements" },
  89. { action: "insertFonts" },
  90. { action: "insertShadowRootContents" },
  91. { action: "setInputValues" },
  92. { option: "removeScripts", action: "removeScripts" },
  93. { option: "selected", action: "removeUnselectedElements" },
  94. { option: "removeVideoSrc", action: "insertVideoPosters" },
  95. { option: "removeFrames", action: "removeFrames" },
  96. { option: "removeVideoSrc", action: "removeVideoSources" },
  97. { option: "removeAudioSrc", action: "removeAudioSources" },
  98. { action: "removeDiscardedResources" },
  99. { option: "removeHiddenElements", action: "removeHiddenElements" },
  100. { action: "resolveHrefs" },
  101. { action: "resolveStyleAttributeURLs" }
  102. ],
  103. parallel: [
  104. { action: "resolveStylesheetURLs" },
  105. { option: "!removeFrames", action: "resolveFrameURLs" },
  106. { action: "resolveHtmlImportURLs" }
  107. ]
  108. }, {
  109. sequential: [
  110. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  111. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  112. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  113. ],
  114. parallel: [
  115. { action: "processStylesheets" },
  116. { action: "processStyleAttributes" },
  117. { action: "processPageResources" },
  118. { option: "!removeScripts", action: "processScripts" }
  119. ]
  120. }, {
  121. sequential: [
  122. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  123. ],
  124. parallel: [
  125. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  126. { option: "!removeFrames", action: "processFrames" },
  127. { option: "!removeImports", action: "processHtmlImports" },
  128. ]
  129. }, {
  130. sequential: [
  131. { action: "replaceStylesheets" },
  132. { action: "replaceStyleAttributes" },
  133. { action: "insertVariables" },
  134. { option: "compressHTML", action: "compressHTML" },
  135. { action: "cleanupPage" }
  136. ],
  137. parallel: [
  138. { option: "enableMaff", action: "insertMAFFMetaData" },
  139. { action: "setDocInfo" }
  140. ]
  141. }];
  142. class Runner {
  143. constructor(options, root) {
  144. const rootDocDefined = root && options.doc;
  145. this.root = root;
  146. this.options = options;
  147. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  148. const matchResourceReferrer = this.options.url.match(/^.*\//);
  149. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  150. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  151. this.options.rootDocument = root;
  152. this.options.updatedResources = this.options.updatedResources || {};
  153. this.options.fontTests = new Map();
  154. this.batchRequest = new BatchRequest();
  155. this.processor = new Processor(options, this.batchRequest);
  156. if (rootDocDefined) {
  157. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  158. this.options.canvases = docData.canvases;
  159. this.options.fonts = docData.fonts;
  160. this.options.stylesheets = docData.stylesheets;
  161. this.options.images = docData.images;
  162. this.options.posters = docData.posters;
  163. this.options.usedFonts = docData.usedFonts;
  164. this.options.shadowRoots = docData.shadowRoots;
  165. this.options.imports = docData.imports;
  166. this.options.referrer = docData.referrer;
  167. this.markedElements = docData.markedElements;
  168. }
  169. if (this.options.saveRawPage) {
  170. this.options.removeFrames = true;
  171. }
  172. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  173. this.onprogress = options.onprogress || (() => { });
  174. }
  175. async loadPage() {
  176. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  177. await this.processor.loadPage(this.options.content);
  178. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  179. }
  180. async initialize() {
  181. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  182. await this.executeStage(RESOLVE_URLS_STAGE);
  183. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  184. if (this.root && this.options.doc) {
  185. util.postProcessDoc(this.options.doc, this.markedElements);
  186. }
  187. }
  188. cancel() {
  189. this.cancelled = true;
  190. this.batchRequest.cancel();
  191. if (this.root) {
  192. if (this.options.frames) {
  193. this.options.frames.forEach(cancelRunner);
  194. }
  195. if (this.options.imports) {
  196. this.options.imports.forEach(cancelRunner);
  197. }
  198. }
  199. function cancelRunner(resourceData) {
  200. if (resourceData.runner) {
  201. resourceData.runner.cancel();
  202. }
  203. }
  204. }
  205. async run() {
  206. if (this.root) {
  207. this.processor.initialize(this.batchRequest);
  208. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  209. }
  210. await this.batchRequest.run(detail => {
  211. detail.pageURL = this.options.url;
  212. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  213. }, this.options);
  214. await this.pendingPromises;
  215. this.options.doc = null;
  216. this.options.win = null;
  217. await this.executeStage(REPLACE_DOCS_STAGE);
  218. await this.executeStage(POST_PROCESS_STAGE);
  219. this.processor.finalize();
  220. }
  221. getDocument() {
  222. return this.processor.doc;
  223. }
  224. getStyleSheets() {
  225. return this.processor.stylesheets;
  226. }
  227. getPageData() {
  228. if (this.root) {
  229. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  230. }
  231. return this.processor.getPageData();
  232. }
  233. async executeStage(step) {
  234. if (DEBUG) {
  235. log("**** STARTED STAGE", step, "****");
  236. }
  237. const frame = !this.root;
  238. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  239. STAGES[step].sequential.forEach(task => {
  240. let startTime;
  241. if (DEBUG) {
  242. startTime = Date.now();
  243. log(" -- STARTED task =", task.action);
  244. }
  245. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  246. if (!this.cancelled) {
  247. this.executeTask(task);
  248. }
  249. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  250. if (DEBUG) {
  251. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  252. }
  253. });
  254. let parallelTasksPromise;
  255. if (STAGES[step].parallel) {
  256. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  257. let startTime;
  258. if (DEBUG) {
  259. startTime = Date.now();
  260. log(" // STARTED task =", task.action);
  261. }
  262. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  263. if (!this.cancelled) {
  264. await this.executeTask(task);
  265. }
  266. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  267. if (DEBUG) {
  268. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  269. }
  270. }));
  271. } else {
  272. parallelTasksPromise = Promise.resolve();
  273. }
  274. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  275. if (DEBUG) {
  276. log("**** ENDED STAGE", step, "****");
  277. }
  278. return parallelTasksPromise;
  279. }
  280. executeTask(task) {
  281. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  282. return this.processor[task.action]();
  283. }
  284. }
  285. }
  286. // ------------
  287. // BatchRequest
  288. // ------------
  289. class BatchRequest {
  290. constructor() {
  291. this.requests = new Map();
  292. this.duplicates = new Map();
  293. }
  294. addURL(resourceURL, asBinary, expectedType, groupDuplicates) {
  295. return new Promise((resolve, reject) => {
  296. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType]);
  297. let resourceRequests = this.requests.get(requestKey);
  298. if (!resourceRequests) {
  299. resourceRequests = [];
  300. this.requests.set(requestKey, resourceRequests);
  301. }
  302. const callbacks = { resolve, reject };
  303. resourceRequests.push(callbacks);
  304. if (groupDuplicates) {
  305. let duplicateRequests = this.duplicates.get(requestKey);
  306. if (!duplicateRequests) {
  307. duplicateRequests = [];
  308. this.duplicates.set(requestKey, duplicateRequests);
  309. }
  310. duplicateRequests.push(callbacks);
  311. }
  312. });
  313. }
  314. getMaxResources() {
  315. return this.requests.size;
  316. }
  317. run(onloadListener, options) {
  318. const resourceURLs = [...this.requests.keys()];
  319. let indexResource = 0;
  320. return Promise.all(resourceURLs.map(async requestKey => {
  321. const [resourceURL, asBinary, expectedType] = JSON.parse(requestKey);
  322. const resourceRequests = this.requests.get(requestKey);
  323. try {
  324. const currentIndexResource = indexResource;
  325. indexResource = indexResource + 1;
  326. const content = await util.getContent(resourceURL, {
  327. asBinary,
  328. expectedType,
  329. maxResourceSize: options.maxResourceSize,
  330. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  331. frameId: options.windowId,
  332. resourceReferrer: options.resourceReferrer
  333. });
  334. onloadListener({ url: resourceURL });
  335. if (!this.cancelled) {
  336. resourceRequests.forEach(callbacks => {
  337. const duplicateCallbacks = this.duplicates.get(requestKey);
  338. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  339. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  340. });
  341. }
  342. } catch (error) {
  343. indexResource = indexResource + 1;
  344. onloadListener({ url: resourceURL });
  345. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  346. }
  347. this.requests.delete(requestKey);
  348. }));
  349. }
  350. cancel() {
  351. this.cancelled = true;
  352. const resourceURLs = [...this.requests.keys()];
  353. resourceURLs.forEach(requestKey => {
  354. const resourceRequests = this.requests.get(requestKey);
  355. resourceRequests.forEach(callbacks => callbacks.reject());
  356. this.requests.delete(requestKey);
  357. });
  358. }
  359. }
  360. // ---------
  361. // Processor
  362. // ---------
  363. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  364. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  365. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  366. const SCRIPT_TAG_FOUND = /<script/gi;
  367. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  368. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  369. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  370. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  371. const UTF8_CHARSET = "utf-8";
  372. class Processor {
  373. constructor(options, batchRequest) {
  374. this.options = options;
  375. this.stats = new Stats(options);
  376. this.baseURI = normalizeURL(options.baseURI || options.url);
  377. this.batchRequest = batchRequest;
  378. this.stylesheets = new Map();
  379. this.styles = new Map();
  380. this.cssVariables = new Map();
  381. this.fontTests = options.fontTests;
  382. }
  383. initialize() {
  384. this.options.saveDate = new Date();
  385. this.options.saveUrl = this.options.url;
  386. if (this.options.enableMaff) {
  387. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url));
  388. }
  389. this.maxResources = this.batchRequest.getMaxResources();
  390. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  391. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  392. }
  393. if (!this.options.removeImports && this.options.imports) {
  394. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  395. }
  396. this.stats.set("processed", "resources", this.maxResources);
  397. }
  398. async loadPage(pageContent, charset) {
  399. let content;
  400. if (!pageContent || this.options.saveRawPage) {
  401. content = await util.getContent(this.baseURI, {
  402. maxResourceSize: this.options.maxResourceSize,
  403. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  404. charset,
  405. frameId: this.options.windowId,
  406. resourceReferrer: this.options.resourceReferrer
  407. });
  408. pageContent = content.data;
  409. }
  410. this.doc = util.parseDocContent(pageContent, this.baseURI);
  411. if (this.options.saveRawPage) {
  412. let charset;
  413. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  414. const charsetDeclaration = element.content.split(";")[1];
  415. if (charsetDeclaration && !charset) {
  416. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  417. }
  418. });
  419. if (charset && charset != content.charset) {
  420. return this.loadPage(pageContent, charset);
  421. }
  422. }
  423. this.workStyleElement = this.doc.createElement("style");
  424. this.doc.body.appendChild(this.workStyleElement);
  425. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  426. }
  427. finalize() {
  428. if (this.workStyleElement.parentNode) {
  429. this.workStyleElement.remove();
  430. }
  431. }
  432. async getPageData() {
  433. util.postProcessDoc(this.doc);
  434. const url = util.parseURL(this.baseURI);
  435. if (this.options.insertSingleFileComment) {
  436. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  437. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  438. " \n url: " + this.options.saveUrl +
  439. " \n saved date: " + this.options.saveDate +
  440. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  441. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  442. }
  443. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  444. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  445. if (!canonicalLink) {
  446. canonicalLink = this.doc.createElement("link");
  447. canonicalLink.setAttribute("rel", "canonical");
  448. this.doc.head.appendChild(canonicalLink);
  449. }
  450. if (canonicalLink && !canonicalLink.href) {
  451. canonicalLink.href = this.options.saveUrl;
  452. }
  453. }
  454. if (this.options.insertMetaNoIndex) {
  455. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  456. if (!metaElement) {
  457. metaElement = this.doc.createElement("meta");
  458. metaElement.setAttribute("name", "robots");
  459. metaElement.setAttribute("content", "noindex");
  460. this.doc.head.appendChild(metaElement);
  461. }
  462. }
  463. let size;
  464. if (this.options.displayStats) {
  465. size = util.getContentSize(this.doc.documentElement.outerHTML);
  466. }
  467. const content = util.serialize(this.doc, this.options.compressHTML);
  468. if (this.options.displayStats) {
  469. const contentSize = util.getContentSize(content);
  470. this.stats.set("processed", "HTML bytes", contentSize);
  471. this.stats.add("discarded", "HTML bytes", size - contentSize);
  472. }
  473. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  474. const replacementCharacter = this.options.filenameReplacementCharacter;
  475. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  476. if (!this.options.backgroundSave) {
  477. filename = filename.replace(/\//g, replacementCharacter);
  478. }
  479. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  480. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  481. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  482. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  483. filename = filename + "…" + extension;
  484. }
  485. if (!filename) {
  486. filename = "Unnamed page";
  487. }
  488. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  489. const pageData = {
  490. stats: this.stats.data,
  491. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  492. filename,
  493. content
  494. };
  495. if (this.options.addProof) {
  496. pageData.hash = await util.digest("SHA-256", content);
  497. }
  498. if (this.options.retrieveLinks) {
  499. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  500. }
  501. return pageData;
  502. }
  503. preProcessPage() {
  504. if (this.options.win) {
  505. 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));
  506. }
  507. if (this.options.images && !this.options.saveRawPage) {
  508. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  509. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  510. if (attributeValue) {
  511. const imageData = this.options.images[Number(attributeValue)];
  512. if (imageData) {
  513. if (this.options.removeHiddenElements && (
  514. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  515. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  516. )) {
  517. imgElement.setAttribute("src", EMPTY_IMAGE);
  518. } else {
  519. if (imageData.currentSrc) {
  520. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  521. imgElement.setAttribute("src", imageData.currentSrc);
  522. }
  523. if (this.options.loadDeferredImages) {
  524. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  525. imageData.src = imgElement.dataset.src;
  526. imgElement.setAttribute("src", imgElement.dataset.src);
  527. imgElement.removeAttribute("data-src");
  528. }
  529. }
  530. }
  531. }
  532. }
  533. });
  534. if (this.options.loadDeferredImages) {
  535. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  536. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  537. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  538. imgElement.removeAttribute("data-srcset");
  539. }
  540. });
  541. }
  542. }
  543. }
  544. replaceStyleContents() {
  545. if (this.options.stylesheets) {
  546. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  547. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  548. if (attributeValue) {
  549. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  550. if (stylesheetContent) {
  551. styleElement.textContent = stylesheetContent;
  552. }
  553. }
  554. });
  555. }
  556. }
  557. removeUnselectedElements() {
  558. removeUnmarkedElements(this.doc.body);
  559. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  560. function removeUnmarkedElements(element) {
  561. let selectedElementFound = false;
  562. Array.from(element.childNodes).forEach(node => {
  563. if (node.nodeType == 1) {
  564. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  565. selectedElementFound = selectedElementFound || isSelectedElement;
  566. if (isSelectedElement) {
  567. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  568. removeUnmarkedElements(node);
  569. } else if (selectedElementFound) {
  570. removeNode(node);
  571. } else {
  572. hideNode(node);
  573. }
  574. }
  575. });
  576. }
  577. function removeNode(node) {
  578. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  579. node.remove();
  580. } else {
  581. hideNode(node);
  582. }
  583. }
  584. function hideNode(node) {
  585. if (canHideNode(node)) {
  586. node.style.setProperty("display", "none", "important");
  587. Array.from(node.childNodes).forEach(removeNode);
  588. }
  589. }
  590. function canHideNode(node) {
  591. if (node.nodeType == 1) {
  592. const tagName = node.tagName && node.tagName.toLowerCase();
  593. return (tagName != "svg" && tagName != "style" && tagName != "link");
  594. }
  595. }
  596. }
  597. insertVideoPosters() {
  598. if (this.options.posters) {
  599. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  600. let videoElement;
  601. if (element.tagName == "VIDEO") {
  602. videoElement = element;
  603. } else {
  604. videoElement = element.parentElement;
  605. }
  606. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  607. if (attributeValue) {
  608. const posterURL = this.options.posters[Number(attributeValue)];
  609. if (!videoElement.poster && posterURL) {
  610. videoElement.setAttribute("poster", posterURL);
  611. }
  612. }
  613. });
  614. }
  615. }
  616. removeFrames() {
  617. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  618. this.stats.set("discarded", "frames", frameElements.length);
  619. this.stats.set("processed", "frames", frameElements.length);
  620. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  621. }
  622. removeScripts() {
  623. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  624. this.doc.querySelectorAll("[href]").forEach(element => {
  625. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  626. element.setAttribute("href", "");
  627. }
  628. });
  629. this.doc.querySelectorAll("[src]").forEach(element => {
  630. if (element.src && element.src.match(/^\s*javascript:/)) {
  631. element.removeAttribute("src");
  632. }
  633. });
  634. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  635. this.stats.set("discarded", "scripts", scriptElements.length);
  636. this.stats.set("processed", "scripts", scriptElements.length);
  637. scriptElements.forEach(element => element.remove());
  638. }
  639. removeVideoSources() {
  640. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  641. this.stats.set("discarded", "video sources", videoSourceElements.length);
  642. this.stats.set("processed", "video sources", videoSourceElements.length);
  643. videoSourceElements.forEach(element => {
  644. if (element.tagName == "SOURCE") {
  645. element.remove();
  646. } else {
  647. videoSourceElements.forEach(element => element.removeAttribute("src"));
  648. }
  649. });
  650. }
  651. removeAudioSources() {
  652. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  653. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  654. this.stats.set("processed", "audio sources", audioSourceElements.length);
  655. audioSourceElements.forEach(element => {
  656. if (element.tagName == "SOURCE") {
  657. element.remove();
  658. } else {
  659. audioSourceElements.forEach(element => element.removeAttribute("src"));
  660. }
  661. });
  662. }
  663. removeDiscardedResources() {
  664. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  665. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  666. 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\"])");
  667. this.stats.set("discarded", "objects", objectElements.length);
  668. this.stats.set("processed", "objects", objectElements.length);
  669. objectElements.forEach(element => element.remove());
  670. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  671. replacedAttributeValue.forEach(element => {
  672. let regExp;
  673. if (this.options.removeScripts) {
  674. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  675. } else {
  676. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  677. }
  678. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  679. if (relValue.length) {
  680. element.setAttribute("rel", relValue);
  681. } else {
  682. element.remove();
  683. }
  684. });
  685. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  686. if (this.options.removeHiddenElements) {
  687. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  688. }
  689. if (!this.options.saveFavicon) {
  690. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  691. }
  692. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  693. }
  694. resetCharsetMeta() {
  695. let charset;
  696. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  697. const charsetDeclaration = element.content.split(";")[1];
  698. if (charsetDeclaration && !charset) {
  699. charset = charsetDeclaration.split("=")[1];
  700. if (charset) {
  701. this.charset = charset.trim().toLowerCase();
  702. }
  703. }
  704. element.remove();
  705. });
  706. const metaElement = this.doc.createElement("meta");
  707. metaElement.setAttribute("charset", UTF8_CHARSET);
  708. if (this.doc.head.firstChild) {
  709. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  710. } else {
  711. this.doc.head.appendChild(metaElement);
  712. }
  713. }
  714. setInputValues() {
  715. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  716. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  717. input.setAttribute("value", value || "");
  718. });
  719. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  720. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  721. if (value == "true") {
  722. input.setAttribute("checked", "");
  723. }
  724. });
  725. this.doc.querySelectorAll("textarea").forEach(textarea => {
  726. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  727. textarea.textContent = value || "";
  728. });
  729. this.doc.querySelectorAll("select").forEach(select => {
  730. select.querySelectorAll("option").forEach(option => {
  731. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  732. if (selected) {
  733. option.setAttribute("selected", "");
  734. }
  735. });
  736. });
  737. }
  738. saveFavicon() {
  739. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  740. if (!faviconElement) {
  741. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  742. }
  743. if (!faviconElement) {
  744. faviconElement = this.doc.createElement("link");
  745. faviconElement.setAttribute("type", "image/x-icon");
  746. faviconElement.setAttribute("rel", "shortcut icon");
  747. faviconElement.setAttribute("href", "/favicon.ico");
  748. }
  749. this.doc.head.appendChild(faviconElement);
  750. }
  751. replaceCanvasElements() {
  752. if (this.options.canvases) {
  753. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  754. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  755. if (attributeValue) {
  756. const canvasData = this.options.canvases[Number(attributeValue)];
  757. if (canvasData) {
  758. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  759. this.stats.add("processed", "canvas", 1);
  760. }
  761. }
  762. });
  763. }
  764. }
  765. insertFonts() {
  766. if (this.options.fonts && this.options.fonts.length) {
  767. let stylesheetContent = "";
  768. this.options.fonts.forEach(fontData => {
  769. if (fontData["font-family"] && fontData.src) {
  770. stylesheetContent += "@font-face{";
  771. let stylesContent = "";
  772. Object.keys(fontData).forEach(fontStyle => {
  773. if (stylesContent) {
  774. stylesContent += ";";
  775. }
  776. stylesContent += fontStyle + ":" + fontData[fontStyle];
  777. });
  778. stylesheetContent += stylesContent + "}";
  779. }
  780. });
  781. if (stylesheetContent) {
  782. const styleElement = this.doc.createElement("style");
  783. styleElement.textContent = stylesheetContent;
  784. const existingStyleElement = this.doc.querySelector("style");
  785. if (existingStyleElement) {
  786. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  787. } else {
  788. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  789. }
  790. }
  791. }
  792. }
  793. removeHiddenElements() {
  794. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  795. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  796. this.stats.set("discarded", "hidden elements", removedElements.length);
  797. this.stats.set("processed", "hidden elements", removedElements.length);
  798. if (hiddenElements.length) {
  799. const styleElement = this.doc.createElement("style");
  800. styleElement.textContent = ".sf-hidden{display:none!important;}";
  801. this.doc.head.appendChild(styleElement);
  802. hiddenElements.forEach(element => {
  803. if (element.style.getPropertyValue("display") != "none") {
  804. if (element.style.getPropertyPriority("display") == "important") {
  805. element.style.setProperty("display", "none", "important");
  806. } else {
  807. element.classList.add("sf-hidden");
  808. }
  809. }
  810. });
  811. }
  812. removedElements.forEach(element => element.remove());
  813. }
  814. resolveHrefs() {
  815. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  816. const href = element.getAttribute("href").trim();
  817. if (!testIgnoredPath(href)) {
  818. let resolvedURL;
  819. try {
  820. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  821. } catch (error) {
  822. // ignored
  823. }
  824. if (resolvedURL) {
  825. const url = normalizeURL(this.options.url);
  826. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  827. resolvedURL = resolvedURL.substring(url.length);
  828. }
  829. try {
  830. element.setAttribute("href", resolvedURL);
  831. } catch (error) {
  832. // ignored
  833. }
  834. }
  835. }
  836. });
  837. }
  838. resolveStyleAttributeURLs() {
  839. this.doc.querySelectorAll("[style]").forEach(element => {
  840. let styleContent = element.getAttribute("style");
  841. if (this.options.compressCSS) {
  842. styleContent = util.compressCSS(styleContent);
  843. }
  844. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement);
  845. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  846. this.styles.set(element, declarationList);
  847. });
  848. }
  849. async resolveStylesheetURLs() {
  850. const options = {
  851. maxResourceSize: this.options.maxResourceSize,
  852. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  853. url: this.options.url,
  854. charset: this.charset,
  855. compressCSS: this.options.compressCSS,
  856. updatedResources: this.options.updatedResources,
  857. rootDocument: this.options.rootDocument,
  858. frameId: this.options.windowId,
  859. resourceReferrer: this.options.resourceReferrer
  860. };
  861. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  862. .map(async element => {
  863. let mediaText;
  864. if (element.media) {
  865. mediaText = element.media.toLowerCase();
  866. }
  867. const stylesheetInfo = { mediaText };
  868. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  869. stylesheetInfo.scoped = true;
  870. }
  871. if (element.tagName.toLowerCase() == "link") {
  872. if (element.charset) {
  873. options.charset = element.charset;
  874. }
  875. }
  876. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  877. }));
  878. if (options.rootDocument) {
  879. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  880. await Promise.all(newResources.map(async resource => {
  881. resource.retrieved = true;
  882. const stylesheetInfo = {};
  883. const element = this.doc.createElement("style");
  884. this.doc.body.appendChild(element);
  885. element.textContent = resource.content;
  886. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  887. }));
  888. }
  889. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  890. stylesheets.set(element, stylesheetInfo);
  891. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  892. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  893. options.charset = getCharset(stylesheetContent);
  894. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  895. }
  896. let stylesheet;
  897. try {
  898. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  899. } catch (error) {
  900. // ignored
  901. }
  902. if (stylesheet && stylesheet.children) {
  903. if (options.compressCSS) {
  904. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  905. }
  906. stylesheetInfo.stylesheet = stylesheet;
  907. } else {
  908. stylesheets.delete(element);
  909. }
  910. }
  911. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  912. let content;
  913. if (element.tagName.toLowerCase() == "link") {
  914. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  915. } else {
  916. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  917. }
  918. return content || "";
  919. }
  920. }
  921. async resolveFrameURLs() {
  922. if (!this.options.saveRawPage) {
  923. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  924. await Promise.all(frameElements.map(async frameElement => {
  925. if (frameElement.tagName == "OBJECT") {
  926. frameElement.setAttribute("data", "data:text/html,");
  927. } else {
  928. frameElement.removeAttribute("src");
  929. frameElement.removeAttribute("srcdoc");
  930. }
  931. Array.from(frameElement.childNodes).forEach(node => node.remove());
  932. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  933. if (this.options.frames && frameWindowId) {
  934. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  935. if (frameData) {
  936. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  937. }
  938. }
  939. }));
  940. }
  941. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  942. options.insertSingleFileComment = false;
  943. options.insertCanonicalLink = false;
  944. options.insertMetaNoIndex = false;
  945. options.saveFavicon = false;
  946. options.url = frameData.baseURI;
  947. options.windowId = frameWindowId;
  948. if (frameData.content) {
  949. options.content = frameData.content;
  950. options.canvases = frameData.canvases;
  951. options.fonts = frameData.fonts;
  952. options.stylesheets = frameData.stylesheets;
  953. options.images = frameData.images;
  954. options.posters = frameData.posters;
  955. options.usedFonts = frameData.usedFonts;
  956. options.shadowRoots = frameData.shadowRoots;
  957. options.imports = frameData.imports;
  958. frameData.runner = new Runner(options);
  959. frameData.frameElement = frameElement;
  960. await frameData.runner.loadPage();
  961. await frameData.runner.initialize();
  962. frameData.maxResources = batchRequest.getMaxResources();
  963. }
  964. }
  965. }
  966. insertShadowRootContents() {
  967. const doc = this.doc;
  968. const options = this.options;
  969. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  970. processElement(this.doc);
  971. if (this.options.removeScripts) {
  972. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  973. }
  974. const scriptElement = doc.createElement("script");
  975. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  976. 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)}}})}})()`;
  977. doc.body.appendChild(scriptElement);
  978. }
  979. function processElement(element) {
  980. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  981. shadowRootElements.forEach(element => {
  982. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  983. if (attributeValue) {
  984. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  985. if (shadowRootData) {
  986. const templateElement = doc.createElement("template");
  987. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  988. if (shadowRootData.delegatesFocus) {
  989. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  990. }
  991. if (shadowRootData.adoptedStyleSheets) {
  992. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  993. const styleElement = doc.createElement("style");
  994. styleElement.textContent = stylesheetContent;
  995. templateElement.appendChild(styleElement);
  996. });
  997. }
  998. const shadowDoc = util.parseDocContent(shadowRootData.content);
  999. if (shadowDoc.head) {
  1000. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  1001. if (metaCharset) {
  1002. metaCharset.remove();
  1003. }
  1004. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1005. }
  1006. if (shadowDoc.body) {
  1007. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1008. }
  1009. processElement(templateElement);
  1010. if (element.firstChild) {
  1011. element.insertBefore(templateElement, element.firstChild);
  1012. } else {
  1013. element.appendChild(templateElement);
  1014. }
  1015. }
  1016. }
  1017. });
  1018. }
  1019. }
  1020. async resolveHtmlImportURLs() {
  1021. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1022. await Promise.all(linkElements.map(async linkElement => {
  1023. const resourceURL = linkElement.href;
  1024. linkElement.removeAttribute("href");
  1025. const options = Object.create(this.options);
  1026. options.insertSingleFileComment = false;
  1027. options.insertCanonicalLink = false;
  1028. options.insertMetaNoIndex = false;
  1029. options.saveFavicon = false;
  1030. options.removeUnusedStyles = false;
  1031. options.removeAlternativeMedias = false;
  1032. options.removeUnusedFonts = false;
  1033. options.removeHiddenElements = false;
  1034. options.url = resourceURL;
  1035. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1036. if (attributeValue) {
  1037. const importData = options.imports[Number(attributeValue)];
  1038. if (importData) {
  1039. options.content = importData.content;
  1040. importData.runner = new Runner(options);
  1041. await importData.runner.loadPage();
  1042. await importData.runner.initialize();
  1043. if (!options.removeImports) {
  1044. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1045. }
  1046. importData.runner.getStyleSheets().forEach(stylesheet => {
  1047. const importedStyleElement = this.doc.createElement("style");
  1048. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1049. this.stylesheets.set(importedStyleElement, stylesheet);
  1050. });
  1051. }
  1052. }
  1053. if (options.removeImports) {
  1054. linkElement.remove();
  1055. this.stats.add("discarded", "HTML imports", 1);
  1056. }
  1057. }));
  1058. }
  1059. removeUnusedStyles() {
  1060. if (!this.mediaAllInfo) {
  1061. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1062. }
  1063. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1064. this.stats.set("processed", "CSS rules", stats.processed);
  1065. this.stats.set("discarded", "CSS rules", stats.discarded);
  1066. }
  1067. removeUnusedFonts() {
  1068. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1069. }
  1070. removeAlternativeMedias() {
  1071. const stats = util.minifyMedias(this.stylesheets);
  1072. this.stats.set("processed", "medias", stats.processed);
  1073. this.stats.set("discarded", "medias", stats.discarded);
  1074. }
  1075. async processStylesheets() {
  1076. this.options.fontURLs = new Map();
  1077. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1078. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1079. ));
  1080. }
  1081. async processStyleAttributes() {
  1082. return Promise.all([...this.styles].map(([, declarationList]) =>
  1083. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1084. ));
  1085. }
  1086. async processPageResources() {
  1087. const processAttributeArgs = [
  1088. ["link[href][rel*=\"icon\"]", "href", false, true],
  1089. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1090. ["img[src], input[src][type=image]", "src", true],
  1091. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1092. ["video[poster]", "poster"],
  1093. ["*[background]", "background"],
  1094. ["image", "xlink:href"],
  1095. ["image", "href"]
  1096. ];
  1097. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1098. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1099. );
  1100. resourcePromises = resourcePromises.concat([
  1101. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1102. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1103. ]);
  1104. if (!this.options.removeAudioSrc) {
  1105. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1106. }
  1107. if (!this.options.removeVideoSrc) {
  1108. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1109. }
  1110. await Promise.all(resourcePromises);
  1111. if (this.options.saveFavicon) {
  1112. ProcessorHelper.processShortcutIcons(this.doc);
  1113. }
  1114. }
  1115. async processScripts() {
  1116. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1117. let resourceURL;
  1118. let scriptSrc;
  1119. if (element.tagName == "SCRIPT") {
  1120. scriptSrc = element.getAttribute("src");
  1121. element.removeAttribute("src");
  1122. } else {
  1123. scriptSrc = element.getAttribute("href");
  1124. element.removeAttribute("href");
  1125. }
  1126. element.removeAttribute("integrity");
  1127. element.textContent = "";
  1128. try {
  1129. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1130. } catch (error) {
  1131. // ignored
  1132. }
  1133. if (testValidURL(resourceURL)) {
  1134. this.stats.add("processed", "scripts", 1);
  1135. const content = await util.getContent(resourceURL, {
  1136. asBinary: true,
  1137. charset: this.charset != UTF8_CHARSET && this.charset,
  1138. maxResourceSize: this.options.maxResourceSize,
  1139. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1140. frameId: this.options.windowId,
  1141. resourceReferrer: this.options.resourceReferrer
  1142. });
  1143. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1144. if (element.tagName == "SCRIPT") {
  1145. element.setAttribute("src", content.data);
  1146. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1147. element.setAttribute("async", "");
  1148. }
  1149. } else {
  1150. const scriptElement = this.doc.createElement("script");
  1151. scriptElement.setAttribute("src", content.data);
  1152. scriptElement.setAttribute("async", "");
  1153. element.parentElement.replaceChild(scriptElement, element);
  1154. }
  1155. }
  1156. }));
  1157. }
  1158. removeAlternativeImages() {
  1159. util.removeAlternativeImages(this.doc);
  1160. }
  1161. async removeAlternativeFonts() {
  1162. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1163. }
  1164. async processFrames() {
  1165. if (this.options.frames) {
  1166. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1167. await Promise.all(frameElements.map(async frameElement => {
  1168. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1169. if (frameWindowId) {
  1170. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1171. if (frameData) {
  1172. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1173. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1174. this.stats.add("processed", "frames", 1);
  1175. await frameData.runner.run();
  1176. const pageData = await frameData.runner.getPageData();
  1177. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1178. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1179. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1180. sandbox += " allow-scripts allow-same-origin";
  1181. }
  1182. frameElement.setAttribute("sandbox", sandbox);
  1183. if (frameElement.tagName == "OBJECT") {
  1184. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1185. } else {
  1186. if (frameElement.tagName == "FRAME") {
  1187. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1188. } else {
  1189. frameElement.setAttribute("srcdoc", pageData.content);
  1190. frameElement.removeAttribute("src");
  1191. }
  1192. }
  1193. this.stats.addAll(pageData);
  1194. } else {
  1195. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1196. this.stats.add("discarded", "frames", 1);
  1197. }
  1198. }
  1199. }
  1200. }));
  1201. }
  1202. }
  1203. async processHtmlImports() {
  1204. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1205. await Promise.all(linkElements.map(async linkElement => {
  1206. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1207. if (attributeValue) {
  1208. const importData = this.options.imports[Number(attributeValue)];
  1209. if (importData.runner) {
  1210. this.stats.add("processed", "HTML imports", 1);
  1211. await importData.runner.run();
  1212. const pageData = await importData.runner.getPageData();
  1213. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1214. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1215. this.stats.addAll(pageData);
  1216. } else {
  1217. this.stats.add("discarded", "HTML imports", 1);
  1218. }
  1219. }
  1220. }));
  1221. }
  1222. replaceStylesheets() {
  1223. this.doc.querySelectorAll("style").forEach(styleElement => {
  1224. const stylesheetInfo = this.stylesheets.get(styleElement);
  1225. if (stylesheetInfo) {
  1226. this.stylesheets.delete(styleElement);
  1227. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1228. styleElement.textContent = stylesheetContent;
  1229. if (stylesheetInfo.mediaText) {
  1230. styleElement.media = stylesheetInfo.mediaText;
  1231. }
  1232. } else {
  1233. styleElement.remove();
  1234. }
  1235. });
  1236. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1237. const stylesheetInfo = this.stylesheets.get(linkElement);
  1238. if (stylesheetInfo) {
  1239. this.stylesheets.delete(linkElement);
  1240. const styleElement = this.doc.createElement("style");
  1241. if (stylesheetInfo.mediaText) {
  1242. styleElement.media = stylesheetInfo.mediaText;
  1243. }
  1244. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1245. styleElement.textContent = stylesheetContent;
  1246. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1247. } else {
  1248. linkElement.remove();
  1249. }
  1250. });
  1251. }
  1252. replaceStyleAttributes() {
  1253. this.doc.querySelectorAll("[style]").forEach(element => {
  1254. const declarations = this.styles.get(element);
  1255. if (declarations) {
  1256. this.styles.delete(element);
  1257. let styleContent = cssTree.generate(declarations);
  1258. element.setAttribute("style", styleContent);
  1259. } else {
  1260. element.setAttribute("style", "");
  1261. }
  1262. });
  1263. }
  1264. insertVariables() {
  1265. if (this.cssVariables.size) {
  1266. const styleElement = this.doc.createElement("style");
  1267. const firstStyleElement = this.doc.head.querySelector("style");
  1268. if (firstStyleElement) {
  1269. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1270. } else {
  1271. this.doc.head.appendChild(styleElement);
  1272. }
  1273. let stylesheetContent = "";
  1274. this.cssVariables.forEach((content, indexResource) => {
  1275. this.cssVariables.delete(indexResource);
  1276. if (stylesheetContent) {
  1277. stylesheetContent += ";";
  1278. }
  1279. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1280. });
  1281. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1282. }
  1283. }
  1284. compressHTML() {
  1285. let size;
  1286. if (this.options.displayStats) {
  1287. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1288. }
  1289. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1290. if (this.options.displayStats) {
  1291. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1292. }
  1293. }
  1294. cleanupPage() {
  1295. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1296. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1297. if (metaCharset) {
  1298. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1299. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1300. this.doc.head.querySelector("meta[charset]").remove();
  1301. }
  1302. }
  1303. }
  1304. resetZoomLevel() {
  1305. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1306. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1307. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1308. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1309. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1310. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1311. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1312. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1313. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1314. this.doc.documentElement.style.removeProperty("-sf-transform");
  1315. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1316. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1317. }
  1318. async insertMAFFMetaData() {
  1319. const maffMetaData = await this.maffMetaDataPromise;
  1320. if (maffMetaData && maffMetaData.content) {
  1321. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1322. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1323. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1324. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1325. if (originalURLElement) {
  1326. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1327. }
  1328. if (archiveTimeElement) {
  1329. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1330. if (value) {
  1331. const date = new Date(value);
  1332. if (!isNaN(date.getTime())) {
  1333. this.options.saveDate = new Date(value);
  1334. }
  1335. }
  1336. }
  1337. }
  1338. }
  1339. async setDocInfo() {
  1340. const titleElement = this.doc.querySelector("title");
  1341. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1342. const authorElement = this.doc.querySelector("meta[name=author]");
  1343. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1344. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1345. const headingElement = this.doc.querySelector("h1");
  1346. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1347. this.options.info = {
  1348. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1349. lang: this.doc.documentElement.lang,
  1350. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1351. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1352. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1353. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1354. };
  1355. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1356. }
  1357. }
  1358. // ---------------
  1359. // ProcessorHelper
  1360. // ---------------
  1361. const DATA_URI_PREFIX = "data:";
  1362. const ABOUT_BLANK_URI = "about:blank";
  1363. const EMPTY_DATA_URI = "data:null;base64,";
  1364. const REGEXP_URL_HASH = /(#.+?)$/;
  1365. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1366. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1367. class ProcessorHelper {
  1368. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1369. const date = options.saveDate;
  1370. const url = util.parseURL(options.saveUrl);
  1371. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1376. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1377. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1387. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1388. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1397. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1398. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1401. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1402. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1403. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1404. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1405. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1406. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1407. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1408. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1409. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1410. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1411. for (const [name, value] of params) {
  1412. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1413. }
  1414. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1415. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1416. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1417. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1418. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1419. if (content) {
  1420. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1421. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1422. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1423. }
  1424. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1425. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1426. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1427. return template.trim();
  1428. function decode(value) {
  1429. try {
  1430. return decodeURI(value);
  1431. } catch (error) {
  1432. return value;
  1433. }
  1434. }
  1435. }
  1436. static setBackgroundImage(element, url, style) {
  1437. element.style.setProperty("background-blend-mode", "normal", "important");
  1438. element.style.setProperty("background-clip", "content-box", "important");
  1439. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1440. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1441. element.style.setProperty("background-image", url, "important");
  1442. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1443. element.style.setProperty("background-origin", "content-box", "important");
  1444. element.style.setProperty("background-repeat", "no-repeat", "important");
  1445. }
  1446. static processShortcutIcons(doc) {
  1447. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1448. if (!shortcutIcon) {
  1449. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1450. if (shortcutIcon) {
  1451. shortcutIcon.rel = "icon";
  1452. }
  1453. }
  1454. if (shortcutIcon) {
  1455. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1456. if (linkElement != shortcutIcon) {
  1457. linkElement.remove();
  1458. }
  1459. });
  1460. }
  1461. }
  1462. static removeSingleLineCssComments(stylesheet) {
  1463. const removedRules = [];
  1464. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1465. const ruleData = cssRule.data;
  1466. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1467. removedRules.push(cssRule);
  1468. }
  1469. }
  1470. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1471. }
  1472. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1473. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet);
  1474. const imports = getImportFunctions(stylesheetContent);
  1475. await Promise.all(imports.map(async cssImport => {
  1476. const match = matchImport(cssImport);
  1477. if (match) {
  1478. const regExpCssImport = getRegExp(cssImport);
  1479. let resourceURL = normalizeURL(match.resourceURL);
  1480. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1481. try {
  1482. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1483. } catch (error) {
  1484. // ignored
  1485. }
  1486. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1487. const content = await util.getContent(resourceURL, {
  1488. maxResourceSize: options.maxResourceSize,
  1489. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1490. validateTextContentType: true,
  1491. frameId: options.frameId,
  1492. resourceReferrer: options.resourceReferrer
  1493. });
  1494. resourceURL = content.resourceURL;
  1495. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1496. let importedStylesheetContent = removeCssComments(content.data);
  1497. if (options.compressCSS) {
  1498. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1499. }
  1500. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1501. if (stylesheetContent.includes(cssImport)) {
  1502. const ancestorStyleSheets = new Set(importedStyleSheets);
  1503. ancestorStyleSheets.add(resourceURL);
  1504. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1505. workStylesheet.textContent = importedStylesheetContent;
  1506. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1507. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1508. } else {
  1509. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1510. }
  1511. }
  1512. } else {
  1513. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1514. }
  1515. } else {
  1516. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1517. }
  1518. }
  1519. }));
  1520. return stylesheetContent;
  1521. }
  1522. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet) {
  1523. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1524. urlFunctions.map(urlFunction => {
  1525. const originalResourceURL = matchURL(urlFunction);
  1526. let resourceURL = normalizeURL(originalResourceURL);
  1527. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1528. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1529. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1530. }
  1531. if (!testIgnoredPath(resourceURL)) {
  1532. if (!resourceURL || testValidPath(resourceURL)) {
  1533. let resolvedURL;
  1534. if (!originalResourceURL.startsWith("#")) {
  1535. try {
  1536. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1537. } catch (error) {
  1538. // ignored
  1539. }
  1540. }
  1541. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1542. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1543. }
  1544. } else {
  1545. let newUrlFunction;
  1546. if (originalResourceURL) {
  1547. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1548. } else {
  1549. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1550. }
  1551. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1552. }
  1553. }
  1554. });
  1555. return stylesheetContent;
  1556. }
  1557. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1558. resourceURL = normalizeURL(resourceURL);
  1559. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1560. const content = await util.getContent(resourceURL, {
  1561. maxResourceSize: options.maxResourceSize,
  1562. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1563. charset: options.charset,
  1564. frameId: options.frameId,
  1565. resourceReferrer: options.resourceReferrer,
  1566. validateTextContentType: true
  1567. });
  1568. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1569. options.charset = getCharset(content.data);
  1570. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1571. }
  1572. resourceURL = content.resourceURL;
  1573. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1574. let stylesheetContent = removeCssComments(content.data);
  1575. if (options.compressCSS) {
  1576. stylesheetContent = util.compressCSS(stylesheetContent);
  1577. }
  1578. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1579. return stylesheetContent;
  1580. }
  1581. }
  1582. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1583. const promises = [];
  1584. const removedRules = [];
  1585. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1586. const ruleData = cssRule.data;
  1587. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1588. removedRules.push(cssRule);
  1589. } else if (ruleData.block && ruleData.block.children) {
  1590. if (ruleData.type == "Rule") {
  1591. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1592. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1593. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1594. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1595. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1596. }
  1597. }
  1598. }
  1599. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1600. await Promise.all(promises);
  1601. async function processFontFaceRule(ruleData, fontURLs) {
  1602. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1603. if (declaration.type == "Declaration" && declaration.value.children) {
  1604. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1605. await Promise.all(urlFunctions.map(async urlFunction => {
  1606. const originalResourceURL = matchURL(urlFunction);
  1607. const resourceURL = normalizeURL(originalResourceURL);
  1608. if (!testIgnoredPath(resourceURL)) {
  1609. if (testValidURL(resourceURL)) {
  1610. let { content } = await batchRequest.addURL(resourceURL, true, "font");
  1611. let resourceURLs = fontURLs.get(declaration);
  1612. if (!resourceURLs) {
  1613. resourceURLs = [];
  1614. fontURLs.set(declaration, resourceURLs);
  1615. }
  1616. resourceURLs.push(resourceURL);
  1617. replaceURLs(declaration, originalResourceURL, content);
  1618. }
  1619. }
  1620. }));
  1621. }
  1622. }));
  1623. function replaceURLs(declaration, oldURL, newURL) {
  1624. declaration.value.children.forEach(token => {
  1625. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1626. token.value.value = newURL;
  1627. }
  1628. });
  1629. }
  1630. }
  1631. }
  1632. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1633. await Promise.all(declarations.map(async declaration => {
  1634. let children;
  1635. if (declaration.value) {
  1636. if (declaration.value.children) {
  1637. children = declaration.value.children;
  1638. } else if (declaration.value.type == "Raw") {
  1639. try {
  1640. children = cssTree.parse(declaration.value.value, { context: "value" }).children;
  1641. } catch (error) {
  1642. // ignored
  1643. }
  1644. }
  1645. }
  1646. if (declaration.type == "Declaration" && children) {
  1647. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1648. await Promise.all(urlFunctions.map(async urlFunction => {
  1649. const originalResourceURL = matchURL(urlFunction);
  1650. const resourceURL = normalizeURL(originalResourceURL);
  1651. if (!testIgnoredPath(resourceURL)) {
  1652. if (testValidURL(resourceURL)) {
  1653. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", true);
  1654. let variableDefined;
  1655. const tokens = [];
  1656. findURLToken(originalResourceURL, children, (token, parent, rootFunction) => {
  1657. if (!originalResourceURL.startsWith("#")) {
  1658. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1659. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1660. tokens.push({ parent, token, value });
  1661. variableDefined = true;
  1662. } else {
  1663. token.data.value.value = content;
  1664. }
  1665. }
  1666. });
  1667. if (variableDefined) {
  1668. cssVariables.set(indexResource, content);
  1669. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1670. }
  1671. }
  1672. }
  1673. }));
  1674. }
  1675. }));
  1676. function findURLToken(url, children, callback, depth = 0) {
  1677. for (let token = children.head; token; token = token.next) {
  1678. if (token.data.children) {
  1679. findURLToken(url, token.data.children, callback, depth + 1);
  1680. }
  1681. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1682. callback(token, children, depth == 0);
  1683. }
  1684. }
  1685. }
  1686. }
  1687. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1688. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1689. let resourceURL = resourceElement.getAttribute(attributeName);
  1690. if (resourceURL != null) {
  1691. resourceURL = normalizeURL(resourceURL);
  1692. let originURL = resourceElement.dataset.singleFileOriginURL;
  1693. delete resourceElement.dataset.singleFileOriginURL;
  1694. if (!testIgnoredPath(resourceURL)) {
  1695. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1696. if (testValidPath(resourceURL)) {
  1697. try {
  1698. resourceURL = util.resolveURL(resourceURL, baseURI);
  1699. } catch (error) {
  1700. // ignored
  1701. }
  1702. if (testValidURL(resourceURL)) {
  1703. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", resourceElement.tagName == "IMG" && attributeName == "src");
  1704. if (originURL) {
  1705. if (content == EMPTY_DATA_URI) {
  1706. try {
  1707. originURL = util.resolveURL(originURL, baseURI);
  1708. } catch (error) {
  1709. // ignored
  1710. }
  1711. try {
  1712. resourceURL = originURL;
  1713. content = (await util.getContent(resourceURL, {
  1714. asBinary: true,
  1715. expectedType: "image",
  1716. maxResourceSize: options.maxResourceSize,
  1717. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1718. frameId: options.windowId,
  1719. resourceReferrer: options.resourceReferrer
  1720. })).data;
  1721. } catch (error) {
  1722. // ignored
  1723. }
  1724. }
  1725. }
  1726. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1727. resourceElement.remove();
  1728. } else {
  1729. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1730. if (!forbiddenPrefixFound) {
  1731. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1732. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1733. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1734. cssVariables.set(indexResource, content);
  1735. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1736. styles.set(resourceElement, declarationList);
  1737. } else {
  1738. resourceElement.setAttribute(attributeName, content);
  1739. }
  1740. } else {
  1741. resourceElement.setAttribute(attributeName, content);
  1742. }
  1743. }
  1744. }
  1745. }
  1746. }
  1747. }
  1748. }
  1749. }));
  1750. }
  1751. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1752. let attributeName = "xlink:href";
  1753. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1754. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1755. if (originalResourceURL == null) {
  1756. attributeName = "href";
  1757. originalResourceURL = resourceElement.getAttribute(attributeName);
  1758. }
  1759. let resourceURL = normalizeURL(originalResourceURL);
  1760. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1761. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1762. try {
  1763. resourceURL = util.resolveURL(resourceURL, baseURI);
  1764. } catch (error) {
  1765. // ignored
  1766. }
  1767. if (testValidURL(resourceURL)) {
  1768. const { content } = await batchRequest.addURL(resourceURL);
  1769. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1770. if (hashMatch && hashMatch[0]) {
  1771. let symbolElement;
  1772. try {
  1773. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1774. } catch (error) {
  1775. // ignored
  1776. }
  1777. if (symbolElement) {
  1778. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1779. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1780. }
  1781. } else {
  1782. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1783. }
  1784. }
  1785. } else if (resourceURL == options.url) {
  1786. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1787. }
  1788. }));
  1789. }
  1790. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1791. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1792. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1793. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1794. let resourceURL = normalizeURL(srcsetValue.url);
  1795. if (!testIgnoredPath(resourceURL)) {
  1796. if (testValidPath(resourceURL)) {
  1797. try {
  1798. resourceURL = util.resolveURL(resourceURL, baseURI);
  1799. } catch (error) {
  1800. // ignored
  1801. }
  1802. if (testValidURL(resourceURL)) {
  1803. const { content } = await batchRequest.addURL(resourceURL, true, "image");
  1804. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1805. if (forbiddenPrefixFound) {
  1806. return "";
  1807. }
  1808. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1809. } else {
  1810. return "";
  1811. }
  1812. } else {
  1813. return "";
  1814. }
  1815. } else {
  1816. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1817. }
  1818. }));
  1819. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1820. }));
  1821. }
  1822. static replaceImageSource(imgElement, variableName, options) {
  1823. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1824. if (attributeValue) {
  1825. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1826. if (imageData && imageData.replaceable) {
  1827. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1828. const backgroundStyle = {};
  1829. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1830. if (backgroundSize) {
  1831. backgroundStyle["background-size"] = imageData.objectFit;
  1832. }
  1833. if (imageData.objectPosition) {
  1834. backgroundStyle["background-position"] = imageData.objectPosition;
  1835. }
  1836. if (imageData.backgroundColor) {
  1837. backgroundStyle["background-color"] = imageData.backgroundColor;
  1838. }
  1839. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1840. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1841. return true;
  1842. }
  1843. }
  1844. }
  1845. }
  1846. // ----
  1847. // Util
  1848. // ----
  1849. const BLOB_URI_PREFIX = "blob:";
  1850. const HTTP_URI_PREFIX = /^https?:\/\//;
  1851. const FILE_URI_PREFIX = /^file:\/\//;
  1852. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1853. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1854. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1855. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1856. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1857. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1858. 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;
  1859. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1860. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1861. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1862. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1863. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1864. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1865. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1866. function getUpdatedResourceContent(resourceURL, content, options) {
  1867. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1868. options.updatedResources[resourceURL].retrieved = true;
  1869. return options.updatedResources[resourceURL].content;
  1870. } else {
  1871. return content.data || "";
  1872. }
  1873. }
  1874. function normalizeURL(url) {
  1875. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1876. return url;
  1877. } else {
  1878. return url.split("#")[0];
  1879. }
  1880. }
  1881. function getCSSValue(value) {
  1882. let result = "";
  1883. try {
  1884. result = cssTree.generate(value);
  1885. } catch (error) {
  1886. // ignored
  1887. }
  1888. return result;
  1889. }
  1890. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  1891. const stylesheetCharset = getCharset(stylesheetContent);
  1892. if (stylesheetCharset) {
  1893. return stylesheetCharset == charset.toLowerCase();
  1894. } else {
  1895. return true;
  1896. }
  1897. }
  1898. function getCharset(stylesheetContent) {
  1899. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  1900. if (match && match[1]) {
  1901. return match[1].toLowerCase().trim();
  1902. }
  1903. }
  1904. function getOnEventAttributeNames(doc) {
  1905. const element = doc.createElement("div");
  1906. const attributeNames = [];
  1907. for (const propertyName in element) {
  1908. if (propertyName.startsWith("on")) {
  1909. attributeNames.push(propertyName);
  1910. }
  1911. }
  1912. attributeNames.push("onunload");
  1913. return attributeNames;
  1914. }
  1915. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1916. let maxLength;
  1917. if (template) {
  1918. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1919. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1920. if (template.match(replaceRegExp)) {
  1921. const matchedLength = template.match(replaceRegExp)[0];
  1922. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1923. if (isNaN(maxLength) || maxLength <= 0) {
  1924. maxLength = null;
  1925. }
  1926. } else {
  1927. replaceRegExp = new RegExp(regExpVariable, "g");
  1928. }
  1929. if (template.match(replaceRegExp)) {
  1930. let value = await valueGetter();
  1931. if (!dontReplaceSlash) {
  1932. value = value.replace(/\/+/g, replacementCharacter);
  1933. }
  1934. if (maxLength) {
  1935. value = await util.truncateText(value, maxLength);
  1936. }
  1937. return template.replace(replaceRegExp, value);
  1938. }
  1939. }
  1940. return template;
  1941. }
  1942. function getLastSegment(url, replacementCharacter) {
  1943. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1944. if (!lastSegment) {
  1945. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1946. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1947. }
  1948. if (!lastSegment) {
  1949. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1950. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1951. }
  1952. if (!lastSegment) {
  1953. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1954. }
  1955. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1956. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1957. lastSegment = lastSegmentMatch[1];
  1958. }
  1959. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1960. return lastSegment;
  1961. }
  1962. function getRegExp(string) {
  1963. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1964. }
  1965. function getUrlFunctions(stylesheetContent, unique) {
  1966. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1967. if (unique) {
  1968. return [...new Set(result)];
  1969. } else {
  1970. return result;
  1971. }
  1972. }
  1973. function getImportFunctions(stylesheetContent) {
  1974. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1975. }
  1976. function findShortcutIcon(shortcutIcons) {
  1977. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1978. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1979. return shortcutIcons[0];
  1980. }
  1981. function matchURL(stylesheetContent) {
  1982. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1983. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1984. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1985. return match && match[1];
  1986. }
  1987. function testIgnoredPath(resourceURL) {
  1988. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1989. }
  1990. function testValidPath(resourceURL) {
  1991. return resourceURL && !resourceURL.match(EMPTY_URL);
  1992. }
  1993. function testValidURL(resourceURL) {
  1994. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1995. }
  1996. function matchImport(stylesheetContent) {
  1997. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1998. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1999. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2000. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2001. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2002. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2003. if (match) {
  2004. const [, resourceURL, media] = match;
  2005. return { resourceURL, media };
  2006. }
  2007. }
  2008. function removeCssComments(stylesheetContent) {
  2009. try {
  2010. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2011. } catch (error) {
  2012. let start, end;
  2013. do {
  2014. start = stylesheetContent.indexOf("/*");
  2015. end = stylesheetContent.indexOf("*/", start + 2);
  2016. if (start != -1 && end != -1) {
  2017. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2018. }
  2019. } while (start != -1 && end != -1);
  2020. return stylesheetContent;
  2021. }
  2022. }
  2023. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2024. if (mediaQuery) {
  2025. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2026. } else {
  2027. return stylesheetContent;
  2028. }
  2029. }
  2030. function log(...args) {
  2031. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2032. }
  2033. // -----
  2034. // Stats
  2035. // -----
  2036. const STATS_DEFAULT_VALUES = {
  2037. discarded: {
  2038. "HTML bytes": 0,
  2039. "hidden elements": 0,
  2040. "HTML imports": 0,
  2041. scripts: 0,
  2042. objects: 0,
  2043. "audio sources": 0,
  2044. "video sources": 0,
  2045. frames: 0,
  2046. "CSS rules": 0,
  2047. canvas: 0,
  2048. stylesheets: 0,
  2049. resources: 0,
  2050. medias: 0
  2051. },
  2052. processed: {
  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. };
  2068. class Stats {
  2069. constructor(options) {
  2070. this.options = options;
  2071. if (options.displayStats) {
  2072. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2073. }
  2074. }
  2075. set(type, subType, value) {
  2076. if (this.options.displayStats) {
  2077. this.data[type][subType] = value;
  2078. }
  2079. }
  2080. add(type, subType, value) {
  2081. if (this.options.displayStats) {
  2082. this.data[type][subType] += value;
  2083. }
  2084. }
  2085. addAll(pageData) {
  2086. if (this.options.displayStats) {
  2087. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2088. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2089. }
  2090. }
  2091. }
  2092. return { getClass };
  2093. })();