single-file-core.js 85 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree, sessionId = 0;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. if (options.sessionId === undefined) {
  34. options.sessionId = sessionId;
  35. sessionId++;
  36. }
  37. }
  38. async run() {
  39. if (this.options.userScriptEnabled) {
  40. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  41. }
  42. this.runner = new Runner(this.options, true);
  43. await this.runner.loadPage();
  44. await this.runner.initialize();
  45. if (this.options.userScriptEnabled) {
  46. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  47. }
  48. await this.runner.run();
  49. }
  50. cancel() {
  51. this.cancelled = true;
  52. if (this.runner) {
  53. this.runner.cancel();
  54. }
  55. }
  56. async getPageData() {
  57. return this.runner.getPageData();
  58. }
  59. }
  60. // -------------
  61. // ProgressEvent
  62. // -------------
  63. const PAGE_LOADING = "page-loading";
  64. const PAGE_LOADED = "page-loaded";
  65. const RESOURCES_INITIALIZING = "resource-initializing";
  66. const RESOURCES_INITIALIZED = "resources-initialized";
  67. const RESOURCE_LOADED = "resource-loaded";
  68. const PAGE_ENDED = "page-ended";
  69. const STAGE_STARTED = "stage-started";
  70. const STAGE_ENDED = "stage-ended";
  71. const STAGE_TASK_STARTED = "stage-task-started";
  72. const STAGE_TASK_ENDED = "stage-task-ended";
  73. class ProgressEvent {
  74. constructor(type, detail) {
  75. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  76. }
  77. }
  78. // ------
  79. // Runner
  80. // ------
  81. const RESOLVE_URLS_STAGE = 0;
  82. const REPLACE_DATA_STAGE = 1;
  83. const REPLACE_DOCS_STAGE = 2;
  84. const POST_PROCESS_STAGE = 3;
  85. const STAGES = [{
  86. sequential: [
  87. { action: "preProcessPage" },
  88. { action: "replaceStyleContents" },
  89. { action: "resetCharsetMeta" },
  90. { option: "saveFavicon", action: "saveFavicon" },
  91. { action: "replaceCanvasElements" },
  92. { action: "insertFonts" },
  93. { action: "insertShadowRootContents" },
  94. { action: "setInputValues" },
  95. { option: "removeScripts", action: "removeScripts" },
  96. { option: "selected", action: "removeUnselectedElements" },
  97. { option: "removeVideoSrc", action: "insertVideoPosters" },
  98. { option: "removeFrames", action: "removeFrames" },
  99. { option: "removeVideoSrc", action: "removeVideoSources" },
  100. { option: "removeAudioSrc", action: "removeAudioSources" },
  101. { action: "removeDiscardedResources" },
  102. { option: "removeHiddenElements", action: "removeHiddenElements" },
  103. { action: "resolveHrefs" },
  104. { action: "resolveStyleAttributeURLs" }
  105. ],
  106. parallel: [
  107. { action: "resolveStylesheetURLs" },
  108. { option: "!removeFrames", action: "resolveFrameURLs" },
  109. { action: "resolveHtmlImportURLs" }
  110. ]
  111. }, {
  112. sequential: [
  113. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  114. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  115. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  116. ],
  117. parallel: [
  118. { action: "processStylesheets" },
  119. { action: "processStyleAttributes" },
  120. { action: "processPageResources" },
  121. { option: "!removeScripts", action: "processScripts" }
  122. ]
  123. }, {
  124. sequential: [
  125. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  126. ],
  127. parallel: [
  128. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  129. { option: "!removeFrames", action: "processFrames" },
  130. { option: "!removeImports", action: "processHtmlImports" },
  131. ]
  132. }, {
  133. sequential: [
  134. { action: "replaceStylesheets" },
  135. { action: "replaceStyleAttributes" },
  136. { action: "insertVariables" },
  137. { option: "compressHTML", action: "compressHTML" },
  138. { action: "cleanupPage" }
  139. ],
  140. parallel: [
  141. { option: "enableMaff", action: "insertMAFFMetaData" },
  142. { action: "setDocInfo" }
  143. ]
  144. }];
  145. class Runner {
  146. constructor(options, root) {
  147. this.root = root;
  148. this.options = options;
  149. this.options.url = this.options.url || (this.options.doc && this.options.doc.location.href);
  150. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  151. this.options.rootDocument = root;
  152. this.options.updatedResources = this.options.updatedResources || {};
  153. this.batchRequest = new BatchRequest();
  154. this.processor = new Processor(options, this.batchRequest);
  155. if (this.options.doc) {
  156. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  157. this.options.canvases = docData.canvases;
  158. this.options.fonts = docData.fonts;
  159. this.options.stylesheets = docData.stylesheets;
  160. this.options.images = docData.images;
  161. this.options.posters = docData.posters;
  162. this.options.usedFonts = docData.usedFonts;
  163. this.options.shadowRoots = docData.shadowRoots;
  164. this.options.imports = docData.imports;
  165. this.options.referrer = docData.referrer;
  166. this.markedElements = docData.markedElements;
  167. }
  168. if (this.options.saveRawPage) {
  169. this.options.removeFrames = true;
  170. }
  171. this.options.content = this.options.content || (this.options.doc ? util.serialize(this.options.doc) : null);
  172. this.onprogress = options.onprogress || (() => { });
  173. }
  174. async loadPage() {
  175. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  176. await this.processor.loadPage(this.options.content);
  177. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  178. }
  179. async initialize() {
  180. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  181. await this.executeStage(RESOLVE_URLS_STAGE);
  182. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  183. if (this.options.doc) {
  184. util.postProcessDoc(this.options.doc, this.markedElements);
  185. }
  186. this.options.doc = null;
  187. this.options.win = null;
  188. }
  189. cancel() {
  190. this.cancelled = true;
  191. this.batchRequest.cancel();
  192. if (this.root) {
  193. if (this.options.frames) {
  194. this.options.frames.forEach(cancelRunner);
  195. }
  196. if (this.options.imports) {
  197. this.options.imports.forEach(cancelRunner);
  198. }
  199. }
  200. function cancelRunner(resourceData) {
  201. if (resourceData.runner) {
  202. resourceData.runner.cancel();
  203. }
  204. }
  205. }
  206. async run() {
  207. if (this.root) {
  208. this.processor.initialize(this.batchRequest);
  209. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  210. }
  211. await this.batchRequest.run(detail => {
  212. detail.pageURL = this.options.url;
  213. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  214. }, this.options);
  215. await this.pendingPromises;
  216. await this.executeStage(REPLACE_DOCS_STAGE);
  217. await this.executeStage(POST_PROCESS_STAGE);
  218. this.processor.finalize();
  219. }
  220. getDocument() {
  221. return this.processor.doc;
  222. }
  223. getStyleSheets() {
  224. return this.processor.stylesheets;
  225. }
  226. async getPageData() {
  227. if (this.root) {
  228. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  229. }
  230. return this.processor.getPageData();
  231. }
  232. async executeStage(step) {
  233. if (DEBUG) {
  234. log("**** STARTED STAGE", step, "****");
  235. }
  236. const frame = !this.root;
  237. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  238. STAGES[step].sequential.forEach(task => {
  239. let startTime;
  240. if (DEBUG) {
  241. startTime = Date.now();
  242. log(" -- STARTED task =", task.action);
  243. }
  244. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  245. if (!this.cancelled) {
  246. this.executeTask(task);
  247. }
  248. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  249. if (DEBUG) {
  250. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  251. }
  252. });
  253. let parallelTasksPromise;
  254. if (STAGES[step].parallel) {
  255. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  256. let startTime;
  257. if (DEBUG) {
  258. startTime = Date.now();
  259. log(" // STARTED task =", task.action);
  260. }
  261. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  262. if (!this.cancelled) {
  263. await this.executeTask(task);
  264. }
  265. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  266. if (DEBUG) {
  267. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  268. }
  269. }));
  270. } else {
  271. parallelTasksPromise = Promise.resolve();
  272. }
  273. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  274. if (DEBUG) {
  275. log("**** ENDED STAGE", step, "****");
  276. }
  277. return parallelTasksPromise;
  278. }
  279. executeTask(task) {
  280. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  281. return this.processor[task.action]();
  282. }
  283. }
  284. }
  285. // ------------
  286. // BatchRequest
  287. // ------------
  288. class BatchRequest {
  289. constructor() {
  290. this.requests = new Map();
  291. this.duplicates = new Map();
  292. }
  293. async addURL(resourceURL, asBinary, groupDuplicates) {
  294. return new Promise((resolve, reject) => {
  295. const requestKey = JSON.stringify([resourceURL, asBinary]);
  296. let resourceRequests = this.requests.get(requestKey);
  297. if (!resourceRequests) {
  298. resourceRequests = [];
  299. this.requests.set(requestKey, resourceRequests);
  300. }
  301. const callbacks = { resolve, reject };
  302. resourceRequests.push(callbacks);
  303. if (groupDuplicates) {
  304. let duplicateRequests = this.duplicates.get(requestKey);
  305. if (!duplicateRequests) {
  306. duplicateRequests = [];
  307. this.duplicates.set(requestKey, duplicateRequests);
  308. }
  309. duplicateRequests.push(callbacks);
  310. }
  311. });
  312. }
  313. getMaxResources() {
  314. return this.requests.size;
  315. }
  316. async run(onloadListener, options) {
  317. const resourceURLs = [...this.requests.keys()];
  318. let indexResource = 0;
  319. return Promise.all(resourceURLs.map(async requestKey => {
  320. const [resourceURL, asBinary] = JSON.parse(requestKey);
  321. const resourceRequests = this.requests.get(requestKey);
  322. try {
  323. const content = await util.getContent(resourceURL, {
  324. asBinary,
  325. maxResourceSize: options.maxResourceSize,
  326. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  327. frameId: options.windowId
  328. });
  329. indexResource = indexResource + 1;
  330. onloadListener({ url: resourceURL });
  331. if (!this.cancelled) {
  332. resourceRequests.forEach(callbacks => {
  333. const duplicateCallbacks = this.duplicates.get(requestKey);
  334. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  335. callbacks.resolve({ content: content.data, indexResource, duplicate });
  336. });
  337. }
  338. } catch (error) {
  339. indexResource = indexResource + 1;
  340. onloadListener({ url: resourceURL });
  341. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  342. }
  343. this.requests.delete(requestKey);
  344. }));
  345. }
  346. cancel() {
  347. this.cancelled = true;
  348. const resourceURLs = [...this.requests.keys()];
  349. resourceURLs.forEach(requestKey => {
  350. const resourceRequests = this.requests.get(requestKey);
  351. resourceRequests.forEach(callbacks => callbacks.reject());
  352. this.requests.delete(requestKey);
  353. });
  354. }
  355. }
  356. // ---------
  357. // Processor
  358. // ---------
  359. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  360. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  361. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  362. const SCRIPT_TAG_FOUND = /<script/gi;
  363. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  364. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  365. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  366. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  367. class Processor {
  368. constructor(options, batchRequest) {
  369. this.options = options;
  370. this.stats = new Stats(options);
  371. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  372. this.batchRequest = batchRequest;
  373. this.stylesheets = new Map();
  374. this.styles = new Map();
  375. this.cssVariables = new Map();
  376. }
  377. initialize() {
  378. this.options.saveDate = new Date();
  379. this.options.saveUrl = this.options.url;
  380. if (this.options.enableMaff) {
  381. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  382. }
  383. this.maxResources = this.batchRequest.getMaxResources();
  384. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  385. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  386. }
  387. if (!this.options.removeImports && this.options.imports) {
  388. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  389. }
  390. this.stats.set("processed", "resources", this.maxResources);
  391. }
  392. async loadPage(pageContent, charset) {
  393. let content;
  394. if (!pageContent || this.options.saveRawPage) {
  395. content = await util.getContent(this.baseURI, {
  396. maxResourceSize: this.options.maxResourceSize,
  397. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  398. charset,
  399. frameId: this.options.windowId
  400. });
  401. pageContent = content.data;
  402. }
  403. this.doc = util.parseDocContent(pageContent, this.baseURI);
  404. if (this.options.saveRawPage) {
  405. let charset;
  406. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  407. const charsetDeclaration = element.content.split(";")[1];
  408. if (charsetDeclaration && !charset) {
  409. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  410. }
  411. });
  412. if (charset && charset != content.charset) {
  413. return this.loadPage(pageContent, charset);
  414. }
  415. }
  416. this.workStyleElement = this.doc.createElement("style");
  417. this.doc.body.appendChild(this.workStyleElement);
  418. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  419. }
  420. finalize() {
  421. if (this.workStyleElement.parentNode) {
  422. this.workStyleElement.remove();
  423. }
  424. }
  425. async getPageData() {
  426. util.postProcessDoc(this.doc);
  427. const url = util.parseURL(this.baseURI);
  428. if (this.options.insertSingleFileComment) {
  429. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  430. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  431. " \n url: " + this.options.saveUrl +
  432. " \n saved date: " + this.options.saveDate +
  433. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  434. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  435. }
  436. let size;
  437. if (this.options.displayStats) {
  438. size = util.getContentSize(this.doc.documentElement.outerHTML);
  439. }
  440. const content = util.serialize(this.doc, this.options.compressHTML);
  441. if (this.options.displayStats) {
  442. const contentSize = util.getContentSize(content);
  443. this.stats.set("processed", "HTML bytes", contentSize);
  444. this.stats.add("discarded", "HTML bytes", size - contentSize);
  445. }
  446. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  447. const replacementCharacter = this.options.filenameReplacementCharacter;
  448. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  449. if (!this.options.backgroundSave) {
  450. filename = filename.replace(/\//g, replacementCharacter);
  451. }
  452. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  453. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  454. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  455. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  456. filename = filename + "…" + extension;
  457. }
  458. if (!filename) {
  459. filename = "Unnamed page";
  460. }
  461. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  462. const pageData = {
  463. stats: this.stats.data,
  464. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  465. filename,
  466. content
  467. };
  468. if (this.options.addProof) {
  469. pageData.hash = await util.digest("SHA-256", content);
  470. }
  471. return pageData;
  472. }
  473. preProcessPage() {
  474. if (this.options.win) {
  475. 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));
  476. }
  477. if (this.options.images && !this.options.saveRawPage) {
  478. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  479. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  480. if (attributeValue) {
  481. const imageData = this.options.images[Number(attributeValue)];
  482. if (imageData) {
  483. if (this.options.removeHiddenElements && (
  484. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  485. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  486. )) {
  487. imgElement.setAttribute("src", EMPTY_IMAGE);
  488. } else {
  489. if (imageData.currentSrc) {
  490. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  491. imgElement.setAttribute("src", imageData.currentSrc);
  492. }
  493. if (this.options.loadDeferredImages) {
  494. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  495. imageData.src = imgElement.dataset.src;
  496. imgElement.setAttribute("src", imgElement.dataset.src);
  497. imgElement.removeAttribute("data-src");
  498. }
  499. }
  500. }
  501. }
  502. }
  503. });
  504. if (this.options.loadDeferredImages) {
  505. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  506. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  507. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  508. imgElement.removeAttribute("data-srcset");
  509. }
  510. });
  511. }
  512. }
  513. }
  514. replaceStyleContents() {
  515. if (this.options.stylesheets) {
  516. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  517. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  518. if (attributeValue) {
  519. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  520. if (stylesheetContent) {
  521. styleElement.textContent = stylesheetContent;
  522. }
  523. }
  524. });
  525. }
  526. }
  527. removeUnselectedElements() {
  528. removeUnmarkedElements(this.doc.body);
  529. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  530. function removeUnmarkedElements(element) {
  531. let selectedElementFound = false;
  532. Array.from(element.childNodes).forEach(node => {
  533. if (node.nodeType == 1) {
  534. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  535. selectedElementFound = selectedElementFound || isSelectedElement;
  536. if (isSelectedElement) {
  537. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  538. removeUnmarkedElements(node);
  539. } else if (selectedElementFound) {
  540. removeNode(node);
  541. } else {
  542. hideNode(node);
  543. }
  544. }
  545. });
  546. }
  547. function removeNode(node) {
  548. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  549. node.remove();
  550. } else {
  551. hideNode(node);
  552. }
  553. }
  554. function hideNode(node) {
  555. if (canHideNode(node)) {
  556. node.style.setProperty("display", "none", "important");
  557. Array.from(node.childNodes).forEach(removeNode);
  558. }
  559. }
  560. function canHideNode(node) {
  561. if (node.nodeType == 1) {
  562. const tagName = node.tagName && node.tagName.toLowerCase();
  563. return (tagName != "svg" && tagName != "style" && tagName != "link");
  564. }
  565. }
  566. }
  567. insertVideoPosters() {
  568. if (this.options.posters) {
  569. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  570. let videoElement;
  571. if (element.tagName == "VIDEO") {
  572. videoElement = element;
  573. } else {
  574. videoElement = element.parentElement;
  575. }
  576. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  577. if (attributeValue) {
  578. const posterURL = this.options.posters[Number(attributeValue)];
  579. if (!videoElement.poster && posterURL) {
  580. videoElement.setAttribute("poster", posterURL);
  581. }
  582. }
  583. });
  584. }
  585. }
  586. removeFrames() {
  587. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  588. this.stats.set("discarded", "frames", frameElements.length);
  589. this.stats.set("processed", "frames", frameElements.length);
  590. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  591. }
  592. removeScripts() {
  593. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  594. this.doc.querySelectorAll("[href]").forEach(element => {
  595. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  596. element.setAttribute("href", "");
  597. }
  598. });
  599. this.doc.querySelectorAll("[src]").forEach(element => {
  600. if (element.src && element.src.match(/^\s*javascript:/)) {
  601. element.removeAttribute("src");
  602. }
  603. });
  604. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  605. this.stats.set("discarded", "scripts", scriptElements.length);
  606. this.stats.set("processed", "scripts", scriptElements.length);
  607. scriptElements.forEach(element => element.remove());
  608. }
  609. removeVideoSources() {
  610. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  611. this.stats.set("discarded", "video sources", videoSourceElements.length);
  612. this.stats.set("processed", "video sources", videoSourceElements.length);
  613. videoSourceElements.forEach(element => {
  614. if (element.tagName == "SOURCE") {
  615. element.remove();
  616. } else {
  617. videoSourceElements.forEach(element => element.removeAttribute("src"));
  618. }
  619. });
  620. }
  621. removeAudioSources() {
  622. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  623. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  624. this.stats.set("processed", "audio sources", audioSourceElements.length);
  625. audioSourceElements.forEach(element => {
  626. if (element.tagName == "SOURCE") {
  627. element.remove();
  628. } else {
  629. audioSourceElements.forEach(element => element.removeAttribute("src"));
  630. }
  631. });
  632. }
  633. removeDiscardedResources() {
  634. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  635. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  636. 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\"])");
  637. this.stats.set("discarded", "objects", objectElements.length);
  638. this.stats.set("processed", "objects", objectElements.length);
  639. objectElements.forEach(element => element.remove());
  640. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  641. replacedAttributeValue.forEach(element => {
  642. let regExp;
  643. if (this.options.removeScripts) {
  644. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  645. } else {
  646. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  647. }
  648. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  649. if (relValue.length) {
  650. element.setAttribute("rel", relValue);
  651. } else {
  652. element.remove();
  653. }
  654. });
  655. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  656. if (this.options.compressHTML) {
  657. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  658. }
  659. if (!this.options.saveFavicon) {
  660. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  661. }
  662. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  663. }
  664. resetCharsetMeta() {
  665. let charset;
  666. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  667. const charsetDeclaration = element.content.split(";")[1];
  668. if (charsetDeclaration && !charset) {
  669. charset = charsetDeclaration.split("=")[1];
  670. if (charset) {
  671. this.charset = charset.trim().toLowerCase();
  672. }
  673. }
  674. element.remove();
  675. });
  676. const metaElement = this.doc.createElement("meta");
  677. metaElement.setAttribute("charset", "utf-8");
  678. if (this.doc.head.firstChild) {
  679. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  680. } else {
  681. this.doc.head.appendChild(metaElement);
  682. }
  683. }
  684. setInputValues() {
  685. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  686. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  687. input.setAttribute("value", value || "");
  688. });
  689. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  690. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  691. if (value == "true") {
  692. input.setAttribute("checked", "");
  693. }
  694. });
  695. this.doc.querySelectorAll("textarea").forEach(textarea => {
  696. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  697. textarea.textContent = value || "";
  698. });
  699. this.doc.querySelectorAll("select").forEach(select => {
  700. select.querySelectorAll("option").forEach(option => {
  701. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  702. if (selected) {
  703. option.setAttribute("selected", "");
  704. }
  705. });
  706. });
  707. }
  708. saveFavicon() {
  709. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  710. if (!faviconElement) {
  711. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  712. }
  713. if (!faviconElement) {
  714. faviconElement = this.doc.createElement("link");
  715. faviconElement.setAttribute("type", "image/x-icon");
  716. faviconElement.setAttribute("rel", "shortcut icon");
  717. faviconElement.setAttribute("href", "/favicon.ico");
  718. }
  719. this.doc.head.appendChild(faviconElement);
  720. }
  721. replaceCanvasElements() {
  722. if (this.options.canvases) {
  723. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  724. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  725. if (attributeValue) {
  726. const canvasData = this.options.canvases[Number(attributeValue)];
  727. if (canvasData) {
  728. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  729. this.stats.add("processed", "canvas", 1);
  730. }
  731. }
  732. });
  733. }
  734. }
  735. insertFonts() {
  736. if (this.options.fonts && this.options.fonts.length) {
  737. let stylesheetContent = "";
  738. this.options.fonts.forEach(fontData => {
  739. if (fontData["font-family"] && fontData.src) {
  740. stylesheetContent += "@font-face{";
  741. let stylesContent = "";
  742. Object.keys(fontData).forEach(fontStyle => {
  743. if (stylesContent) {
  744. stylesContent += ";";
  745. }
  746. stylesContent += fontStyle + ":" + fontData[fontStyle];
  747. });
  748. stylesheetContent += stylesContent + "}";
  749. }
  750. });
  751. if (stylesheetContent) {
  752. const styleElement = this.doc.createElement("style");
  753. styleElement.textContent = stylesheetContent;
  754. const existingStyleElement = this.doc.querySelector("style");
  755. if (existingStyleElement) {
  756. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  757. } else {
  758. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  759. }
  760. }
  761. }
  762. }
  763. removeHiddenElements() {
  764. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  765. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  766. this.stats.set("discarded", "hidden elements", removedElements.length);
  767. this.stats.set("processed", "hidden elements", removedElements.length);
  768. if (hiddenElements.length) {
  769. const styleElement = this.doc.createElement("style");
  770. styleElement.textContent = ".sf-hidden{display:none!important;}";
  771. this.doc.head.appendChild(styleElement);
  772. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  773. }
  774. removedElements.forEach(element => element.remove());
  775. }
  776. resolveHrefs() {
  777. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  778. const href = element.getAttribute("href").trim();
  779. if (!Util.testIgnoredPath(href)) {
  780. let resolvedURL;
  781. try {
  782. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  783. } catch (error) {
  784. // ignored
  785. }
  786. if (resolvedURL) {
  787. const url = Util.normalizeURL(this.options.url);
  788. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  789. resolvedURL = resolvedURL.substring(url.length);
  790. }
  791. try {
  792. element.setAttribute("href", resolvedURL);
  793. } catch (error) {
  794. // ignored
  795. }
  796. }
  797. }
  798. });
  799. }
  800. resolveStyleAttributeURLs() {
  801. this.doc.querySelectorAll("[style]").forEach(element => {
  802. let styleContent = element.getAttribute("style");
  803. if (this.options.compressCSS) {
  804. styleContent = util.compressCSS(styleContent);
  805. }
  806. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  807. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  808. this.styles.set(element, declarationList);
  809. });
  810. }
  811. async resolveStylesheetURLs() {
  812. const options = {
  813. maxResourceSize: this.options.maxResourceSize,
  814. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  815. url: this.options.url,
  816. charset: this.charset,
  817. compressCSS: this.options.compressCSS,
  818. updatedResources: this.options.updatedResources,
  819. rootDocument: this.options.rootDocument,
  820. frameId: this.options.windowId
  821. };
  822. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  823. .map(async element => {
  824. let mediaText;
  825. if (element.media) {
  826. mediaText = element.media.toLowerCase();
  827. }
  828. const stylesheetInfo = { mediaText };
  829. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  830. stylesheetInfo.scoped = true;
  831. }
  832. if (element.tagName.toLowerCase() == "link") {
  833. if (element.charset) {
  834. options.charset = element.charset;
  835. }
  836. }
  837. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  838. }));
  839. if (options.rootDocument) {
  840. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  841. await Promise.all(newResources.map(async resource => {
  842. resource.retrieved = true;
  843. const stylesheetInfo = {};
  844. const element = this.doc.createElement("style");
  845. this.doc.body.appendChild(element);
  846. element.textContent = resource.content;
  847. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  848. }));
  849. }
  850. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  851. stylesheets.set(element, stylesheetInfo);
  852. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  853. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  854. if (match && match[1] && match[1] != options.charset) {
  855. options.charset = match[1];
  856. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  857. }
  858. let stylesheet;
  859. try {
  860. stylesheet = cssTree.parse(Util.removeCssComments(stylesheetContent));
  861. } catch (error) {
  862. // ignored
  863. }
  864. if (stylesheet && stylesheet.children) {
  865. if (options.compressCSS) {
  866. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  867. }
  868. stylesheetInfo.stylesheet = stylesheet;
  869. } else {
  870. stylesheets.delete(element);
  871. }
  872. }
  873. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  874. let content;
  875. if (element.tagName.toLowerCase() == "link") {
  876. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  877. } else {
  878. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  879. }
  880. return content || "";
  881. }
  882. }
  883. async resolveFrameURLs() {
  884. if (!this.options.saveRawPage && this.options.frames) {
  885. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  886. await Promise.all(frameElements.map(async frameElement => {
  887. if (frameElement.tagName == "OBJECT") {
  888. frameElement.setAttribute("data", "data:text/html,");
  889. } else {
  890. frameElement.removeAttribute("src");
  891. frameElement.removeAttribute("srcdoc");
  892. }
  893. Array.from(frameElement.childNodes).forEach(node => node.remove());
  894. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  895. if (frameWindowId) {
  896. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  897. if (frameData) {
  898. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  899. }
  900. }
  901. }));
  902. }
  903. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  904. options.insertSingleFileComment = false;
  905. options.saveFavicon = false;
  906. options.doc = null;
  907. options.win = null;
  908. options.url = frameData.baseURI;
  909. options.windowId = frameWindowId;
  910. if (frameData.content) {
  911. options.content = frameData.content;
  912. options.canvases = frameData.canvases;
  913. options.fonts = frameData.fonts;
  914. options.stylesheets = frameData.stylesheets;
  915. options.images = frameData.images;
  916. options.posters = frameData.posters;
  917. options.usedFonts = frameData.usedFonts;
  918. options.shadowRoots = frameData.shadowRoots;
  919. options.imports = frameData.imports;
  920. frameData.runner = new Runner(options);
  921. frameData.frameElement = frameElement;
  922. await frameData.runner.loadPage();
  923. await frameData.runner.initialize();
  924. frameData.maxResources = batchRequest.getMaxResources();
  925. }
  926. }
  927. }
  928. insertShadowRootContents() {
  929. const doc = this.doc;
  930. const options = this.options;
  931. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  932. processElement(this.doc);
  933. if (this.options.removeScripts) {
  934. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  935. }
  936. const scriptElement = doc.createElement("script");
  937. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  938. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{if (!element.parentElement.shadowRoot) {const shadowRoot=element.parentElement.attachShadow({mode:element.getAttribute("${SHADOW_MODE_ATTRIBUTE_NAME}"),delegatesFocus:Boolean(element.getAttribute("${SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME}"))});shadowRoot.innerHTML=element.innerHTML;element.remove();processNode(shadowRoot)}})}})()`;
  939. doc.body.appendChild(scriptElement);
  940. }
  941. function processElement(element) {
  942. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  943. shadowRootElements.forEach(element => {
  944. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  945. if (attributeValue) {
  946. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  947. if (shadowRootData) {
  948. const templateElement = doc.createElement("template");
  949. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  950. if (shadowRootData.delegatesFocus) {
  951. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  952. }
  953. if (shadowRootData.adoptedStyleSheets) {
  954. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  955. const styleElement = doc.createElement("style");
  956. styleElement.textContent = stylesheetContent;
  957. templateElement.appendChild(styleElement);
  958. });
  959. }
  960. const shadowDoc = util.parseDocContent(shadowRootData.content);
  961. if (shadowDoc.head) {
  962. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  963. if (metaCharset) {
  964. metaCharset.remove();
  965. }
  966. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  967. }
  968. if (shadowDoc.body) {
  969. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  970. }
  971. processElement(templateElement);
  972. if (element.firstChild) {
  973. element.insertBefore(templateElement, element.firstChild);
  974. } else {
  975. element.appendChild(templateElement);
  976. }
  977. }
  978. }
  979. });
  980. }
  981. }
  982. async resolveHtmlImportURLs() {
  983. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  984. await Promise.all(linkElements.map(async linkElement => {
  985. const resourceURL = linkElement.href;
  986. linkElement.removeAttribute("href");
  987. const options = Object.create(this.options);
  988. options.insertSingleFileComment = false;
  989. options.saveFavicon = false;
  990. options.removeUnusedStyles = false;
  991. options.removeAlternativeMedias = false;
  992. options.removeUnusedFonts = false;
  993. options.removeHiddenElements = false;
  994. options.doc = null;
  995. options.win = null;
  996. options.url = resourceURL;
  997. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  998. if (attributeValue) {
  999. const importData = options.imports[Number(attributeValue)];
  1000. if (importData) {
  1001. options.content = importData.content;
  1002. importData.runner = new Runner(options);
  1003. await importData.runner.loadPage();
  1004. await importData.runner.initialize();
  1005. if (!options.removeImports) {
  1006. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1007. }
  1008. importData.runner.getStyleSheets().forEach(stylesheet => {
  1009. const importedStyleElement = this.doc.createElement("style");
  1010. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1011. this.stylesheets.set(importedStyleElement, stylesheet);
  1012. });
  1013. }
  1014. }
  1015. if (options.removeImports) {
  1016. linkElement.remove();
  1017. this.stats.add("discarded", "HTML imports", 1);
  1018. }
  1019. }));
  1020. }
  1021. removeUnusedStyles() {
  1022. if (!this.mediaAllInfo) {
  1023. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1024. }
  1025. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1026. this.stats.set("processed", "CSS rules", stats.processed);
  1027. this.stats.set("discarded", "CSS rules", stats.discarded);
  1028. }
  1029. removeUnusedFonts() {
  1030. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1031. }
  1032. removeAlternativeMedias() {
  1033. const stats = util.minifyMedias(this.stylesheets);
  1034. this.stats.set("processed", "medias", stats.processed);
  1035. this.stats.set("discarded", "medias", stats.discarded);
  1036. }
  1037. async processStylesheets() {
  1038. await Promise.all(Array.from(this.stylesheets).map(([, stylesheetInfo]) =>
  1039. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1040. ));
  1041. }
  1042. async processStyleAttributes() {
  1043. return Promise.all(Array.from(this.styles).map(([, declarationList]) =>
  1044. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1045. ));
  1046. }
  1047. async processPageResources() {
  1048. const processAttributeArgs = [
  1049. ["link[href][rel*=\"icon\"]", "href", false, true],
  1050. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1051. ["img[src], input[src][type=image]", "src", true],
  1052. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1053. ["video[poster]", "poster"],
  1054. ["*[background]", "background"],
  1055. ["image", "xlink:href"],
  1056. ["image", "href"]
  1057. ];
  1058. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1059. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1060. );
  1061. resourcePromises = resourcePromises.concat([
  1062. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1063. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1064. ]);
  1065. if (!this.options.removeAudioSrc) {
  1066. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1067. }
  1068. if (!this.options.removeVideoSrc) {
  1069. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1070. }
  1071. await Promise.all(resourcePromises);
  1072. if (this.options.saveFavicon) {
  1073. ProcessorHelper.processShortcutIcons(this.doc);
  1074. }
  1075. }
  1076. async processScripts() {
  1077. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1078. let resourceURL;
  1079. let scriptSrc;
  1080. if (element.tagName == "SCRIPT") {
  1081. scriptSrc = element.getAttribute("src");
  1082. element.removeAttribute("src");
  1083. } else {
  1084. scriptSrc = element.getAttribute("href");
  1085. element.removeAttribute("href");
  1086. }
  1087. element.removeAttribute("integrity");
  1088. element.textContent = "";
  1089. try {
  1090. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1091. } catch (error) {
  1092. // ignored
  1093. }
  1094. if (Util.testValidURL(resourceURL)) {
  1095. this.stats.add("processed", "scripts", 1);
  1096. const content = await util.getContent(resourceURL, {
  1097. asBinary: true,
  1098. maxResourceSize: this.options.maxResourceSize,
  1099. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1100. frameId: this.options.windowId
  1101. });
  1102. content.data = Util.getUpdatedResourceContent(resourceURL, content, this.options);
  1103. if (element.tagName == "SCRIPT") {
  1104. element.setAttribute("src", content.data);
  1105. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1106. element.setAttribute("async", "");
  1107. }
  1108. } else {
  1109. const scriptElement = this.doc.createElement("script");
  1110. scriptElement.setAttribute("src", content.data);
  1111. scriptElement.setAttribute("async", "");
  1112. element.parentElement.replaceChild(scriptElement, element);
  1113. }
  1114. }
  1115. }));
  1116. }
  1117. removeAlternativeImages() {
  1118. util.removeAlternativeImages(this.doc);
  1119. }
  1120. async removeAlternativeFonts() {
  1121. await util.removeAlternativeFonts(this.doc, this.stylesheets);
  1122. }
  1123. async processFrames() {
  1124. if (this.options.frames) {
  1125. const frameElements = Array.from(this.doc.querySelectorAll("iframe:not([" + util.HIDDEN_FRAME_ATTRIBUTE_NAME + "]), frame, object[type=\"text/html\"][data]"));
  1126. await Promise.all(frameElements.map(async frameElement => {
  1127. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1128. if (frameWindowId) {
  1129. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1130. if (frameData) {
  1131. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1132. if (frameData.runner) {
  1133. this.stats.add("processed", "frames", 1);
  1134. await frameData.runner.run();
  1135. const pageData = await frameData.runner.getPageData();
  1136. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1137. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1138. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1139. sandbox += " allow-scripts allow-same-origin";
  1140. }
  1141. frameElement.setAttribute("sandbox", sandbox);
  1142. if (frameElement.tagName == "OBJECT") {
  1143. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1144. } else {
  1145. if (frameElement.tagName == "FRAME") {
  1146. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1147. } else {
  1148. frameElement.setAttribute("srcdoc", pageData.content);
  1149. frameElement.removeAttribute("src");
  1150. }
  1151. }
  1152. this.stats.addAll(pageData);
  1153. } else {
  1154. this.stats.add("discarded", "frames", 1);
  1155. }
  1156. }
  1157. }
  1158. }));
  1159. }
  1160. }
  1161. async processHtmlImports() {
  1162. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1163. await Promise.all(linkElements.map(async linkElement => {
  1164. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1165. if (attributeValue) {
  1166. const importData = this.options.imports[Number(attributeValue)];
  1167. if (importData.runner) {
  1168. this.stats.add("processed", "HTML imports", 1);
  1169. await importData.runner.run();
  1170. const pageData = await importData.runner.getPageData();
  1171. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1172. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1173. this.stats.addAll(pageData);
  1174. } else {
  1175. this.stats.add("discarded", "HTML imports", 1);
  1176. }
  1177. }
  1178. }));
  1179. }
  1180. replaceStylesheets() {
  1181. this.doc.querySelectorAll("style").forEach(styleElement => {
  1182. const stylesheetInfo = this.stylesheets.get(styleElement);
  1183. if (stylesheetInfo) {
  1184. this.stylesheets.delete(styleElement);
  1185. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1186. styleElement.textContent = stylesheetContent;
  1187. if (stylesheetInfo.mediaText) {
  1188. styleElement.media = stylesheetInfo.mediaText;
  1189. }
  1190. } else {
  1191. styleElement.remove();
  1192. }
  1193. });
  1194. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1195. const stylesheetInfo = this.stylesheets.get(linkElement);
  1196. if (stylesheetInfo) {
  1197. this.stylesheets.delete(linkElement);
  1198. const styleElement = this.doc.createElement("style");
  1199. if (stylesheetInfo.mediaText) {
  1200. styleElement.media = stylesheetInfo.mediaText;
  1201. }
  1202. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1203. styleElement.textContent = stylesheetContent;
  1204. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1205. } else {
  1206. linkElement.remove();
  1207. }
  1208. });
  1209. }
  1210. replaceStyleAttributes() {
  1211. this.doc.querySelectorAll("[style]").forEach(async element => {
  1212. const declarations = this.styles.get(element);
  1213. if (declarations) {
  1214. this.styles.delete(element);
  1215. let styleContent = cssTree.generate(declarations);
  1216. element.setAttribute("style", styleContent);
  1217. } else {
  1218. element.setAttribute("style", "");
  1219. }
  1220. });
  1221. }
  1222. insertVariables() {
  1223. if (this.cssVariables.size) {
  1224. const styleElement = this.doc.createElement("style");
  1225. const firstStyleElement = this.doc.head.querySelector("style");
  1226. if (firstStyleElement) {
  1227. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1228. } else {
  1229. this.doc.head.appendChild(styleElement);
  1230. }
  1231. let stylesheetContent = "";
  1232. this.cssVariables.forEach((content, indexResource) => {
  1233. this.cssVariables.delete(indexResource);
  1234. if (stylesheetContent) {
  1235. stylesheetContent += ";";
  1236. }
  1237. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1238. });
  1239. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1240. }
  1241. }
  1242. compressHTML() {
  1243. let size;
  1244. if (this.options.displayStats) {
  1245. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1246. }
  1247. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1248. if (this.options.displayStats) {
  1249. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1250. }
  1251. }
  1252. cleanupPage() {
  1253. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1254. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1255. if (metaCharset) {
  1256. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1257. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1258. this.doc.head.querySelector("meta[charset]").remove();
  1259. }
  1260. }
  1261. }
  1262. async insertMAFFMetaData() {
  1263. const maffMetaData = await this.maffMetaDataPromise;
  1264. if (maffMetaData && maffMetaData.content) {
  1265. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1266. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1267. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1268. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1269. if (originalURLElement) {
  1270. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1271. }
  1272. if (archiveTimeElement) {
  1273. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1274. if (value) {
  1275. const date = new Date(value);
  1276. if (!isNaN(date.getTime())) {
  1277. this.options.saveDate = new Date(value);
  1278. }
  1279. }
  1280. }
  1281. }
  1282. }
  1283. async setDocInfo() {
  1284. const titleElement = this.doc.querySelector("title");
  1285. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1286. const authorElement = this.doc.querySelector("meta[name=author]");
  1287. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1288. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1289. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1290. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1291. this.options.info = {
  1292. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1293. lang: this.doc.documentElement.lang,
  1294. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1295. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1296. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : ""
  1297. };
  1298. }
  1299. }
  1300. // ---------------
  1301. // ProcessorHelper
  1302. // ---------------
  1303. const DATA_URI_PREFIX = "data:";
  1304. const ABOUT_BLANK_URI = "about:blank";
  1305. const EMPTY_DATA_URI = "data:null;base64,";
  1306. const REGEXP_URL_HASH = /(#.+?)$/;
  1307. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1308. const VAR_MAX_SIZE = 1024 * 1024 * 2;
  1309. class ProcessorHelper {
  1310. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1311. const date = options.saveDate;
  1312. const url = util.parseURL(options.saveUrl);
  1313. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1314. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1315. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1316. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1317. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1318. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1319. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1320. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1321. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1322. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1323. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1324. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1325. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1326. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1327. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1328. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1329. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1330. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1331. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1332. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1333. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1334. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1335. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1336. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1337. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1338. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1339. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1340. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1341. template = await Util.evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1342. template = await Util.evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1343. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1344. template = await Util.evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1345. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1346. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1347. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1348. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1349. for (const [name, value] of params) {
  1350. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1351. }
  1352. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1353. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1354. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decode(Util.getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1357. if (content) {
  1358. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1359. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1360. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1361. }
  1362. return template.trim();
  1363. function decode(value) {
  1364. try {
  1365. return decodeURI(value);
  1366. } catch (error) {
  1367. return value;
  1368. }
  1369. }
  1370. }
  1371. static setBackgroundImage(element, url, style) {
  1372. element.style.setProperty("background-blend-mode", "normal", "important");
  1373. element.style.setProperty("background-clip", "content-box", "important");
  1374. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1375. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1376. element.style.setProperty("background-image", url, "important");
  1377. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1378. element.style.setProperty("background-origin", "content-box", "important");
  1379. element.style.setProperty("background-repeat", "no-repeat", "important");
  1380. }
  1381. static processShortcutIcons(doc) {
  1382. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1383. if (!shortcutIcon) {
  1384. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1385. if (shortcutIcon) {
  1386. shortcutIcon.rel = "icon";
  1387. }
  1388. }
  1389. if (shortcutIcon) {
  1390. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1391. if (linkElement != shortcutIcon) {
  1392. linkElement.remove();
  1393. }
  1394. });
  1395. }
  1396. }
  1397. static removeSingleLineCssComments(stylesheet) {
  1398. const removedRules = [];
  1399. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1400. const ruleData = cssRule.data;
  1401. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1402. removedRules.push(cssRule);
  1403. }
  1404. }
  1405. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1406. }
  1407. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1408. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1409. const imports = Util.getImportFunctions(stylesheetContent);
  1410. await Promise.all(imports.map(async cssImport => {
  1411. const match = Util.matchImport(cssImport);
  1412. if (match) {
  1413. const regExpCssImport = Util.getRegExp(cssImport);
  1414. let resourceURL = Util.normalizeURL(match.resourceURL);
  1415. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1416. try {
  1417. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1418. } catch (error) {
  1419. // ignored
  1420. }
  1421. if (Util.testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1422. const content = await util.getContent(resourceURL, {
  1423. maxResourceSize: options.maxResourceSize,
  1424. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1425. validateTextContentType: true,
  1426. frameId: options.windowId
  1427. });
  1428. resourceURL = content.resourceURL;
  1429. content.data = Util.getUpdatedResourceContent(resourceURL, content, options);
  1430. let importedStylesheetContent = Util.removeCssComments(content.data);
  1431. if (options.compressCSS) {
  1432. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1433. }
  1434. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1435. if (stylesheetContent.includes(cssImport)) {
  1436. const ancestorStyleSheets = new Set(importedStyleSheets);
  1437. ancestorStyleSheets.add(resourceURL);
  1438. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1439. workStylesheet.textContent = importedStylesheetContent;
  1440. if (workStylesheet.sheet.cssRules.length) {
  1441. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1442. } else {
  1443. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1444. }
  1445. }
  1446. } else {
  1447. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1448. }
  1449. } else {
  1450. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1451. }
  1452. }
  1453. }));
  1454. return stylesheetContent;
  1455. }
  1456. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1457. const urlFunctions = Util.getUrlFunctions(stylesheetContent, true);
  1458. urlFunctions.map(urlFunction => {
  1459. const originalResourceURL = Util.matchURL(urlFunction);
  1460. const resourceURL = Util.normalizeURL(originalResourceURL);
  1461. if (!Util.testIgnoredPath(resourceURL)) {
  1462. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1463. let resolvedURL;
  1464. if (!originalResourceURL.startsWith("#")) {
  1465. try {
  1466. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1467. } catch (error) {
  1468. // ignored
  1469. }
  1470. }
  1471. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1472. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1473. }
  1474. } else {
  1475. let newUrlFunction;
  1476. if (originalResourceURL) {
  1477. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1478. } else {
  1479. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1480. }
  1481. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1482. }
  1483. }
  1484. });
  1485. return stylesheetContent;
  1486. }
  1487. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1488. resourceURL = Util.normalizeURL(resourceURL);
  1489. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1490. const content = await util.getContent(resourceURL, {
  1491. maxResourceSize: options.maxResourceSize,
  1492. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1493. charset: options.charset,
  1494. frameId: options.windowId
  1495. });
  1496. resourceURL = content.resourceURL;
  1497. content.data = Util.getUpdatedResourceContent(content.resourceURL, content, options);
  1498. let stylesheetContent = Util.removeCssComments(content.data);
  1499. if (options.compressCSS) {
  1500. stylesheetContent = util.compressCSS(stylesheetContent);
  1501. }
  1502. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1503. return stylesheetContent;
  1504. }
  1505. }
  1506. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1507. const promises = [];
  1508. const removedRules = [];
  1509. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1510. const ruleData = cssRule.data;
  1511. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1512. removedRules.push(cssRule);
  1513. } else if (ruleData.block && ruleData.block.children) {
  1514. if (ruleData.type == "Rule") {
  1515. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1516. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1517. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1518. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1519. promises.push(processFontFaceRule(ruleData));
  1520. }
  1521. }
  1522. }
  1523. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1524. await Promise.all(promises);
  1525. async function processFontFaceRule(ruleData) {
  1526. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1527. if (declaration.type == "Declaration" && declaration.value.children) {
  1528. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value), true);
  1529. await Promise.all(urlFunctions.map(async urlFunction => {
  1530. const originalResourceURL = Util.matchURL(urlFunction);
  1531. const resourceURL = Util.normalizeURL(originalResourceURL);
  1532. if (!Util.testIgnoredPath(resourceURL)) {
  1533. if (Util.testValidURL(resourceURL)) {
  1534. let { content } = await batchRequest.addURL(resourceURL, true);
  1535. replaceURLs(declaration, originalResourceURL, content);
  1536. }
  1537. }
  1538. }));
  1539. }
  1540. }));
  1541. function replaceURLs(declaration, oldURL, newURL) {
  1542. declaration.value.children.forEach(token => {
  1543. if (token.type == "Url" && util.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1544. token.value.value = newURL;
  1545. }
  1546. });
  1547. }
  1548. }
  1549. }
  1550. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1551. await Promise.all(declarations.map(async declaration => {
  1552. if (declaration.type == "Declaration" && declaration.value.children) {
  1553. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1554. await Promise.all(urlFunctions.map(async urlFunction => {
  1555. const originalResourceURL = Util.matchURL(urlFunction);
  1556. const resourceURL = Util.normalizeURL(originalResourceURL);
  1557. if (!Util.testIgnoredPath(resourceURL)) {
  1558. if (Util.testValidURL(resourceURL)) {
  1559. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1560. let variableDefined;
  1561. const tokens = [];
  1562. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1563. if (!originalResourceURL.startsWith("#")) {
  1564. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < VAR_MAX_SIZE) {
  1565. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1566. tokens.push({ parent, token, value });
  1567. variableDefined = true;
  1568. } else {
  1569. token.data.value.value = content;
  1570. }
  1571. }
  1572. });
  1573. if (variableDefined) {
  1574. cssVariables.set(indexResource, content);
  1575. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1576. }
  1577. }
  1578. }
  1579. }));
  1580. }
  1581. }));
  1582. function findURLToken(url, children, callback, depth = 0) {
  1583. for (let token = children.head; token; token = token.next) {
  1584. if (token.data.children) {
  1585. findURLToken(url, token.data.children, callback, depth + 1);
  1586. }
  1587. if (token.data.type == "Url" && util.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1588. callback(token, children, depth == 0);
  1589. }
  1590. }
  1591. }
  1592. }
  1593. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1594. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1595. let resourceURL = resourceElement.getAttribute(attributeName);
  1596. if (resourceURL != null) {
  1597. resourceURL = Util.normalizeURL(resourceURL);
  1598. if (!Util.testIgnoredPath(resourceURL)) {
  1599. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1600. if (Util.testValidPath(resourceURL)) {
  1601. try {
  1602. resourceURL = util.resolveURL(resourceURL, baseURI);
  1603. } catch (error) {
  1604. // ignored
  1605. }
  1606. if (Util.testValidURL(resourceURL)) {
  1607. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1608. if (resourceElement.dataset.singleFileOriginURL) {
  1609. let originURL = resourceElement.dataset.singleFileOriginURL;
  1610. delete resourceElement.dataset.singleFileOriginURL;
  1611. if (content == EMPTY_DATA_URI) {
  1612. try {
  1613. originURL = util.resolveURL(originURL, baseURI);
  1614. } catch (error) {
  1615. // ignored
  1616. }
  1617. try {
  1618. resourceURL = originURL;
  1619. content = (await util.getContent(resourceURL, {
  1620. asBinary: true,
  1621. maxResourceSize: options.maxResourceSize,
  1622. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1623. frameId: options.windowId
  1624. })).data;
  1625. } catch (error) {
  1626. // ignored
  1627. }
  1628. }
  1629. }
  1630. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1631. resourceElement.remove();
  1632. } else {
  1633. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1634. if (!forbiddenPrefixFound) {
  1635. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1636. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < VAR_MAX_SIZE) {
  1637. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1638. cssVariables.set(indexResource, content);
  1639. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1640. styles.set(resourceElement, declarationList);
  1641. } else {
  1642. resourceElement.setAttribute(attributeName, content);
  1643. }
  1644. } else {
  1645. resourceElement.setAttribute(attributeName, content);
  1646. }
  1647. }
  1648. }
  1649. }
  1650. }
  1651. }
  1652. }
  1653. }));
  1654. }
  1655. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1656. const attributeName = "xlink:href";
  1657. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1658. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1659. let resourceURL = Util.normalizeURL(originalResourceURL);
  1660. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1661. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1662. try {
  1663. resourceURL = util.resolveURL(resourceURL, baseURI);
  1664. } catch (error) {
  1665. // ignored
  1666. }
  1667. if (Util.testValidURL(resourceURL)) {
  1668. const { content } = await batchRequest.addURL(resourceURL);
  1669. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1670. if (hashMatch && hashMatch[0]) {
  1671. let symbolElement;
  1672. try {
  1673. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1674. } catch (error) {
  1675. // ignored
  1676. }
  1677. if (symbolElement) {
  1678. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1679. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1680. }
  1681. } else {
  1682. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1683. }
  1684. }
  1685. } else if (resourceURL == options.url) {
  1686. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1687. }
  1688. }));
  1689. }
  1690. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1691. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1692. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1693. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1694. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1695. if (!Util.testIgnoredPath(resourceURL)) {
  1696. if (Util.testValidPath(resourceURL)) {
  1697. try {
  1698. resourceURL = util.resolveURL(resourceURL, baseURI);
  1699. } catch (error) {
  1700. // ignored
  1701. }
  1702. if (Util.testValidURL(resourceURL)) {
  1703. const { content } = await batchRequest.addURL(resourceURL, true);
  1704. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1705. if (forbiddenPrefixFound) {
  1706. return "";
  1707. }
  1708. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1709. } else {
  1710. return "";
  1711. }
  1712. } else {
  1713. return "";
  1714. }
  1715. } else {
  1716. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1717. }
  1718. }));
  1719. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1720. }));
  1721. }
  1722. static replaceImageSource(imgElement, variableName, options) {
  1723. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1724. if (attributeValue) {
  1725. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1726. if (imageData && imageData.replaceable) {
  1727. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1728. const backgroundStyle = {};
  1729. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1730. if (backgroundSize) {
  1731. backgroundStyle["background-size"] = imageData.objectFit;
  1732. }
  1733. if (imageData.objectPosition) {
  1734. backgroundStyle["background-position"] = imageData.objectPosition;
  1735. }
  1736. if (imageData.backgroundColor) {
  1737. backgroundStyle["background-color"] = imageData.backgroundColor;
  1738. }
  1739. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1740. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1741. return true;
  1742. }
  1743. }
  1744. }
  1745. }
  1746. // ----
  1747. // Util
  1748. // ----
  1749. const BLOB_URI_PREFIX = "blob:";
  1750. const HTTP_URI_PREFIX = /^https?:\/\//;
  1751. const FILE_URI_PREFIX = /^file:\/\//;
  1752. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1753. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1754. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1755. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1756. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1757. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1758. 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;
  1759. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1760. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1761. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1762. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1763. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1764. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1765. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1766. class Util {
  1767. static getUpdatedResourceContent(resourceURL, content, options) {
  1768. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1769. options.updatedResources[resourceURL].retrieved = true;
  1770. return options.updatedResources[resourceURL].content;
  1771. } else {
  1772. return content.data || "";
  1773. }
  1774. }
  1775. static normalizeURL(url) {
  1776. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1777. return url;
  1778. } else {
  1779. return url.split("#")[0];
  1780. }
  1781. }
  1782. static getCSSValue(value) {
  1783. let result = "";
  1784. try {
  1785. result = cssTree.generate(value);
  1786. } catch (error) {
  1787. // ignored
  1788. }
  1789. return result;
  1790. }
  1791. static getOnEventAttributeNames(doc) {
  1792. const element = doc.createElement("div");
  1793. const attributeNames = [];
  1794. for (const propertyName in element) {
  1795. if (propertyName.startsWith("on")) {
  1796. attributeNames.push(propertyName);
  1797. }
  1798. }
  1799. attributeNames.push("onunload");
  1800. return attributeNames;
  1801. }
  1802. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1803. let maxLength;
  1804. if (template) {
  1805. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1806. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1807. if (template.match(replaceRegExp)) {
  1808. const matchedLength = template.match(replaceRegExp)[0];
  1809. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1810. if (isNaN(maxLength) || maxLength <= 0) {
  1811. maxLength = null;
  1812. }
  1813. } else {
  1814. replaceRegExp = new RegExp(regExpVariable, "g");
  1815. }
  1816. if (template.match(replaceRegExp)) {
  1817. let value = await valueGetter();
  1818. if (!dontReplaceSlash) {
  1819. value = value.replace(/\/+/g, replacementCharacter);
  1820. }
  1821. if (maxLength) {
  1822. value = await util.truncateText(value, maxLength);
  1823. }
  1824. return template.replace(replaceRegExp, value);
  1825. }
  1826. }
  1827. return template;
  1828. }
  1829. static getLastSegment(url, replacementCharacter) {
  1830. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1831. if (!lastSegment) {
  1832. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1833. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1834. }
  1835. if (!lastSegment) {
  1836. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1837. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1838. }
  1839. if (!lastSegment) {
  1840. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1841. }
  1842. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1843. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1844. lastSegment = lastSegmentMatch[1];
  1845. }
  1846. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1847. return lastSegment;
  1848. }
  1849. static getRegExp(string) {
  1850. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1851. }
  1852. static getUrlFunctions(stylesheetContent, unique) {
  1853. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1854. if (unique) {
  1855. return [...new Set(result)];
  1856. } else {
  1857. return result;
  1858. }
  1859. }
  1860. static getImportFunctions(stylesheetContent) {
  1861. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1862. }
  1863. static findShortcutIcon(shortcutIcons) {
  1864. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1865. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1866. return shortcutIcons[0];
  1867. }
  1868. static matchURL(stylesheetContent) {
  1869. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1870. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1871. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1872. return match && match[1];
  1873. }
  1874. static testIgnoredPath(resourceURL) {
  1875. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1876. }
  1877. static testValidPath(resourceURL) {
  1878. return resourceURL && !resourceURL.match(EMPTY_URL);
  1879. }
  1880. static testValidURL(resourceURL) {
  1881. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1882. }
  1883. static matchImport(stylesheetContent) {
  1884. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1885. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1886. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1887. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1888. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1889. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1890. if (match) {
  1891. const [, resourceURL, media] = match;
  1892. return { resourceURL, media };
  1893. }
  1894. }
  1895. static removeCssComments(stylesheetContent) {
  1896. let start, end;
  1897. do {
  1898. start = stylesheetContent.indexOf("/*");
  1899. end = stylesheetContent.indexOf("*/", start + 2);
  1900. if (start != -1 && end != -1) {
  1901. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1902. }
  1903. } while (start != -1 && end != -1);
  1904. return stylesheetContent;
  1905. }
  1906. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1907. if (mediaQuery) {
  1908. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1909. } else {
  1910. return stylesheetContent;
  1911. }
  1912. }
  1913. }
  1914. function log(...args) {
  1915. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1916. }
  1917. // -----
  1918. // Stats
  1919. // -----
  1920. const STATS_DEFAULT_VALUES = {
  1921. discarded: {
  1922. "HTML bytes": 0,
  1923. "hidden elements": 0,
  1924. "HTML imports": 0,
  1925. scripts: 0,
  1926. objects: 0,
  1927. "audio sources": 0,
  1928. "video sources": 0,
  1929. frames: 0,
  1930. "CSS rules": 0,
  1931. canvas: 0,
  1932. stylesheets: 0,
  1933. resources: 0,
  1934. medias: 0
  1935. },
  1936. processed: {
  1937. "HTML bytes": 0,
  1938. "hidden elements": 0,
  1939. "HTML imports": 0,
  1940. scripts: 0,
  1941. objects: 0,
  1942. "audio sources": 0,
  1943. "video sources": 0,
  1944. frames: 0,
  1945. "CSS rules": 0,
  1946. canvas: 0,
  1947. stylesheets: 0,
  1948. resources: 0,
  1949. medias: 0
  1950. }
  1951. };
  1952. class Stats {
  1953. constructor(options) {
  1954. this.options = options;
  1955. if (options.displayStats) {
  1956. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1957. }
  1958. }
  1959. set(type, subType, value) {
  1960. if (this.options.displayStats) {
  1961. this.data[type][subType] = value;
  1962. }
  1963. }
  1964. add(type, subType, value) {
  1965. if (this.options.displayStats) {
  1966. this.data[type][subType] += value;
  1967. }
  1968. }
  1969. addAll(pageData) {
  1970. if (this.options.displayStats) {
  1971. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1972. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1973. }
  1974. }
  1975. }
  1976. return { getClass };
  1977. })();