single-file-core.js 86 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree, sessionId = 0;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. if (options.sessionId === undefined) {
  34. options.sessionId = sessionId;
  35. sessionId++;
  36. }
  37. }
  38. async run() {
  39. if (this.options.userScriptEnabled) {
  40. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  41. }
  42. this.runner = new Runner(this.options, true);
  43. await this.runner.loadPage();
  44. await this.runner.initialize();
  45. if (this.options.userScriptEnabled) {
  46. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  47. }
  48. await this.runner.run();
  49. }
  50. cancel() {
  51. this.cancelled = true;
  52. if (this.runner) {
  53. this.runner.cancel();
  54. }
  55. }
  56. getPageData() {
  57. return this.runner.getPageData();
  58. }
  59. }
  60. // -------------
  61. // ProgressEvent
  62. // -------------
  63. const PAGE_LOADING = "page-loading";
  64. const PAGE_LOADED = "page-loaded";
  65. const RESOURCES_INITIALIZING = "resource-initializing";
  66. const RESOURCES_INITIALIZED = "resources-initialized";
  67. const RESOURCE_LOADED = "resource-loaded";
  68. const PAGE_ENDED = "page-ended";
  69. const STAGE_STARTED = "stage-started";
  70. const STAGE_ENDED = "stage-ended";
  71. const STAGE_TASK_STARTED = "stage-task-started";
  72. const STAGE_TASK_ENDED = "stage-task-ended";
  73. class ProgressEvent {
  74. constructor(type, detail) {
  75. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  76. }
  77. }
  78. // ------
  79. // Runner
  80. // ------
  81. const RESOLVE_URLS_STAGE = 0;
  82. const REPLACE_DATA_STAGE = 1;
  83. const REPLACE_DOCS_STAGE = 2;
  84. const POST_PROCESS_STAGE = 3;
  85. const STAGES = [{
  86. sequential: [
  87. { action: "preProcessPage" },
  88. { action: "replaceStyleContents" },
  89. { action: "resetCharsetMeta" },
  90. { option: "saveFavicon", action: "saveFavicon" },
  91. { action: "replaceCanvasElements" },
  92. { action: "insertFonts" },
  93. { action: "insertShadowRootContents" },
  94. { action: "setInputValues" },
  95. { option: "removeScripts", action: "removeScripts" },
  96. { option: "selected", action: "removeUnselectedElements" },
  97. { option: "removeVideoSrc", action: "insertVideoPosters" },
  98. { option: "removeFrames", action: "removeFrames" },
  99. { option: "removeVideoSrc", action: "removeVideoSources" },
  100. { option: "removeAudioSrc", action: "removeAudioSources" },
  101. { action: "removeDiscardedResources" },
  102. { option: "removeHiddenElements", action: "removeHiddenElements" },
  103. { action: "resolveHrefs" },
  104. { action: "resolveStyleAttributeURLs" }
  105. ],
  106. parallel: [
  107. { action: "resolveStylesheetURLs" },
  108. { option: "!removeFrames", action: "resolveFrameURLs" },
  109. { action: "resolveHtmlImportURLs" }
  110. ]
  111. }, {
  112. sequential: [
  113. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  114. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  115. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  116. ],
  117. parallel: [
  118. { action: "processStylesheets" },
  119. { action: "processStyleAttributes" },
  120. { action: "processPageResources" },
  121. { option: "!removeScripts", action: "processScripts" }
  122. ]
  123. }, {
  124. sequential: [
  125. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  126. ],
  127. parallel: [
  128. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  129. { option: "!removeFrames", action: "processFrames" },
  130. { option: "!removeImports", action: "processHtmlImports" },
  131. ]
  132. }, {
  133. sequential: [
  134. { action: "replaceStylesheets" },
  135. { action: "replaceStyleAttributes" },
  136. { action: "insertVariables" },
  137. { option: "compressHTML", action: "compressHTML" },
  138. { action: "cleanupPage" }
  139. ],
  140. parallel: [
  141. { option: "enableMaff", action: "insertMAFFMetaData" },
  142. { action: "setDocInfo" }
  143. ]
  144. }];
  145. class Runner {
  146. constructor(options, root) {
  147. const rootDocDefined = root && options.doc;
  148. this.root = root;
  149. this.options = options;
  150. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  151. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  152. this.options.rootDocument = root;
  153. this.options.updatedResources = this.options.updatedResources || {};
  154. this.batchRequest = new BatchRequest();
  155. this.processor = new Processor(options, this.batchRequest);
  156. if (rootDocDefined) {
  157. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  158. this.options.canvases = docData.canvases;
  159. this.options.fonts = docData.fonts;
  160. this.options.stylesheets = docData.stylesheets;
  161. this.options.images = docData.images;
  162. this.options.posters = docData.posters;
  163. this.options.usedFonts = docData.usedFonts;
  164. this.options.shadowRoots = docData.shadowRoots;
  165. this.options.imports = docData.imports;
  166. this.options.referrer = docData.referrer;
  167. this.markedElements = docData.markedElements;
  168. }
  169. if (this.options.saveRawPage) {
  170. this.options.removeFrames = true;
  171. }
  172. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  173. this.onprogress = options.onprogress || (() => { });
  174. }
  175. async loadPage() {
  176. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  177. await this.processor.loadPage(this.options.content);
  178. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  179. }
  180. async initialize() {
  181. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  182. await this.executeStage(RESOLVE_URLS_STAGE);
  183. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  184. if (this.root && this.options.doc) {
  185. util.postProcessDoc(this.options.doc, this.markedElements);
  186. }
  187. }
  188. cancel() {
  189. this.cancelled = true;
  190. this.batchRequest.cancel();
  191. if (this.root) {
  192. if (this.options.frames) {
  193. this.options.frames.forEach(cancelRunner);
  194. }
  195. if (this.options.imports) {
  196. this.options.imports.forEach(cancelRunner);
  197. }
  198. }
  199. function cancelRunner(resourceData) {
  200. if (resourceData.runner) {
  201. resourceData.runner.cancel();
  202. }
  203. }
  204. }
  205. async run() {
  206. if (this.root) {
  207. this.processor.initialize(this.batchRequest);
  208. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  209. }
  210. await this.batchRequest.run(detail => {
  211. detail.pageURL = this.options.url;
  212. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  213. }, this.options);
  214. await this.pendingPromises;
  215. this.options.doc = null;
  216. this.options.win = null;
  217. await this.executeStage(REPLACE_DOCS_STAGE);
  218. await this.executeStage(POST_PROCESS_STAGE);
  219. this.processor.finalize();
  220. }
  221. getDocument() {
  222. return this.processor.doc;
  223. }
  224. getStyleSheets() {
  225. return this.processor.stylesheets;
  226. }
  227. getPageData() {
  228. if (this.root) {
  229. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  230. }
  231. return this.processor.getPageData();
  232. }
  233. async executeStage(step) {
  234. if (DEBUG) {
  235. log("**** STARTED STAGE", step, "****");
  236. }
  237. const frame = !this.root;
  238. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  239. STAGES[step].sequential.forEach(task => {
  240. let startTime;
  241. if (DEBUG) {
  242. startTime = Date.now();
  243. log(" -- STARTED task =", task.action);
  244. }
  245. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  246. if (!this.cancelled) {
  247. this.executeTask(task);
  248. }
  249. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  250. if (DEBUG) {
  251. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  252. }
  253. });
  254. let parallelTasksPromise;
  255. if (STAGES[step].parallel) {
  256. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  257. let startTime;
  258. if (DEBUG) {
  259. startTime = Date.now();
  260. log(" // STARTED task =", task.action);
  261. }
  262. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  263. if (!this.cancelled) {
  264. await this.executeTask(task);
  265. }
  266. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  267. if (DEBUG) {
  268. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  269. }
  270. }));
  271. } else {
  272. parallelTasksPromise = Promise.resolve();
  273. }
  274. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  275. if (DEBUG) {
  276. log("**** ENDED STAGE", step, "****");
  277. }
  278. return parallelTasksPromise;
  279. }
  280. executeTask(task) {
  281. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  282. return this.processor[task.action]();
  283. }
  284. }
  285. }
  286. // ------------
  287. // BatchRequest
  288. // ------------
  289. class BatchRequest {
  290. constructor() {
  291. this.requests = new Map();
  292. this.duplicates = new Map();
  293. }
  294. addURL(resourceURL, asBinary, groupDuplicates) {
  295. return new Promise((resolve, reject) => {
  296. const requestKey = JSON.stringify([resourceURL, asBinary]);
  297. let resourceRequests = this.requests.get(requestKey);
  298. if (!resourceRequests) {
  299. resourceRequests = [];
  300. this.requests.set(requestKey, resourceRequests);
  301. }
  302. const callbacks = { resolve, reject };
  303. resourceRequests.push(callbacks);
  304. if (groupDuplicates) {
  305. let duplicateRequests = this.duplicates.get(requestKey);
  306. if (!duplicateRequests) {
  307. duplicateRequests = [];
  308. this.duplicates.set(requestKey, duplicateRequests);
  309. }
  310. duplicateRequests.push(callbacks);
  311. }
  312. });
  313. }
  314. getMaxResources() {
  315. return this.requests.size;
  316. }
  317. run(onloadListener, options) {
  318. const resourceURLs = [...this.requests.keys()];
  319. let indexResource = 0;
  320. return Promise.all(resourceURLs.map(async requestKey => {
  321. const [resourceURL, asBinary] = JSON.parse(requestKey);
  322. const resourceRequests = this.requests.get(requestKey);
  323. try {
  324. const content = await util.getContent(resourceURL, {
  325. asBinary,
  326. maxResourceSize: options.maxResourceSize,
  327. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  328. frameId: options.windowId
  329. });
  330. indexResource = indexResource + 1;
  331. onloadListener({ url: resourceURL });
  332. if (!this.cancelled) {
  333. resourceRequests.forEach(callbacks => {
  334. const duplicateCallbacks = this.duplicates.get(requestKey);
  335. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  336. callbacks.resolve({ content: content.data, indexResource, duplicate });
  337. });
  338. }
  339. } catch (error) {
  340. indexResource = indexResource + 1;
  341. onloadListener({ url: resourceURL });
  342. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  343. }
  344. this.requests.delete(requestKey);
  345. }));
  346. }
  347. cancel() {
  348. this.cancelled = true;
  349. const resourceURLs = [...this.requests.keys()];
  350. resourceURLs.forEach(requestKey => {
  351. const resourceRequests = this.requests.get(requestKey);
  352. resourceRequests.forEach(callbacks => callbacks.reject());
  353. this.requests.delete(requestKey);
  354. });
  355. }
  356. }
  357. // ---------
  358. // Processor
  359. // ---------
  360. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  361. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  362. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  363. const SCRIPT_TAG_FOUND = /<script/gi;
  364. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  365. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  366. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  367. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  368. class Processor {
  369. constructor(options, batchRequest) {
  370. this.options = options;
  371. this.stats = new Stats(options);
  372. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  373. this.batchRequest = batchRequest;
  374. this.stylesheets = new Map();
  375. this.styles = new Map();
  376. this.cssVariables = new Map();
  377. }
  378. initialize() {
  379. this.options.saveDate = new Date();
  380. this.options.saveUrl = this.options.url;
  381. if (this.options.enableMaff) {
  382. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  383. }
  384. this.maxResources = this.batchRequest.getMaxResources();
  385. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  386. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  387. }
  388. if (!this.options.removeImports && this.options.imports) {
  389. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  390. }
  391. this.stats.set("processed", "resources", this.maxResources);
  392. }
  393. async loadPage(pageContent, charset) {
  394. let content;
  395. if (!pageContent || this.options.saveRawPage) {
  396. content = await util.getContent(this.baseURI, {
  397. maxResourceSize: this.options.maxResourceSize,
  398. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  399. charset,
  400. frameId: this.options.windowId
  401. });
  402. pageContent = content.data;
  403. }
  404. this.doc = util.parseDocContent(pageContent, this.baseURI);
  405. if (this.options.saveRawPage) {
  406. let charset;
  407. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  408. const charsetDeclaration = element.content.split(";")[1];
  409. if (charsetDeclaration && !charset) {
  410. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  411. }
  412. });
  413. if (charset && charset != content.charset) {
  414. return this.loadPage(pageContent, charset);
  415. }
  416. }
  417. this.workStyleElement = this.doc.createElement("style");
  418. this.doc.body.appendChild(this.workStyleElement);
  419. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  420. }
  421. finalize() {
  422. if (this.workStyleElement.parentNode) {
  423. this.workStyleElement.remove();
  424. }
  425. }
  426. async getPageData() {
  427. util.postProcessDoc(this.doc);
  428. const url = util.parseURL(this.baseURI);
  429. if (this.options.insertSingleFileComment) {
  430. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  431. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  432. " \n url: " + this.options.saveUrl +
  433. " \n saved date: " + this.options.saveDate +
  434. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  435. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  436. }
  437. if (this.options.insertCanonicalLink) {
  438. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  439. if (!canonicalLink) {
  440. canonicalLink = this.doc.createElement("link");
  441. canonicalLink.setAttribute("rel", "canonical");
  442. this.doc.head.appendChild(canonicalLink);
  443. }
  444. if (canonicalLink && !canonicalLink.href) {
  445. canonicalLink.href = this.options.saveUrl;
  446. }
  447. }
  448. let size;
  449. if (this.options.displayStats) {
  450. size = util.getContentSize(this.doc.documentElement.outerHTML);
  451. }
  452. const content = util.serialize(this.doc, this.options.compressHTML);
  453. if (this.options.displayStats) {
  454. const contentSize = util.getContentSize(content);
  455. this.stats.set("processed", "HTML bytes", contentSize);
  456. this.stats.add("discarded", "HTML bytes", size - contentSize);
  457. }
  458. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  459. const replacementCharacter = this.options.filenameReplacementCharacter;
  460. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  461. if (!this.options.backgroundSave) {
  462. filename = filename.replace(/\//g, replacementCharacter);
  463. }
  464. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  465. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  466. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  467. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  468. filename = filename + "…" + extension;
  469. }
  470. if (!filename) {
  471. filename = "Unnamed page";
  472. }
  473. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  474. const pageData = {
  475. stats: this.stats.data,
  476. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  477. filename,
  478. content
  479. };
  480. if (this.options.addProof) {
  481. pageData.hash = await util.digest("SHA-256", content);
  482. }
  483. return pageData;
  484. }
  485. preProcessPage() {
  486. if (this.options.win) {
  487. 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));
  488. }
  489. if (this.options.images && !this.options.saveRawPage) {
  490. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  491. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  492. if (attributeValue) {
  493. const imageData = this.options.images[Number(attributeValue)];
  494. if (imageData) {
  495. if (this.options.removeHiddenElements && (
  496. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  497. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  498. )) {
  499. imgElement.setAttribute("src", EMPTY_IMAGE);
  500. } else {
  501. if (imageData.currentSrc) {
  502. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  503. imgElement.setAttribute("src", imageData.currentSrc);
  504. }
  505. if (this.options.loadDeferredImages) {
  506. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  507. imageData.src = imgElement.dataset.src;
  508. imgElement.setAttribute("src", imgElement.dataset.src);
  509. imgElement.removeAttribute("data-src");
  510. }
  511. }
  512. }
  513. }
  514. }
  515. });
  516. if (this.options.loadDeferredImages) {
  517. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  518. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  519. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  520. imgElement.removeAttribute("data-srcset");
  521. }
  522. });
  523. }
  524. }
  525. }
  526. replaceStyleContents() {
  527. if (this.options.stylesheets) {
  528. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  529. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  530. if (attributeValue) {
  531. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  532. if (stylesheetContent) {
  533. styleElement.textContent = stylesheetContent;
  534. }
  535. }
  536. });
  537. }
  538. }
  539. removeUnselectedElements() {
  540. removeUnmarkedElements(this.doc.body);
  541. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  542. function removeUnmarkedElements(element) {
  543. let selectedElementFound = false;
  544. Array.from(element.childNodes).forEach(node => {
  545. if (node.nodeType == 1) {
  546. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  547. selectedElementFound = selectedElementFound || isSelectedElement;
  548. if (isSelectedElement) {
  549. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  550. removeUnmarkedElements(node);
  551. } else if (selectedElementFound) {
  552. removeNode(node);
  553. } else {
  554. hideNode(node);
  555. }
  556. }
  557. });
  558. }
  559. function removeNode(node) {
  560. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  561. node.remove();
  562. } else {
  563. hideNode(node);
  564. }
  565. }
  566. function hideNode(node) {
  567. if (canHideNode(node)) {
  568. node.style.setProperty("display", "none", "important");
  569. Array.from(node.childNodes).forEach(removeNode);
  570. }
  571. }
  572. function canHideNode(node) {
  573. if (node.nodeType == 1) {
  574. const tagName = node.tagName && node.tagName.toLowerCase();
  575. return (tagName != "svg" && tagName != "style" && tagName != "link");
  576. }
  577. }
  578. }
  579. insertVideoPosters() {
  580. if (this.options.posters) {
  581. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  582. let videoElement;
  583. if (element.tagName == "VIDEO") {
  584. videoElement = element;
  585. } else {
  586. videoElement = element.parentElement;
  587. }
  588. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  589. if (attributeValue) {
  590. const posterURL = this.options.posters[Number(attributeValue)];
  591. if (!videoElement.poster && posterURL) {
  592. videoElement.setAttribute("poster", posterURL);
  593. }
  594. }
  595. });
  596. }
  597. }
  598. removeFrames() {
  599. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  600. this.stats.set("discarded", "frames", frameElements.length);
  601. this.stats.set("processed", "frames", frameElements.length);
  602. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  603. }
  604. removeScripts() {
  605. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  606. this.doc.querySelectorAll("[href]").forEach(element => {
  607. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  608. element.setAttribute("href", "");
  609. }
  610. });
  611. this.doc.querySelectorAll("[src]").forEach(element => {
  612. if (element.src && element.src.match(/^\s*javascript:/)) {
  613. element.removeAttribute("src");
  614. }
  615. });
  616. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  617. this.stats.set("discarded", "scripts", scriptElements.length);
  618. this.stats.set("processed", "scripts", scriptElements.length);
  619. scriptElements.forEach(element => element.remove());
  620. }
  621. removeVideoSources() {
  622. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  623. this.stats.set("discarded", "video sources", videoSourceElements.length);
  624. this.stats.set("processed", "video sources", videoSourceElements.length);
  625. videoSourceElements.forEach(element => {
  626. if (element.tagName == "SOURCE") {
  627. element.remove();
  628. } else {
  629. videoSourceElements.forEach(element => element.removeAttribute("src"));
  630. }
  631. });
  632. }
  633. removeAudioSources() {
  634. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  635. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  636. this.stats.set("processed", "audio sources", audioSourceElements.length);
  637. audioSourceElements.forEach(element => {
  638. if (element.tagName == "SOURCE") {
  639. element.remove();
  640. } else {
  641. audioSourceElements.forEach(element => element.removeAttribute("src"));
  642. }
  643. });
  644. }
  645. removeDiscardedResources() {
  646. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  647. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  648. 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\"])");
  649. this.stats.set("discarded", "objects", objectElements.length);
  650. this.stats.set("processed", "objects", objectElements.length);
  651. objectElements.forEach(element => element.remove());
  652. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  653. replacedAttributeValue.forEach(element => {
  654. let regExp;
  655. if (this.options.removeScripts) {
  656. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  657. } else {
  658. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  659. }
  660. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  661. if (relValue.length) {
  662. element.setAttribute("rel", relValue);
  663. } else {
  664. element.remove();
  665. }
  666. });
  667. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  668. if (this.options.compressHTML) {
  669. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  670. }
  671. if (!this.options.saveFavicon) {
  672. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  673. }
  674. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  675. }
  676. resetCharsetMeta() {
  677. let charset;
  678. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  679. const charsetDeclaration = element.content.split(";")[1];
  680. if (charsetDeclaration && !charset) {
  681. charset = charsetDeclaration.split("=")[1];
  682. if (charset) {
  683. this.charset = charset.trim().toLowerCase();
  684. }
  685. }
  686. element.remove();
  687. });
  688. const metaElement = this.doc.createElement("meta");
  689. metaElement.setAttribute("charset", "utf-8");
  690. if (this.doc.head.firstChild) {
  691. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  692. } else {
  693. this.doc.head.appendChild(metaElement);
  694. }
  695. }
  696. setInputValues() {
  697. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  698. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  699. input.setAttribute("value", value || "");
  700. });
  701. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  702. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  703. if (value == "true") {
  704. input.setAttribute("checked", "");
  705. }
  706. });
  707. this.doc.querySelectorAll("textarea").forEach(textarea => {
  708. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  709. textarea.textContent = value || "";
  710. });
  711. this.doc.querySelectorAll("select").forEach(select => {
  712. select.querySelectorAll("option").forEach(option => {
  713. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  714. if (selected) {
  715. option.setAttribute("selected", "");
  716. }
  717. });
  718. });
  719. }
  720. saveFavicon() {
  721. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  722. if (!faviconElement) {
  723. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  724. }
  725. if (!faviconElement) {
  726. faviconElement = this.doc.createElement("link");
  727. faviconElement.setAttribute("type", "image/x-icon");
  728. faviconElement.setAttribute("rel", "shortcut icon");
  729. faviconElement.setAttribute("href", "/favicon.ico");
  730. }
  731. this.doc.head.appendChild(faviconElement);
  732. }
  733. replaceCanvasElements() {
  734. if (this.options.canvases) {
  735. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  736. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  737. if (attributeValue) {
  738. const canvasData = this.options.canvases[Number(attributeValue)];
  739. if (canvasData) {
  740. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  741. this.stats.add("processed", "canvas", 1);
  742. }
  743. }
  744. });
  745. }
  746. }
  747. insertFonts() {
  748. if (this.options.fonts && this.options.fonts.length) {
  749. let stylesheetContent = "";
  750. this.options.fonts.forEach(fontData => {
  751. if (fontData["font-family"] && fontData.src) {
  752. stylesheetContent += "@font-face{";
  753. let stylesContent = "";
  754. Object.keys(fontData).forEach(fontStyle => {
  755. if (stylesContent) {
  756. stylesContent += ";";
  757. }
  758. stylesContent += fontStyle + ":" + fontData[fontStyle];
  759. });
  760. stylesheetContent += stylesContent + "}";
  761. }
  762. });
  763. if (stylesheetContent) {
  764. const styleElement = this.doc.createElement("style");
  765. styleElement.textContent = stylesheetContent;
  766. const existingStyleElement = this.doc.querySelector("style");
  767. if (existingStyleElement) {
  768. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  769. } else {
  770. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  771. }
  772. }
  773. }
  774. }
  775. removeHiddenElements() {
  776. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  777. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  778. this.stats.set("discarded", "hidden elements", removedElements.length);
  779. this.stats.set("processed", "hidden elements", removedElements.length);
  780. if (hiddenElements.length) {
  781. const styleElement = this.doc.createElement("style");
  782. styleElement.textContent = ".sf-hidden{display:none!important;}";
  783. this.doc.head.appendChild(styleElement);
  784. hiddenElements.forEach(element => element.classList.add("sf-hidden"));
  785. }
  786. removedElements.forEach(element => element.remove());
  787. }
  788. resolveHrefs() {
  789. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  790. const href = element.getAttribute("href").trim();
  791. if (!Util.testIgnoredPath(href)) {
  792. let resolvedURL;
  793. try {
  794. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  795. } catch (error) {
  796. // ignored
  797. }
  798. if (resolvedURL) {
  799. const url = Util.normalizeURL(this.options.url);
  800. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  801. resolvedURL = resolvedURL.substring(url.length);
  802. }
  803. try {
  804. element.setAttribute("href", resolvedURL);
  805. } catch (error) {
  806. // ignored
  807. }
  808. }
  809. }
  810. });
  811. }
  812. resolveStyleAttributeURLs() {
  813. this.doc.querySelectorAll("[style]").forEach(element => {
  814. let styleContent = element.getAttribute("style");
  815. if (this.options.compressCSS) {
  816. styleContent = util.compressCSS(styleContent);
  817. }
  818. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  819. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  820. this.styles.set(element, declarationList);
  821. });
  822. }
  823. async resolveStylesheetURLs() {
  824. const options = {
  825. maxResourceSize: this.options.maxResourceSize,
  826. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  827. url: this.options.url,
  828. charset: this.charset,
  829. compressCSS: this.options.compressCSS,
  830. updatedResources: this.options.updatedResources,
  831. rootDocument: this.options.rootDocument,
  832. frameId: this.options.windowId
  833. };
  834. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  835. .map(async element => {
  836. let mediaText;
  837. if (element.media) {
  838. mediaText = element.media.toLowerCase();
  839. }
  840. const stylesheetInfo = { mediaText };
  841. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  842. stylesheetInfo.scoped = true;
  843. }
  844. if (element.tagName.toLowerCase() == "link") {
  845. if (element.charset) {
  846. options.charset = element.charset;
  847. }
  848. }
  849. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  850. }));
  851. if (options.rootDocument) {
  852. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  853. await Promise.all(newResources.map(async resource => {
  854. resource.retrieved = true;
  855. const stylesheetInfo = {};
  856. const element = this.doc.createElement("style");
  857. this.doc.body.appendChild(element);
  858. element.textContent = resource.content;
  859. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  860. }));
  861. }
  862. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  863. stylesheets.set(element, stylesheetInfo);
  864. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  865. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  866. if (match && match[1] && match[1] != options.charset) {
  867. options.charset = match[1];
  868. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  869. }
  870. let stylesheet;
  871. try {
  872. stylesheet = cssTree.parse(Util.removeCssComments(stylesheetContent));
  873. } catch (error) {
  874. // ignored
  875. }
  876. if (stylesheet && stylesheet.children) {
  877. if (options.compressCSS) {
  878. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  879. }
  880. stylesheetInfo.stylesheet = stylesheet;
  881. } else {
  882. stylesheets.delete(element);
  883. }
  884. }
  885. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  886. let content;
  887. if (element.tagName.toLowerCase() == "link") {
  888. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  889. } else {
  890. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  891. }
  892. return content || "";
  893. }
  894. }
  895. async resolveFrameURLs() {
  896. if (!this.options.saveRawPage && this.options.frames) {
  897. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  898. await Promise.all(frameElements.map(async frameElement => {
  899. if (frameElement.tagName == "OBJECT") {
  900. frameElement.setAttribute("data", "data:text/html,");
  901. } else {
  902. frameElement.removeAttribute("src");
  903. frameElement.removeAttribute("srcdoc");
  904. }
  905. Array.from(frameElement.childNodes).forEach(node => node.remove());
  906. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  907. if (frameWindowId) {
  908. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  909. if (frameData) {
  910. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  911. }
  912. }
  913. }));
  914. }
  915. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  916. options.insertSingleFileComment = false;
  917. options.insertCanonicalLink = false;
  918. options.saveFavicon = false;
  919. options.url = frameData.baseURI;
  920. options.windowId = frameWindowId;
  921. if (frameData.content) {
  922. options.content = frameData.content;
  923. options.canvases = frameData.canvases;
  924. options.fonts = frameData.fonts;
  925. options.stylesheets = frameData.stylesheets;
  926. options.images = frameData.images;
  927. options.posters = frameData.posters;
  928. options.usedFonts = frameData.usedFonts;
  929. options.shadowRoots = frameData.shadowRoots;
  930. options.imports = frameData.imports;
  931. frameData.runner = new Runner(options);
  932. frameData.frameElement = frameElement;
  933. await frameData.runner.loadPage();
  934. await frameData.runner.initialize();
  935. frameData.maxResources = batchRequest.getMaxResources();
  936. }
  937. }
  938. }
  939. insertShadowRootContents() {
  940. const doc = this.doc;
  941. const options = this.options;
  942. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  943. processElement(this.doc);
  944. if (this.options.removeScripts) {
  945. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  946. }
  947. const scriptElement = doc.createElement("script");
  948. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  949. 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)}})}})()`;
  950. doc.body.appendChild(scriptElement);
  951. }
  952. function processElement(element) {
  953. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  954. shadowRootElements.forEach(element => {
  955. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  956. if (attributeValue) {
  957. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  958. if (shadowRootData) {
  959. const templateElement = doc.createElement("template");
  960. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  961. if (shadowRootData.delegatesFocus) {
  962. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  963. }
  964. if (shadowRootData.adoptedStyleSheets) {
  965. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  966. const styleElement = doc.createElement("style");
  967. styleElement.textContent = stylesheetContent;
  968. templateElement.appendChild(styleElement);
  969. });
  970. }
  971. const shadowDoc = util.parseDocContent(shadowRootData.content);
  972. if (shadowDoc.head) {
  973. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  974. if (metaCharset) {
  975. metaCharset.remove();
  976. }
  977. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  978. }
  979. if (shadowDoc.body) {
  980. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  981. }
  982. processElement(templateElement);
  983. if (element.firstChild) {
  984. element.insertBefore(templateElement, element.firstChild);
  985. } else {
  986. element.appendChild(templateElement);
  987. }
  988. }
  989. }
  990. });
  991. }
  992. }
  993. async resolveHtmlImportURLs() {
  994. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  995. await Promise.all(linkElements.map(async linkElement => {
  996. const resourceURL = linkElement.href;
  997. linkElement.removeAttribute("href");
  998. const options = Object.create(this.options);
  999. options.insertSingleFileComment = false;
  1000. options.insertCanonicalLink = false;
  1001. options.saveFavicon = false;
  1002. options.removeUnusedStyles = false;
  1003. options.removeAlternativeMedias = false;
  1004. options.removeUnusedFonts = false;
  1005. options.removeHiddenElements = false;
  1006. options.url = resourceURL;
  1007. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1008. if (attributeValue) {
  1009. const importData = options.imports[Number(attributeValue)];
  1010. if (importData) {
  1011. options.content = importData.content;
  1012. importData.runner = new Runner(options);
  1013. await importData.runner.loadPage();
  1014. await importData.runner.initialize();
  1015. if (!options.removeImports) {
  1016. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1017. }
  1018. importData.runner.getStyleSheets().forEach(stylesheet => {
  1019. const importedStyleElement = this.doc.createElement("style");
  1020. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1021. this.stylesheets.set(importedStyleElement, stylesheet);
  1022. });
  1023. }
  1024. }
  1025. if (options.removeImports) {
  1026. linkElement.remove();
  1027. this.stats.add("discarded", "HTML imports", 1);
  1028. }
  1029. }));
  1030. }
  1031. removeUnusedStyles() {
  1032. if (!this.mediaAllInfo) {
  1033. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1034. }
  1035. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1036. this.stats.set("processed", "CSS rules", stats.processed);
  1037. this.stats.set("discarded", "CSS rules", stats.discarded);
  1038. }
  1039. removeUnusedFonts() {
  1040. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1041. }
  1042. removeAlternativeMedias() {
  1043. const stats = util.minifyMedias(this.stylesheets);
  1044. this.stats.set("processed", "medias", stats.processed);
  1045. this.stats.set("discarded", "medias", stats.discarded);
  1046. }
  1047. async processStylesheets() {
  1048. this.options.fontURLs = new Map();
  1049. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1050. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1051. ));
  1052. }
  1053. async processStyleAttributes() {
  1054. return Promise.all([...this.styles].map(([, declarationList]) =>
  1055. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1056. ));
  1057. }
  1058. async processPageResources() {
  1059. const processAttributeArgs = [
  1060. ["link[href][rel*=\"icon\"]", "href", false, true],
  1061. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1062. ["img[src], input[src][type=image]", "src", true],
  1063. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1064. ["video[poster]", "poster"],
  1065. ["*[background]", "background"],
  1066. ["image", "xlink:href"],
  1067. ["image", "href"]
  1068. ];
  1069. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1070. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1071. );
  1072. resourcePromises = resourcePromises.concat([
  1073. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1074. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1075. ]);
  1076. if (!this.options.removeAudioSrc) {
  1077. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1078. }
  1079. if (!this.options.removeVideoSrc) {
  1080. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1081. }
  1082. await Promise.all(resourcePromises);
  1083. if (this.options.saveFavicon) {
  1084. ProcessorHelper.processShortcutIcons(this.doc);
  1085. }
  1086. }
  1087. async processScripts() {
  1088. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1089. let resourceURL;
  1090. let scriptSrc;
  1091. if (element.tagName == "SCRIPT") {
  1092. scriptSrc = element.getAttribute("src");
  1093. element.removeAttribute("src");
  1094. } else {
  1095. scriptSrc = element.getAttribute("href");
  1096. element.removeAttribute("href");
  1097. }
  1098. element.removeAttribute("integrity");
  1099. element.textContent = "";
  1100. try {
  1101. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1102. } catch (error) {
  1103. // ignored
  1104. }
  1105. if (Util.testValidURL(resourceURL)) {
  1106. this.stats.add("processed", "scripts", 1);
  1107. const content = await util.getContent(resourceURL, {
  1108. asBinary: true,
  1109. maxResourceSize: this.options.maxResourceSize,
  1110. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1111. frameId: this.options.windowId
  1112. });
  1113. content.data = Util.getUpdatedResourceContent(resourceURL, content, this.options);
  1114. if (element.tagName == "SCRIPT") {
  1115. element.setAttribute("src", content.data);
  1116. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1117. element.setAttribute("async", "");
  1118. }
  1119. } else {
  1120. const scriptElement = this.doc.createElement("script");
  1121. scriptElement.setAttribute("src", content.data);
  1122. scriptElement.setAttribute("async", "");
  1123. element.parentElement.replaceChild(scriptElement, element);
  1124. }
  1125. }
  1126. }));
  1127. }
  1128. removeAlternativeImages() {
  1129. util.removeAlternativeImages(this.doc);
  1130. }
  1131. async removeAlternativeFonts() {
  1132. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs);
  1133. }
  1134. async processFrames() {
  1135. if (this.options.frames) {
  1136. const frameElements = Array.from(this.doc.querySelectorAll("iframe:not([" + util.HIDDEN_FRAME_ATTRIBUTE_NAME + "]), frame, object[type=\"text/html\"][data]"));
  1137. await Promise.all(frameElements.map(async frameElement => {
  1138. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1139. if (frameWindowId) {
  1140. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1141. if (frameData) {
  1142. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1143. if (frameData.runner) {
  1144. this.stats.add("processed", "frames", 1);
  1145. await frameData.runner.run();
  1146. const pageData = await frameData.runner.getPageData();
  1147. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1148. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1149. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1150. sandbox += " allow-scripts allow-same-origin";
  1151. }
  1152. frameElement.setAttribute("sandbox", sandbox);
  1153. if (frameElement.tagName == "OBJECT") {
  1154. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1155. } else {
  1156. if (frameElement.tagName == "FRAME") {
  1157. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1158. } else {
  1159. frameElement.setAttribute("srcdoc", pageData.content);
  1160. frameElement.removeAttribute("src");
  1161. }
  1162. }
  1163. this.stats.addAll(pageData);
  1164. } else {
  1165. this.stats.add("discarded", "frames", 1);
  1166. }
  1167. }
  1168. }
  1169. }));
  1170. }
  1171. }
  1172. async processHtmlImports() {
  1173. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1174. await Promise.all(linkElements.map(async linkElement => {
  1175. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1176. if (attributeValue) {
  1177. const importData = this.options.imports[Number(attributeValue)];
  1178. if (importData.runner) {
  1179. this.stats.add("processed", "HTML imports", 1);
  1180. await importData.runner.run();
  1181. const pageData = await importData.runner.getPageData();
  1182. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1183. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1184. this.stats.addAll(pageData);
  1185. } else {
  1186. this.stats.add("discarded", "HTML imports", 1);
  1187. }
  1188. }
  1189. }));
  1190. }
  1191. replaceStylesheets() {
  1192. this.doc.querySelectorAll("style").forEach(styleElement => {
  1193. const stylesheetInfo = this.stylesheets.get(styleElement);
  1194. if (stylesheetInfo) {
  1195. this.stylesheets.delete(styleElement);
  1196. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1197. styleElement.textContent = stylesheetContent;
  1198. if (stylesheetInfo.mediaText) {
  1199. styleElement.media = stylesheetInfo.mediaText;
  1200. }
  1201. } else {
  1202. styleElement.remove();
  1203. }
  1204. });
  1205. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1206. const stylesheetInfo = this.stylesheets.get(linkElement);
  1207. if (stylesheetInfo) {
  1208. this.stylesheets.delete(linkElement);
  1209. const styleElement = this.doc.createElement("style");
  1210. if (stylesheetInfo.mediaText) {
  1211. styleElement.media = stylesheetInfo.mediaText;
  1212. }
  1213. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1214. styleElement.textContent = stylesheetContent;
  1215. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1216. } else {
  1217. linkElement.remove();
  1218. }
  1219. });
  1220. }
  1221. replaceStyleAttributes() {
  1222. this.doc.querySelectorAll("[style]").forEach(element => {
  1223. const declarations = this.styles.get(element);
  1224. if (declarations) {
  1225. this.styles.delete(element);
  1226. let styleContent = cssTree.generate(declarations);
  1227. element.setAttribute("style", styleContent);
  1228. } else {
  1229. element.setAttribute("style", "");
  1230. }
  1231. });
  1232. }
  1233. insertVariables() {
  1234. if (this.cssVariables.size) {
  1235. const styleElement = this.doc.createElement("style");
  1236. const firstStyleElement = this.doc.head.querySelector("style");
  1237. if (firstStyleElement) {
  1238. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1239. } else {
  1240. this.doc.head.appendChild(styleElement);
  1241. }
  1242. let stylesheetContent = "";
  1243. this.cssVariables.forEach((content, indexResource) => {
  1244. this.cssVariables.delete(indexResource);
  1245. if (stylesheetContent) {
  1246. stylesheetContent += ";";
  1247. }
  1248. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1249. });
  1250. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1251. }
  1252. }
  1253. compressHTML() {
  1254. let size;
  1255. if (this.options.displayStats) {
  1256. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1257. }
  1258. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1259. if (this.options.displayStats) {
  1260. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1261. }
  1262. }
  1263. cleanupPage() {
  1264. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1265. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1266. if (metaCharset) {
  1267. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1268. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1269. this.doc.head.querySelector("meta[charset]").remove();
  1270. }
  1271. }
  1272. }
  1273. async insertMAFFMetaData() {
  1274. const maffMetaData = await this.maffMetaDataPromise;
  1275. if (maffMetaData && maffMetaData.content) {
  1276. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1277. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1278. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1279. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1280. if (originalURLElement) {
  1281. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1282. }
  1283. if (archiveTimeElement) {
  1284. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1285. if (value) {
  1286. const date = new Date(value);
  1287. if (!isNaN(date.getTime())) {
  1288. this.options.saveDate = new Date(value);
  1289. }
  1290. }
  1291. }
  1292. }
  1293. }
  1294. async setDocInfo() {
  1295. const titleElement = this.doc.querySelector("title");
  1296. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1297. const authorElement = this.doc.querySelector("meta[name=author]");
  1298. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1299. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1300. const headingElement = this.doc.querySelector("h1");
  1301. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1302. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1303. this.options.info = {
  1304. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1305. lang: this.doc.documentElement.lang,
  1306. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1307. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1308. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1309. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1310. };
  1311. }
  1312. }
  1313. // ---------------
  1314. // ProcessorHelper
  1315. // ---------------
  1316. const DATA_URI_PREFIX = "data:";
  1317. const ABOUT_BLANK_URI = "about:blank";
  1318. const EMPTY_DATA_URI = "data:null;base64,";
  1319. const REGEXP_URL_HASH = /(#.+?)$/;
  1320. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1321. const VAR_MAX_SIZE = 1024 * 1024 * 2;
  1322. class ProcessorHelper {
  1323. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1324. const date = options.saveDate;
  1325. const url = util.parseURL(options.saveUrl);
  1326. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1327. template = await Util.evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1328. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1329. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1330. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1331. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1332. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1333. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1334. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1335. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1336. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1337. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1338. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1339. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1340. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1341. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1342. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1343. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1344. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1345. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1346. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1347. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1348. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1349. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1350. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1351. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1352. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1353. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1354. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await Util.evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await Util.evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1357. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1358. template = await Util.evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1359. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1360. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1361. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1362. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1363. for (const [name, value] of params) {
  1364. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1365. }
  1366. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1367. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1368. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1370. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decode(Util.getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1371. if (content) {
  1372. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1375. }
  1376. return template.trim();
  1377. function decode(value) {
  1378. try {
  1379. return decodeURI(value);
  1380. } catch (error) {
  1381. return value;
  1382. }
  1383. }
  1384. }
  1385. static setBackgroundImage(element, url, style) {
  1386. element.style.setProperty("background-blend-mode", "normal", "important");
  1387. element.style.setProperty("background-clip", "content-box", "important");
  1388. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1389. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1390. element.style.setProperty("background-image", url, "important");
  1391. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1392. element.style.setProperty("background-origin", "content-box", "important");
  1393. element.style.setProperty("background-repeat", "no-repeat", "important");
  1394. }
  1395. static processShortcutIcons(doc) {
  1396. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1397. if (!shortcutIcon) {
  1398. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1399. if (shortcutIcon) {
  1400. shortcutIcon.rel = "icon";
  1401. }
  1402. }
  1403. if (shortcutIcon) {
  1404. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1405. if (linkElement != shortcutIcon) {
  1406. linkElement.remove();
  1407. }
  1408. });
  1409. }
  1410. }
  1411. static removeSingleLineCssComments(stylesheet) {
  1412. const removedRules = [];
  1413. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1414. const ruleData = cssRule.data;
  1415. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1416. removedRules.push(cssRule);
  1417. }
  1418. }
  1419. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1420. }
  1421. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1422. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1423. const imports = Util.getImportFunctions(stylesheetContent);
  1424. await Promise.all(imports.map(async cssImport => {
  1425. const match = Util.matchImport(cssImport);
  1426. if (match) {
  1427. const regExpCssImport = Util.getRegExp(cssImport);
  1428. let resourceURL = Util.normalizeURL(match.resourceURL);
  1429. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1430. try {
  1431. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1432. } catch (error) {
  1433. // ignored
  1434. }
  1435. if (Util.testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1436. const content = await util.getContent(resourceURL, {
  1437. maxResourceSize: options.maxResourceSize,
  1438. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1439. validateTextContentType: true,
  1440. frameId: options.windowId
  1441. });
  1442. resourceURL = content.resourceURL;
  1443. content.data = Util.getUpdatedResourceContent(resourceURL, content, options);
  1444. let importedStylesheetContent = Util.removeCssComments(content.data);
  1445. if (options.compressCSS) {
  1446. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1447. }
  1448. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1449. if (stylesheetContent.includes(cssImport)) {
  1450. const ancestorStyleSheets = new Set(importedStyleSheets);
  1451. ancestorStyleSheets.add(resourceURL);
  1452. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1453. workStylesheet.textContent = importedStylesheetContent;
  1454. if (workStylesheet.sheet.cssRules.length) {
  1455. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1456. } else {
  1457. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1458. }
  1459. }
  1460. } else {
  1461. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1462. }
  1463. } else {
  1464. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1465. }
  1466. }
  1467. }));
  1468. return stylesheetContent;
  1469. }
  1470. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1471. const urlFunctions = Util.getUrlFunctions(stylesheetContent, true);
  1472. urlFunctions.map(urlFunction => {
  1473. const originalResourceURL = Util.matchURL(urlFunction);
  1474. const resourceURL = Util.normalizeURL(originalResourceURL);
  1475. if (!Util.testIgnoredPath(resourceURL)) {
  1476. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1477. let resolvedURL;
  1478. if (!originalResourceURL.startsWith("#")) {
  1479. try {
  1480. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1481. } catch (error) {
  1482. // ignored
  1483. }
  1484. }
  1485. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1486. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1487. }
  1488. } else {
  1489. let newUrlFunction;
  1490. if (originalResourceURL) {
  1491. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1492. } else {
  1493. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1494. }
  1495. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1496. }
  1497. }
  1498. });
  1499. return stylesheetContent;
  1500. }
  1501. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1502. resourceURL = Util.normalizeURL(resourceURL);
  1503. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1504. const content = await util.getContent(resourceURL, {
  1505. maxResourceSize: options.maxResourceSize,
  1506. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1507. charset: options.charset,
  1508. frameId: options.windowId
  1509. });
  1510. resourceURL = content.resourceURL;
  1511. content.data = Util.getUpdatedResourceContent(content.resourceURL, content, options);
  1512. let stylesheetContent = Util.removeCssComments(content.data);
  1513. if (options.compressCSS) {
  1514. stylesheetContent = util.compressCSS(stylesheetContent);
  1515. }
  1516. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1517. return stylesheetContent;
  1518. }
  1519. }
  1520. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1521. const promises = [];
  1522. const removedRules = [];
  1523. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1524. const ruleData = cssRule.data;
  1525. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1526. removedRules.push(cssRule);
  1527. } else if (ruleData.block && ruleData.block.children) {
  1528. if (ruleData.type == "Rule") {
  1529. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1530. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1531. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1532. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1533. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1534. }
  1535. }
  1536. }
  1537. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1538. await Promise.all(promises);
  1539. async function processFontFaceRule(ruleData, fontURLs) {
  1540. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1541. if (declaration.type == "Declaration" && declaration.value.children) {
  1542. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value), true);
  1543. await Promise.all(urlFunctions.map(async urlFunction => {
  1544. const originalResourceURL = Util.matchURL(urlFunction);
  1545. const resourceURL = Util.normalizeURL(originalResourceURL);
  1546. if (!Util.testIgnoredPath(resourceURL)) {
  1547. if (Util.testValidURL(resourceURL)) {
  1548. let { content } = await batchRequest.addURL(resourceURL, true);
  1549. fontURLs.set(declaration, resourceURL);
  1550. replaceURLs(declaration, originalResourceURL, content);
  1551. }
  1552. }
  1553. }));
  1554. }
  1555. }));
  1556. function replaceURLs(declaration, oldURL, newURL) {
  1557. declaration.value.children.forEach(token => {
  1558. if (token.type == "Url" && util.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1559. token.value.value = newURL;
  1560. }
  1561. });
  1562. }
  1563. }
  1564. }
  1565. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1566. await Promise.all(declarations.map(async declaration => {
  1567. if (declaration.type == "Declaration" && declaration.value.children) {
  1568. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1569. await Promise.all(urlFunctions.map(async urlFunction => {
  1570. const originalResourceURL = Util.matchURL(urlFunction);
  1571. const resourceURL = Util.normalizeURL(originalResourceURL);
  1572. if (!Util.testIgnoredPath(resourceURL)) {
  1573. if (Util.testValidURL(resourceURL)) {
  1574. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1575. let variableDefined;
  1576. const tokens = [];
  1577. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1578. if (!originalResourceURL.startsWith("#")) {
  1579. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < VAR_MAX_SIZE) {
  1580. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1581. tokens.push({ parent, token, value });
  1582. variableDefined = true;
  1583. } else {
  1584. token.data.value.value = content;
  1585. }
  1586. }
  1587. });
  1588. if (variableDefined) {
  1589. cssVariables.set(indexResource, content);
  1590. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1591. }
  1592. }
  1593. }
  1594. }));
  1595. }
  1596. }));
  1597. function findURLToken(url, children, callback, depth = 0) {
  1598. for (let token = children.head; token; token = token.next) {
  1599. if (token.data.children) {
  1600. findURLToken(url, token.data.children, callback, depth + 1);
  1601. }
  1602. if (token.data.type == "Url" && util.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1603. callback(token, children, depth == 0);
  1604. }
  1605. }
  1606. }
  1607. }
  1608. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1609. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1610. let resourceURL = resourceElement.getAttribute(attributeName);
  1611. if (resourceURL != null) {
  1612. resourceURL = Util.normalizeURL(resourceURL);
  1613. if (!Util.testIgnoredPath(resourceURL)) {
  1614. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1615. if (Util.testValidPath(resourceURL)) {
  1616. try {
  1617. resourceURL = util.resolveURL(resourceURL, baseURI);
  1618. } catch (error) {
  1619. // ignored
  1620. }
  1621. if (Util.testValidURL(resourceURL)) {
  1622. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1623. if (resourceElement.dataset.singleFileOriginURL) {
  1624. let originURL = resourceElement.dataset.singleFileOriginURL;
  1625. delete resourceElement.dataset.singleFileOriginURL;
  1626. if (content == EMPTY_DATA_URI) {
  1627. try {
  1628. originURL = util.resolveURL(originURL, baseURI);
  1629. } catch (error) {
  1630. // ignored
  1631. }
  1632. try {
  1633. resourceURL = originURL;
  1634. content = (await util.getContent(resourceURL, {
  1635. asBinary: true,
  1636. maxResourceSize: options.maxResourceSize,
  1637. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1638. frameId: options.windowId
  1639. })).data;
  1640. } catch (error) {
  1641. // ignored
  1642. }
  1643. }
  1644. }
  1645. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1646. resourceElement.remove();
  1647. } else {
  1648. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1649. if (!forbiddenPrefixFound) {
  1650. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1651. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < VAR_MAX_SIZE) {
  1652. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1653. cssVariables.set(indexResource, content);
  1654. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1655. styles.set(resourceElement, declarationList);
  1656. } else {
  1657. resourceElement.setAttribute(attributeName, content);
  1658. }
  1659. } else {
  1660. resourceElement.setAttribute(attributeName, content);
  1661. }
  1662. }
  1663. }
  1664. }
  1665. }
  1666. }
  1667. }
  1668. }));
  1669. }
  1670. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1671. const attributeName = "xlink:href";
  1672. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1673. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1674. let resourceURL = Util.normalizeURL(originalResourceURL);
  1675. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1676. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1677. try {
  1678. resourceURL = util.resolveURL(resourceURL, baseURI);
  1679. } catch (error) {
  1680. // ignored
  1681. }
  1682. if (Util.testValidURL(resourceURL)) {
  1683. const { content } = await batchRequest.addURL(resourceURL);
  1684. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1685. if (hashMatch && hashMatch[0]) {
  1686. let symbolElement;
  1687. try {
  1688. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1689. } catch (error) {
  1690. // ignored
  1691. }
  1692. if (symbolElement) {
  1693. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1694. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1695. }
  1696. } else {
  1697. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1698. }
  1699. }
  1700. } else if (resourceURL == options.url) {
  1701. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1702. }
  1703. }));
  1704. }
  1705. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1706. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1707. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1708. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1709. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1710. if (!Util.testIgnoredPath(resourceURL)) {
  1711. if (Util.testValidPath(resourceURL)) {
  1712. try {
  1713. resourceURL = util.resolveURL(resourceURL, baseURI);
  1714. } catch (error) {
  1715. // ignored
  1716. }
  1717. if (Util.testValidURL(resourceURL)) {
  1718. const { content } = await batchRequest.addURL(resourceURL, true);
  1719. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1720. if (forbiddenPrefixFound) {
  1721. return "";
  1722. }
  1723. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1724. } else {
  1725. return "";
  1726. }
  1727. } else {
  1728. return "";
  1729. }
  1730. } else {
  1731. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1732. }
  1733. }));
  1734. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1735. }));
  1736. }
  1737. static replaceImageSource(imgElement, variableName, options) {
  1738. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1739. if (attributeValue) {
  1740. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1741. if (imageData && imageData.replaceable) {
  1742. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1743. const backgroundStyle = {};
  1744. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1745. if (backgroundSize) {
  1746. backgroundStyle["background-size"] = imageData.objectFit;
  1747. }
  1748. if (imageData.objectPosition) {
  1749. backgroundStyle["background-position"] = imageData.objectPosition;
  1750. }
  1751. if (imageData.backgroundColor) {
  1752. backgroundStyle["background-color"] = imageData.backgroundColor;
  1753. }
  1754. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1755. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1756. return true;
  1757. }
  1758. }
  1759. }
  1760. }
  1761. // ----
  1762. // Util
  1763. // ----
  1764. const BLOB_URI_PREFIX = "blob:";
  1765. const HTTP_URI_PREFIX = /^https?:\/\//;
  1766. const FILE_URI_PREFIX = /^file:\/\//;
  1767. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1768. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1769. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1770. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1771. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1772. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1773. 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;
  1774. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1775. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1776. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1777. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1778. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1779. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1780. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1781. class Util {
  1782. static getUpdatedResourceContent(resourceURL, content, options) {
  1783. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1784. options.updatedResources[resourceURL].retrieved = true;
  1785. return options.updatedResources[resourceURL].content;
  1786. } else {
  1787. return content.data || "";
  1788. }
  1789. }
  1790. static normalizeURL(url) {
  1791. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1792. return url;
  1793. } else {
  1794. return url.split("#")[0];
  1795. }
  1796. }
  1797. static getCSSValue(value) {
  1798. let result = "";
  1799. try {
  1800. result = cssTree.generate(value);
  1801. } catch (error) {
  1802. // ignored
  1803. }
  1804. return result;
  1805. }
  1806. static getOnEventAttributeNames(doc) {
  1807. const element = doc.createElement("div");
  1808. const attributeNames = [];
  1809. for (const propertyName in element) {
  1810. if (propertyName.startsWith("on")) {
  1811. attributeNames.push(propertyName);
  1812. }
  1813. }
  1814. attributeNames.push("onunload");
  1815. return attributeNames;
  1816. }
  1817. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1818. let maxLength;
  1819. if (template) {
  1820. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1821. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1822. if (template.match(replaceRegExp)) {
  1823. const matchedLength = template.match(replaceRegExp)[0];
  1824. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1825. if (isNaN(maxLength) || maxLength <= 0) {
  1826. maxLength = null;
  1827. }
  1828. } else {
  1829. replaceRegExp = new RegExp(regExpVariable, "g");
  1830. }
  1831. if (template.match(replaceRegExp)) {
  1832. let value = await valueGetter();
  1833. if (!dontReplaceSlash) {
  1834. value = value.replace(/\/+/g, replacementCharacter);
  1835. }
  1836. if (maxLength) {
  1837. value = await util.truncateText(value, maxLength);
  1838. }
  1839. return template.replace(replaceRegExp, value);
  1840. }
  1841. }
  1842. return template;
  1843. }
  1844. static getLastSegment(url, replacementCharacter) {
  1845. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1846. if (!lastSegment) {
  1847. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1848. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1849. }
  1850. if (!lastSegment) {
  1851. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1852. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1853. }
  1854. if (!lastSegment) {
  1855. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1856. }
  1857. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1858. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1859. lastSegment = lastSegmentMatch[1];
  1860. }
  1861. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1862. return lastSegment;
  1863. }
  1864. static getRegExp(string) {
  1865. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1866. }
  1867. static getUrlFunctions(stylesheetContent, unique) {
  1868. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1869. if (unique) {
  1870. return [...new Set(result)];
  1871. } else {
  1872. return result;
  1873. }
  1874. }
  1875. static getImportFunctions(stylesheetContent) {
  1876. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1877. }
  1878. static findShortcutIcon(shortcutIcons) {
  1879. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1880. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1881. return shortcutIcons[0];
  1882. }
  1883. static matchURL(stylesheetContent) {
  1884. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1885. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1886. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1887. return match && match[1];
  1888. }
  1889. static testIgnoredPath(resourceURL) {
  1890. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1891. }
  1892. static testValidPath(resourceURL) {
  1893. return resourceURL && !resourceURL.match(EMPTY_URL);
  1894. }
  1895. static testValidURL(resourceURL) {
  1896. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1897. }
  1898. static matchImport(stylesheetContent) {
  1899. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1900. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1901. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1902. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1903. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1904. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1905. if (match) {
  1906. const [, resourceURL, media] = match;
  1907. return { resourceURL, media };
  1908. }
  1909. }
  1910. static removeCssComments(stylesheetContent) {
  1911. let start, end;
  1912. do {
  1913. start = stylesheetContent.indexOf("/*");
  1914. end = stylesheetContent.indexOf("*/", start + 2);
  1915. if (start != -1 && end != -1) {
  1916. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1917. }
  1918. } while (start != -1 && end != -1);
  1919. return stylesheetContent;
  1920. }
  1921. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1922. if (mediaQuery) {
  1923. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1924. } else {
  1925. return stylesheetContent;
  1926. }
  1927. }
  1928. }
  1929. function log(...args) {
  1930. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1931. }
  1932. // -----
  1933. // Stats
  1934. // -----
  1935. const STATS_DEFAULT_VALUES = {
  1936. discarded: {
  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. processed: {
  1952. "HTML bytes": 0,
  1953. "hidden elements": 0,
  1954. "HTML imports": 0,
  1955. scripts: 0,
  1956. objects: 0,
  1957. "audio sources": 0,
  1958. "video sources": 0,
  1959. frames: 0,
  1960. "CSS rules": 0,
  1961. canvas: 0,
  1962. stylesheets: 0,
  1963. resources: 0,
  1964. medias: 0
  1965. }
  1966. };
  1967. class Stats {
  1968. constructor(options) {
  1969. this.options = options;
  1970. if (options.displayStats) {
  1971. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1972. }
  1973. }
  1974. set(type, subType, value) {
  1975. if (this.options.displayStats) {
  1976. this.data[type][subType] = value;
  1977. }
  1978. }
  1979. add(type, subType, value) {
  1980. if (this.options.displayStats) {
  1981. this.data[type][subType] += value;
  1982. }
  1983. }
  1984. addAll(pageData) {
  1985. if (this.options.displayStats) {
  1986. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1987. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1988. }
  1989. }
  1990. }
  1991. return { getClass };
  1992. })();