single-file-core.js 90 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. }
  34. async run() {
  35. if (this.options.userScriptEnabled) {
  36. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  37. }
  38. this.runner = new Runner(this.options, true);
  39. await this.runner.loadPage();
  40. await this.runner.initialize();
  41. if (this.options.userScriptEnabled) {
  42. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  43. }
  44. await this.runner.run();
  45. }
  46. cancel() {
  47. this.cancelled = true;
  48. if (this.runner) {
  49. this.runner.cancel();
  50. }
  51. }
  52. getPageData() {
  53. return this.runner.getPageData();
  54. }
  55. }
  56. // -------------
  57. // ProgressEvent
  58. // -------------
  59. const PAGE_LOADING = "page-loading";
  60. const PAGE_LOADED = "page-loaded";
  61. const RESOURCES_INITIALIZING = "resource-initializing";
  62. const RESOURCES_INITIALIZED = "resources-initialized";
  63. const RESOURCE_LOADED = "resource-loaded";
  64. const PAGE_ENDED = "page-ended";
  65. const STAGE_STARTED = "stage-started";
  66. const STAGE_ENDED = "stage-ended";
  67. const STAGE_TASK_STARTED = "stage-task-started";
  68. const STAGE_TASK_ENDED = "stage-task-ended";
  69. class ProgressEvent {
  70. constructor(type, detail) {
  71. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  72. }
  73. }
  74. // ------
  75. // Runner
  76. // ------
  77. const RESOLVE_URLS_STAGE = 0;
  78. const REPLACE_DATA_STAGE = 1;
  79. const REPLACE_DOCS_STAGE = 2;
  80. const POST_PROCESS_STAGE = 3;
  81. const STAGES = [{
  82. sequential: [
  83. { action: "preProcessPage" },
  84. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  85. { action: "replaceStyleContents" },
  86. { action: "resetCharsetMeta" },
  87. { option: "saveFavicon", action: "saveFavicon" },
  88. { action: "replaceCanvasElements" },
  89. { action: "insertFonts" },
  90. { action: "insertShadowRootContents" },
  91. { action: "setInputValues" },
  92. { option: "removeScripts", action: "removeScripts" },
  93. { option: "selected", action: "removeUnselectedElements" },
  94. { option: "removeVideoSrc", action: "insertVideoPosters" },
  95. { option: "removeFrames", action: "removeFrames" },
  96. { option: "removeVideoSrc", action: "removeVideoSources" },
  97. { option: "removeAudioSrc", action: "removeAudioSources" },
  98. { action: "removeDiscardedResources" },
  99. { option: "removeHiddenElements", action: "removeHiddenElements" },
  100. { action: "resolveHrefs" },
  101. { action: "resolveStyleAttributeURLs" }
  102. ],
  103. parallel: [
  104. { action: "resolveStylesheetURLs" },
  105. { option: "!removeFrames", action: "resolveFrameURLs" },
  106. { action: "resolveHtmlImportURLs" }
  107. ]
  108. }, {
  109. sequential: [
  110. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  111. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  112. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  113. ],
  114. parallel: [
  115. { action: "processStylesheets" },
  116. { action: "processStyleAttributes" },
  117. { action: "processPageResources" },
  118. { option: "!removeScripts", action: "processScripts" }
  119. ]
  120. }, {
  121. sequential: [
  122. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  123. ],
  124. parallel: [
  125. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  126. { option: "!removeFrames", action: "processFrames" },
  127. { option: "!removeImports", action: "processHtmlImports" },
  128. ]
  129. }, {
  130. sequential: [
  131. { action: "replaceStylesheets" },
  132. { action: "replaceStyleAttributes" },
  133. { action: "insertVariables" },
  134. { option: "compressHTML", action: "compressHTML" },
  135. { action: "cleanupPage" }
  136. ],
  137. parallel: [
  138. { option: "enableMaff", action: "insertMAFFMetaData" },
  139. { action: "setDocInfo" }
  140. ]
  141. }];
  142. class Runner {
  143. constructor(options, root) {
  144. const rootDocDefined = root && options.doc;
  145. this.root = root;
  146. this.options = options;
  147. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  148. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  149. this.options.rootDocument = root;
  150. this.options.updatedResources = this.options.updatedResources || {};
  151. this.options.fontTests = new Map();
  152. this.batchRequest = new BatchRequest();
  153. this.processor = new Processor(options, this.batchRequest);
  154. if (rootDocDefined) {
  155. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  156. this.options.canvases = docData.canvases;
  157. this.options.fonts = docData.fonts;
  158. this.options.stylesheets = docData.stylesheets;
  159. this.options.images = docData.images;
  160. this.options.posters = docData.posters;
  161. this.options.usedFonts = docData.usedFonts;
  162. this.options.shadowRoots = docData.shadowRoots;
  163. this.options.imports = docData.imports;
  164. this.options.referrer = docData.referrer;
  165. this.markedElements = docData.markedElements;
  166. }
  167. if (this.options.saveRawPage) {
  168. this.options.removeFrames = true;
  169. }
  170. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  171. this.onprogress = options.onprogress || (() => { });
  172. }
  173. async loadPage() {
  174. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  175. await this.processor.loadPage(this.options.content);
  176. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  177. }
  178. async initialize() {
  179. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  180. await this.executeStage(RESOLVE_URLS_STAGE);
  181. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  182. if (this.root && this.options.doc) {
  183. util.postProcessDoc(this.options.doc, this.markedElements);
  184. }
  185. }
  186. cancel() {
  187. this.cancelled = true;
  188. this.batchRequest.cancel();
  189. if (this.root) {
  190. if (this.options.frames) {
  191. this.options.frames.forEach(cancelRunner);
  192. }
  193. if (this.options.imports) {
  194. this.options.imports.forEach(cancelRunner);
  195. }
  196. }
  197. function cancelRunner(resourceData) {
  198. if (resourceData.runner) {
  199. resourceData.runner.cancel();
  200. }
  201. }
  202. }
  203. async run() {
  204. if (this.root) {
  205. this.processor.initialize(this.batchRequest);
  206. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  207. }
  208. await this.batchRequest.run(detail => {
  209. detail.pageURL = this.options.url;
  210. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  211. }, this.options);
  212. await this.pendingPromises;
  213. this.options.doc = null;
  214. this.options.win = null;
  215. await this.executeStage(REPLACE_DOCS_STAGE);
  216. await this.executeStage(POST_PROCESS_STAGE);
  217. this.processor.finalize();
  218. }
  219. getDocument() {
  220. return this.processor.doc;
  221. }
  222. getStyleSheets() {
  223. return this.processor.stylesheets;
  224. }
  225. getPageData() {
  226. if (this.root) {
  227. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  228. }
  229. return this.processor.getPageData();
  230. }
  231. async executeStage(step) {
  232. if (DEBUG) {
  233. log("**** STARTED STAGE", step, "****");
  234. }
  235. const frame = !this.root;
  236. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  237. STAGES[step].sequential.forEach(task => {
  238. let startTime;
  239. if (DEBUG) {
  240. startTime = Date.now();
  241. log(" -- STARTED task =", task.action);
  242. }
  243. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  244. if (!this.cancelled) {
  245. this.executeTask(task);
  246. }
  247. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  248. if (DEBUG) {
  249. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  250. }
  251. });
  252. let parallelTasksPromise;
  253. if (STAGES[step].parallel) {
  254. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  255. let startTime;
  256. if (DEBUG) {
  257. startTime = Date.now();
  258. log(" // STARTED task =", task.action);
  259. }
  260. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  261. if (!this.cancelled) {
  262. await this.executeTask(task);
  263. }
  264. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  265. if (DEBUG) {
  266. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  267. }
  268. }));
  269. } else {
  270. parallelTasksPromise = Promise.resolve();
  271. }
  272. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  273. if (DEBUG) {
  274. log("**** ENDED STAGE", step, "****");
  275. }
  276. return parallelTasksPromise;
  277. }
  278. executeTask(task) {
  279. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  280. return this.processor[task.action]();
  281. }
  282. }
  283. }
  284. // ------------
  285. // BatchRequest
  286. // ------------
  287. class BatchRequest {
  288. constructor() {
  289. this.requests = new Map();
  290. this.duplicates = new Map();
  291. }
  292. addURL(resourceURL, asBinary, expectedType, groupDuplicates) {
  293. return new Promise((resolve, reject) => {
  294. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType]);
  295. let resourceRequests = this.requests.get(requestKey);
  296. if (!resourceRequests) {
  297. resourceRequests = [];
  298. this.requests.set(requestKey, resourceRequests);
  299. }
  300. const callbacks = { resolve, reject };
  301. resourceRequests.push(callbacks);
  302. if (groupDuplicates) {
  303. let duplicateRequests = this.duplicates.get(requestKey);
  304. if (!duplicateRequests) {
  305. duplicateRequests = [];
  306. this.duplicates.set(requestKey, duplicateRequests);
  307. }
  308. duplicateRequests.push(callbacks);
  309. }
  310. });
  311. }
  312. getMaxResources() {
  313. return this.requests.size;
  314. }
  315. run(onloadListener, options) {
  316. const resourceURLs = [...this.requests.keys()];
  317. let indexResource = 0;
  318. return Promise.all(resourceURLs.map(async requestKey => {
  319. const [resourceURL, asBinary, expectedType] = JSON.parse(requestKey);
  320. const resourceRequests = this.requests.get(requestKey);
  321. try {
  322. const currentIndexResource = indexResource;
  323. indexResource = indexResource + 1;
  324. const content = await util.getContent(resourceURL, {
  325. asBinary,
  326. expectedType,
  327. maxResourceSize: options.maxResourceSize,
  328. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  329. frameId: options.windowId
  330. });
  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: currentIndexResource, duplicate });
  337. });
  338. }
  339. } catch (error) {
  340. indexResource = indexResource + 1;
  341. onloadListener({ url: resourceURL });
  342. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  343. }
  344. this.requests.delete(requestKey);
  345. }));
  346. }
  347. cancel() {
  348. this.cancelled = true;
  349. const resourceURLs = [...this.requests.keys()];
  350. resourceURLs.forEach(requestKey => {
  351. const resourceRequests = this.requests.get(requestKey);
  352. resourceRequests.forEach(callbacks => callbacks.reject());
  353. this.requests.delete(requestKey);
  354. });
  355. }
  356. }
  357. // ---------
  358. // Processor
  359. // ---------
  360. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  361. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  362. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  363. const SCRIPT_TAG_FOUND = /<script/gi;
  364. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  365. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  366. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  367. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  368. const UTF8_CHARSET = "utf-8";
  369. class Processor {
  370. constructor(options, batchRequest) {
  371. this.options = options;
  372. this.stats = new Stats(options);
  373. this.baseURI = normalizeURL(options.baseURI || options.url);
  374. this.batchRequest = batchRequest;
  375. this.stylesheets = new Map();
  376. this.styles = new Map();
  377. this.cssVariables = new Map();
  378. this.fontTests = options.fontTests;
  379. }
  380. initialize() {
  381. this.options.saveDate = new Date();
  382. this.options.saveUrl = this.options.url;
  383. if (this.options.enableMaff) {
  384. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url));
  385. }
  386. this.maxResources = this.batchRequest.getMaxResources();
  387. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  388. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  389. }
  390. if (!this.options.removeImports && this.options.imports) {
  391. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  392. }
  393. this.stats.set("processed", "resources", this.maxResources);
  394. }
  395. async loadPage(pageContent, charset) {
  396. let content;
  397. if (!pageContent || this.options.saveRawPage) {
  398. content = await util.getContent(this.baseURI, {
  399. maxResourceSize: this.options.maxResourceSize,
  400. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  401. charset,
  402. frameId: this.options.windowId
  403. });
  404. pageContent = content.data;
  405. }
  406. this.doc = util.parseDocContent(pageContent, this.baseURI);
  407. if (this.options.saveRawPage) {
  408. let charset;
  409. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  410. const charsetDeclaration = element.content.split(";")[1];
  411. if (charsetDeclaration && !charset) {
  412. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  413. }
  414. });
  415. if (charset && charset != content.charset) {
  416. return this.loadPage(pageContent, charset);
  417. }
  418. }
  419. this.workStyleElement = this.doc.createElement("style");
  420. this.doc.body.appendChild(this.workStyleElement);
  421. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  422. }
  423. finalize() {
  424. if (this.workStyleElement.parentNode) {
  425. this.workStyleElement.remove();
  426. }
  427. }
  428. async getPageData() {
  429. util.postProcessDoc(this.doc);
  430. const url = util.parseURL(this.baseURI);
  431. if (this.options.insertSingleFileComment) {
  432. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  433. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  434. " \n url: " + this.options.saveUrl +
  435. " \n saved date: " + this.options.saveDate +
  436. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  437. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  438. }
  439. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  440. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  441. if (!canonicalLink) {
  442. canonicalLink = this.doc.createElement("link");
  443. canonicalLink.setAttribute("rel", "canonical");
  444. this.doc.head.appendChild(canonicalLink);
  445. }
  446. if (canonicalLink && !canonicalLink.href) {
  447. canonicalLink.href = this.options.saveUrl;
  448. }
  449. }
  450. if (this.options.insertMetaNoIndex) {
  451. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  452. if (!metaElement) {
  453. metaElement = this.doc.createElement("meta");
  454. metaElement.setAttribute("name", "robots");
  455. metaElement.setAttribute("content", "noindex");
  456. this.doc.head.appendChild(metaElement);
  457. }
  458. }
  459. let size;
  460. if (this.options.displayStats) {
  461. size = util.getContentSize(this.doc.documentElement.outerHTML);
  462. }
  463. const content = util.serialize(this.doc, this.options.compressHTML);
  464. if (this.options.displayStats) {
  465. const contentSize = util.getContentSize(content);
  466. this.stats.set("processed", "HTML bytes", contentSize);
  467. this.stats.add("discarded", "HTML bytes", size - contentSize);
  468. }
  469. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  470. const replacementCharacter = this.options.filenameReplacementCharacter;
  471. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  472. if (!this.options.backgroundSave) {
  473. filename = filename.replace(/\//g, replacementCharacter);
  474. }
  475. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  476. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  477. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  478. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  479. filename = filename + "…" + extension;
  480. }
  481. if (!filename) {
  482. filename = "Unnamed page";
  483. }
  484. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  485. const pageData = {
  486. stats: this.stats.data,
  487. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  488. filename,
  489. content
  490. };
  491. if (this.options.addProof) {
  492. pageData.hash = await util.digest("SHA-256", content);
  493. }
  494. if (this.options.retrieveLinks) {
  495. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  496. }
  497. return pageData;
  498. }
  499. preProcessPage() {
  500. if (this.options.win) {
  501. 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));
  502. }
  503. if (this.options.images && !this.options.saveRawPage) {
  504. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  505. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  506. if (attributeValue) {
  507. const imageData = this.options.images[Number(attributeValue)];
  508. if (imageData) {
  509. if (this.options.removeHiddenElements && (
  510. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  511. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  512. )) {
  513. imgElement.setAttribute("src", EMPTY_IMAGE);
  514. } else {
  515. if (imageData.currentSrc) {
  516. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  517. imgElement.setAttribute("src", imageData.currentSrc);
  518. }
  519. if (this.options.loadDeferredImages) {
  520. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  521. imageData.src = imgElement.dataset.src;
  522. imgElement.setAttribute("src", imgElement.dataset.src);
  523. imgElement.removeAttribute("data-src");
  524. }
  525. }
  526. }
  527. }
  528. }
  529. });
  530. if (this.options.loadDeferredImages) {
  531. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  532. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  533. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  534. imgElement.removeAttribute("data-srcset");
  535. }
  536. });
  537. }
  538. }
  539. }
  540. replaceStyleContents() {
  541. if (this.options.stylesheets) {
  542. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  543. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  544. if (attributeValue) {
  545. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  546. if (stylesheetContent) {
  547. styleElement.textContent = stylesheetContent;
  548. }
  549. }
  550. });
  551. }
  552. }
  553. removeUnselectedElements() {
  554. removeUnmarkedElements(this.doc.body);
  555. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  556. function removeUnmarkedElements(element) {
  557. let selectedElementFound = false;
  558. Array.from(element.childNodes).forEach(node => {
  559. if (node.nodeType == 1) {
  560. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  561. selectedElementFound = selectedElementFound || isSelectedElement;
  562. if (isSelectedElement) {
  563. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  564. removeUnmarkedElements(node);
  565. } else if (selectedElementFound) {
  566. removeNode(node);
  567. } else {
  568. hideNode(node);
  569. }
  570. }
  571. });
  572. }
  573. function removeNode(node) {
  574. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  575. node.remove();
  576. } else {
  577. hideNode(node);
  578. }
  579. }
  580. function hideNode(node) {
  581. if (canHideNode(node)) {
  582. node.style.setProperty("display", "none", "important");
  583. Array.from(node.childNodes).forEach(removeNode);
  584. }
  585. }
  586. function canHideNode(node) {
  587. if (node.nodeType == 1) {
  588. const tagName = node.tagName && node.tagName.toLowerCase();
  589. return (tagName != "svg" && tagName != "style" && tagName != "link");
  590. }
  591. }
  592. }
  593. insertVideoPosters() {
  594. if (this.options.posters) {
  595. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  596. let videoElement;
  597. if (element.tagName == "VIDEO") {
  598. videoElement = element;
  599. } else {
  600. videoElement = element.parentElement;
  601. }
  602. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  603. if (attributeValue) {
  604. const posterURL = this.options.posters[Number(attributeValue)];
  605. if (!videoElement.poster && posterURL) {
  606. videoElement.setAttribute("poster", posterURL);
  607. }
  608. }
  609. });
  610. }
  611. }
  612. removeFrames() {
  613. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  614. this.stats.set("discarded", "frames", frameElements.length);
  615. this.stats.set("processed", "frames", frameElements.length);
  616. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  617. }
  618. removeScripts() {
  619. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  620. this.doc.querySelectorAll("[href]").forEach(element => {
  621. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  622. element.setAttribute("href", "");
  623. }
  624. });
  625. this.doc.querySelectorAll("[src]").forEach(element => {
  626. if (element.src && element.src.match(/^\s*javascript:/)) {
  627. element.removeAttribute("src");
  628. }
  629. });
  630. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  631. this.stats.set("discarded", "scripts", scriptElements.length);
  632. this.stats.set("processed", "scripts", scriptElements.length);
  633. scriptElements.forEach(element => element.remove());
  634. }
  635. removeVideoSources() {
  636. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  637. this.stats.set("discarded", "video sources", videoSourceElements.length);
  638. this.stats.set("processed", "video sources", videoSourceElements.length);
  639. videoSourceElements.forEach(element => {
  640. if (element.tagName == "SOURCE") {
  641. element.remove();
  642. } else {
  643. videoSourceElements.forEach(element => element.removeAttribute("src"));
  644. }
  645. });
  646. }
  647. removeAudioSources() {
  648. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  649. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  650. this.stats.set("processed", "audio sources", audioSourceElements.length);
  651. audioSourceElements.forEach(element => {
  652. if (element.tagName == "SOURCE") {
  653. element.remove();
  654. } else {
  655. audioSourceElements.forEach(element => element.removeAttribute("src"));
  656. }
  657. });
  658. }
  659. removeDiscardedResources() {
  660. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  661. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  662. 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\"])");
  663. this.stats.set("discarded", "objects", objectElements.length);
  664. this.stats.set("processed", "objects", objectElements.length);
  665. objectElements.forEach(element => element.remove());
  666. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  667. replacedAttributeValue.forEach(element => {
  668. let regExp;
  669. if (this.options.removeScripts) {
  670. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  671. } else {
  672. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  673. }
  674. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  675. if (relValue.length) {
  676. element.setAttribute("rel", relValue);
  677. } else {
  678. element.remove();
  679. }
  680. });
  681. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  682. if (this.options.removeHiddenElements) {
  683. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  684. }
  685. if (!this.options.saveFavicon) {
  686. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  687. }
  688. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  689. }
  690. resetCharsetMeta() {
  691. let charset;
  692. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  693. const charsetDeclaration = element.content.split(";")[1];
  694. if (charsetDeclaration && !charset) {
  695. charset = charsetDeclaration.split("=")[1];
  696. if (charset) {
  697. this.charset = charset.trim().toLowerCase();
  698. }
  699. }
  700. element.remove();
  701. });
  702. const metaElement = this.doc.createElement("meta");
  703. metaElement.setAttribute("charset", UTF8_CHARSET);
  704. if (this.doc.head.firstChild) {
  705. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  706. } else {
  707. this.doc.head.appendChild(metaElement);
  708. }
  709. }
  710. setInputValues() {
  711. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  712. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  713. input.setAttribute("value", value || "");
  714. });
  715. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  716. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  717. if (value == "true") {
  718. input.setAttribute("checked", "");
  719. }
  720. });
  721. this.doc.querySelectorAll("textarea").forEach(textarea => {
  722. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  723. textarea.textContent = value || "";
  724. });
  725. this.doc.querySelectorAll("select").forEach(select => {
  726. select.querySelectorAll("option").forEach(option => {
  727. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  728. if (selected) {
  729. option.setAttribute("selected", "");
  730. }
  731. });
  732. });
  733. }
  734. saveFavicon() {
  735. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  736. if (!faviconElement) {
  737. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  738. }
  739. if (!faviconElement) {
  740. faviconElement = this.doc.createElement("link");
  741. faviconElement.setAttribute("type", "image/x-icon");
  742. faviconElement.setAttribute("rel", "shortcut icon");
  743. faviconElement.setAttribute("href", "/favicon.ico");
  744. }
  745. this.doc.head.appendChild(faviconElement);
  746. }
  747. replaceCanvasElements() {
  748. if (this.options.canvases) {
  749. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  750. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  751. if (attributeValue) {
  752. const canvasData = this.options.canvases[Number(attributeValue)];
  753. if (canvasData) {
  754. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  755. this.stats.add("processed", "canvas", 1);
  756. }
  757. }
  758. });
  759. }
  760. }
  761. insertFonts() {
  762. if (this.options.fonts && this.options.fonts.length) {
  763. let stylesheetContent = "";
  764. this.options.fonts.forEach(fontData => {
  765. if (fontData["font-family"] && fontData.src) {
  766. stylesheetContent += "@font-face{";
  767. let stylesContent = "";
  768. Object.keys(fontData).forEach(fontStyle => {
  769. if (stylesContent) {
  770. stylesContent += ";";
  771. }
  772. stylesContent += fontStyle + ":" + fontData[fontStyle];
  773. });
  774. stylesheetContent += stylesContent + "}";
  775. }
  776. });
  777. if (stylesheetContent) {
  778. const styleElement = this.doc.createElement("style");
  779. styleElement.textContent = stylesheetContent;
  780. const existingStyleElement = this.doc.querySelector("style");
  781. if (existingStyleElement) {
  782. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  783. } else {
  784. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  785. }
  786. }
  787. }
  788. }
  789. removeHiddenElements() {
  790. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  791. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  792. this.stats.set("discarded", "hidden elements", removedElements.length);
  793. this.stats.set("processed", "hidden elements", removedElements.length);
  794. if (hiddenElements.length) {
  795. const styleElement = this.doc.createElement("style");
  796. styleElement.textContent = ".sf-hidden{display:none!important;}";
  797. this.doc.head.appendChild(styleElement);
  798. hiddenElements.forEach(element => {
  799. if (element.style.getPropertyValue("display") != "none") {
  800. if (element.style.getPropertyPriority("display") == "important") {
  801. element.style.setProperty("display", "none", "important");
  802. } else {
  803. element.classList.add("sf-hidden");
  804. }
  805. }
  806. });
  807. }
  808. removedElements.forEach(element => element.remove());
  809. }
  810. resolveHrefs() {
  811. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  812. const href = element.getAttribute("href").trim();
  813. if (!testIgnoredPath(href)) {
  814. let resolvedURL;
  815. try {
  816. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  817. } catch (error) {
  818. // ignored
  819. }
  820. if (resolvedURL) {
  821. const url = normalizeURL(this.options.url);
  822. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  823. resolvedURL = resolvedURL.substring(url.length);
  824. }
  825. try {
  826. element.setAttribute("href", resolvedURL);
  827. } catch (error) {
  828. // ignored
  829. }
  830. }
  831. }
  832. });
  833. }
  834. resolveStyleAttributeURLs() {
  835. this.doc.querySelectorAll("[style]").forEach(element => {
  836. let styleContent = element.getAttribute("style");
  837. if (this.options.compressCSS) {
  838. styleContent = util.compressCSS(styleContent);
  839. }
  840. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement);
  841. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  842. this.styles.set(element, declarationList);
  843. });
  844. }
  845. async resolveStylesheetURLs() {
  846. const options = {
  847. maxResourceSize: this.options.maxResourceSize,
  848. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  849. url: this.options.url,
  850. charset: this.charset,
  851. compressCSS: this.options.compressCSS,
  852. updatedResources: this.options.updatedResources,
  853. rootDocument: this.options.rootDocument,
  854. frameId: this.options.windowId
  855. };
  856. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  857. .map(async element => {
  858. let mediaText;
  859. if (element.media) {
  860. mediaText = element.media.toLowerCase();
  861. }
  862. const stylesheetInfo = { mediaText };
  863. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  864. stylesheetInfo.scoped = true;
  865. }
  866. if (element.tagName.toLowerCase() == "link") {
  867. if (element.charset) {
  868. options.charset = element.charset;
  869. }
  870. }
  871. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  872. }));
  873. if (options.rootDocument) {
  874. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  875. await Promise.all(newResources.map(async resource => {
  876. resource.retrieved = true;
  877. const stylesheetInfo = {};
  878. const element = this.doc.createElement("style");
  879. this.doc.body.appendChild(element);
  880. element.textContent = resource.content;
  881. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  882. }));
  883. }
  884. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  885. stylesheets.set(element, stylesheetInfo);
  886. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  887. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  888. if (match && match[1] && match[1] != options.charset) {
  889. options.charset = match[1];
  890. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  891. }
  892. let stylesheet;
  893. try {
  894. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  895. } catch (error) {
  896. // ignored
  897. }
  898. if (stylesheet && stylesheet.children) {
  899. if (options.compressCSS) {
  900. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  901. }
  902. stylesheetInfo.stylesheet = stylesheet;
  903. } else {
  904. stylesheets.delete(element);
  905. }
  906. }
  907. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  908. let content;
  909. if (element.tagName.toLowerCase() == "link") {
  910. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  911. } else {
  912. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  913. }
  914. return content || "";
  915. }
  916. }
  917. async resolveFrameURLs() {
  918. if (!this.options.saveRawPage) {
  919. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  920. await Promise.all(frameElements.map(async frameElement => {
  921. if (frameElement.tagName == "OBJECT") {
  922. frameElement.setAttribute("data", "data:text/html,");
  923. } else {
  924. frameElement.removeAttribute("src");
  925. frameElement.removeAttribute("srcdoc");
  926. }
  927. Array.from(frameElement.childNodes).forEach(node => node.remove());
  928. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  929. if (this.options.frames && frameWindowId) {
  930. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  931. if (frameData) {
  932. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  933. }
  934. }
  935. }));
  936. }
  937. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  938. options.insertSingleFileComment = false;
  939. options.insertCanonicalLink = false;
  940. options.insertMetaNoIndex = false;
  941. options.saveFavicon = false;
  942. options.url = frameData.baseURI;
  943. options.windowId = frameWindowId;
  944. if (frameData.content) {
  945. options.content = frameData.content;
  946. options.canvases = frameData.canvases;
  947. options.fonts = frameData.fonts;
  948. options.stylesheets = frameData.stylesheets;
  949. options.images = frameData.images;
  950. options.posters = frameData.posters;
  951. options.usedFonts = frameData.usedFonts;
  952. options.shadowRoots = frameData.shadowRoots;
  953. options.imports = frameData.imports;
  954. frameData.runner = new Runner(options);
  955. frameData.frameElement = frameElement;
  956. await frameData.runner.loadPage();
  957. await frameData.runner.initialize();
  958. frameData.maxResources = batchRequest.getMaxResources();
  959. }
  960. }
  961. }
  962. insertShadowRootContents() {
  963. const doc = this.doc;
  964. const options = this.options;
  965. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  966. processElement(this.doc);
  967. if (this.options.removeScripts) {
  968. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  969. }
  970. const scriptElement = doc.createElement("script");
  971. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  972. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{let shadowRoot = element.parentElement.shadowRoot;if (!shadowRoot) {try {shadowRoot=element.parentElement.attachShadow({mode:element.getAttribute("${SHADOW_MODE_ATTRIBUTE_NAME}"),delegatesFocus:Boolean(element.getAttribute("${SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME}"))});shadowRoot.innerHTML=element.innerHTML;element.remove()} catch (error) {} if (shadowRoot) {processNode(shadowRoot)}}})}})()`;
  973. doc.body.appendChild(scriptElement);
  974. }
  975. function processElement(element) {
  976. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  977. shadowRootElements.forEach(element => {
  978. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  979. if (attributeValue) {
  980. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  981. if (shadowRootData) {
  982. const templateElement = doc.createElement("template");
  983. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  984. if (shadowRootData.delegatesFocus) {
  985. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  986. }
  987. if (shadowRootData.adoptedStyleSheets) {
  988. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  989. const styleElement = doc.createElement("style");
  990. styleElement.textContent = stylesheetContent;
  991. templateElement.appendChild(styleElement);
  992. });
  993. }
  994. const shadowDoc = util.parseDocContent(shadowRootData.content);
  995. if (shadowDoc.head) {
  996. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  997. if (metaCharset) {
  998. metaCharset.remove();
  999. }
  1000. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1001. }
  1002. if (shadowDoc.body) {
  1003. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1004. }
  1005. processElement(templateElement);
  1006. if (element.firstChild) {
  1007. element.insertBefore(templateElement, element.firstChild);
  1008. } else {
  1009. element.appendChild(templateElement);
  1010. }
  1011. }
  1012. }
  1013. });
  1014. }
  1015. }
  1016. async resolveHtmlImportURLs() {
  1017. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1018. await Promise.all(linkElements.map(async linkElement => {
  1019. const resourceURL = linkElement.href;
  1020. linkElement.removeAttribute("href");
  1021. const options = Object.create(this.options);
  1022. options.insertSingleFileComment = false;
  1023. options.insertCanonicalLink = false;
  1024. options.insertMetaNoIndex = false;
  1025. options.saveFavicon = false;
  1026. options.removeUnusedStyles = false;
  1027. options.removeAlternativeMedias = false;
  1028. options.removeUnusedFonts = false;
  1029. options.removeHiddenElements = false;
  1030. options.url = resourceURL;
  1031. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1032. if (attributeValue) {
  1033. const importData = options.imports[Number(attributeValue)];
  1034. if (importData) {
  1035. options.content = importData.content;
  1036. importData.runner = new Runner(options);
  1037. await importData.runner.loadPage();
  1038. await importData.runner.initialize();
  1039. if (!options.removeImports) {
  1040. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1041. }
  1042. importData.runner.getStyleSheets().forEach(stylesheet => {
  1043. const importedStyleElement = this.doc.createElement("style");
  1044. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1045. this.stylesheets.set(importedStyleElement, stylesheet);
  1046. });
  1047. }
  1048. }
  1049. if (options.removeImports) {
  1050. linkElement.remove();
  1051. this.stats.add("discarded", "HTML imports", 1);
  1052. }
  1053. }));
  1054. }
  1055. removeUnusedStyles() {
  1056. if (!this.mediaAllInfo) {
  1057. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1058. }
  1059. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1060. this.stats.set("processed", "CSS rules", stats.processed);
  1061. this.stats.set("discarded", "CSS rules", stats.discarded);
  1062. }
  1063. removeUnusedFonts() {
  1064. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1065. }
  1066. removeAlternativeMedias() {
  1067. const stats = util.minifyMedias(this.stylesheets);
  1068. this.stats.set("processed", "medias", stats.processed);
  1069. this.stats.set("discarded", "medias", stats.discarded);
  1070. }
  1071. async processStylesheets() {
  1072. this.options.fontURLs = new Map();
  1073. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1074. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1075. ));
  1076. }
  1077. async processStyleAttributes() {
  1078. return Promise.all([...this.styles].map(([, declarationList]) =>
  1079. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1080. ));
  1081. }
  1082. async processPageResources() {
  1083. const processAttributeArgs = [
  1084. ["link[href][rel*=\"icon\"]", "href", false, true],
  1085. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1086. ["img[src], input[src][type=image]", "src", true],
  1087. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1088. ["video[poster]", "poster"],
  1089. ["*[background]", "background"],
  1090. ["image", "xlink:href"],
  1091. ["image", "href"]
  1092. ];
  1093. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1094. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1095. );
  1096. resourcePromises = resourcePromises.concat([
  1097. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1098. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1099. ]);
  1100. if (!this.options.removeAudioSrc) {
  1101. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1102. }
  1103. if (!this.options.removeVideoSrc) {
  1104. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1105. }
  1106. await Promise.all(resourcePromises);
  1107. if (this.options.saveFavicon) {
  1108. ProcessorHelper.processShortcutIcons(this.doc);
  1109. }
  1110. }
  1111. async processScripts() {
  1112. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1113. let resourceURL;
  1114. let scriptSrc;
  1115. if (element.tagName == "SCRIPT") {
  1116. scriptSrc = element.getAttribute("src");
  1117. element.removeAttribute("src");
  1118. } else {
  1119. scriptSrc = element.getAttribute("href");
  1120. element.removeAttribute("href");
  1121. }
  1122. element.removeAttribute("integrity");
  1123. element.textContent = "";
  1124. try {
  1125. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1126. } catch (error) {
  1127. // ignored
  1128. }
  1129. if (testValidURL(resourceURL)) {
  1130. this.stats.add("processed", "scripts", 1);
  1131. const content = await util.getContent(resourceURL, {
  1132. asBinary: true,
  1133. charset: this.charset != UTF8_CHARSET && this.charset,
  1134. maxResourceSize: this.options.maxResourceSize,
  1135. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1136. frameId: this.options.windowId
  1137. });
  1138. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1139. if (element.tagName == "SCRIPT") {
  1140. element.setAttribute("src", content.data);
  1141. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1142. element.setAttribute("async", "");
  1143. }
  1144. } else {
  1145. const scriptElement = this.doc.createElement("script");
  1146. scriptElement.setAttribute("src", content.data);
  1147. scriptElement.setAttribute("async", "");
  1148. element.parentElement.replaceChild(scriptElement, element);
  1149. }
  1150. }
  1151. }));
  1152. }
  1153. removeAlternativeImages() {
  1154. util.removeAlternativeImages(this.doc);
  1155. }
  1156. async removeAlternativeFonts() {
  1157. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1158. }
  1159. async processFrames() {
  1160. if (this.options.frames) {
  1161. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1162. await Promise.all(frameElements.map(async frameElement => {
  1163. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1164. if (frameWindowId) {
  1165. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1166. if (frameData) {
  1167. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1168. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1169. this.stats.add("processed", "frames", 1);
  1170. await frameData.runner.run();
  1171. const pageData = await frameData.runner.getPageData();
  1172. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1173. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1174. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1175. sandbox += " allow-scripts allow-same-origin";
  1176. }
  1177. frameElement.setAttribute("sandbox", sandbox);
  1178. if (frameElement.tagName == "OBJECT") {
  1179. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1180. } else {
  1181. if (frameElement.tagName == "FRAME") {
  1182. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1183. } else {
  1184. frameElement.setAttribute("srcdoc", pageData.content);
  1185. frameElement.removeAttribute("src");
  1186. }
  1187. }
  1188. this.stats.addAll(pageData);
  1189. } else {
  1190. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1191. this.stats.add("discarded", "frames", 1);
  1192. }
  1193. }
  1194. }
  1195. }));
  1196. }
  1197. }
  1198. async processHtmlImports() {
  1199. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1200. await Promise.all(linkElements.map(async linkElement => {
  1201. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1202. if (attributeValue) {
  1203. const importData = this.options.imports[Number(attributeValue)];
  1204. if (importData.runner) {
  1205. this.stats.add("processed", "HTML imports", 1);
  1206. await importData.runner.run();
  1207. const pageData = await importData.runner.getPageData();
  1208. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1209. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1210. this.stats.addAll(pageData);
  1211. } else {
  1212. this.stats.add("discarded", "HTML imports", 1);
  1213. }
  1214. }
  1215. }));
  1216. }
  1217. replaceStylesheets() {
  1218. this.doc.querySelectorAll("style").forEach(styleElement => {
  1219. const stylesheetInfo = this.stylesheets.get(styleElement);
  1220. if (stylesheetInfo) {
  1221. this.stylesheets.delete(styleElement);
  1222. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1223. styleElement.textContent = stylesheetContent;
  1224. if (stylesheetInfo.mediaText) {
  1225. styleElement.media = stylesheetInfo.mediaText;
  1226. }
  1227. } else {
  1228. styleElement.remove();
  1229. }
  1230. });
  1231. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1232. const stylesheetInfo = this.stylesheets.get(linkElement);
  1233. if (stylesheetInfo) {
  1234. this.stylesheets.delete(linkElement);
  1235. const styleElement = this.doc.createElement("style");
  1236. if (stylesheetInfo.mediaText) {
  1237. styleElement.media = stylesheetInfo.mediaText;
  1238. }
  1239. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1240. styleElement.textContent = stylesheetContent;
  1241. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1242. } else {
  1243. linkElement.remove();
  1244. }
  1245. });
  1246. }
  1247. replaceStyleAttributes() {
  1248. this.doc.querySelectorAll("[style]").forEach(element => {
  1249. const declarations = this.styles.get(element);
  1250. if (declarations) {
  1251. this.styles.delete(element);
  1252. let styleContent = cssTree.generate(declarations);
  1253. element.setAttribute("style", styleContent);
  1254. } else {
  1255. element.setAttribute("style", "");
  1256. }
  1257. });
  1258. }
  1259. insertVariables() {
  1260. if (this.cssVariables.size) {
  1261. const styleElement = this.doc.createElement("style");
  1262. const firstStyleElement = this.doc.head.querySelector("style");
  1263. if (firstStyleElement) {
  1264. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1265. } else {
  1266. this.doc.head.appendChild(styleElement);
  1267. }
  1268. let stylesheetContent = "";
  1269. this.cssVariables.forEach((content, indexResource) => {
  1270. this.cssVariables.delete(indexResource);
  1271. if (stylesheetContent) {
  1272. stylesheetContent += ";";
  1273. }
  1274. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1275. });
  1276. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1277. }
  1278. }
  1279. compressHTML() {
  1280. let size;
  1281. if (this.options.displayStats) {
  1282. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1283. }
  1284. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1285. if (this.options.displayStats) {
  1286. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1287. }
  1288. }
  1289. cleanupPage() {
  1290. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1291. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1292. if (metaCharset) {
  1293. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1294. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1295. this.doc.head.querySelector("meta[charset]").remove();
  1296. }
  1297. }
  1298. }
  1299. resetZoomLevel() {
  1300. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1301. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1302. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1303. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1304. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1305. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1306. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1307. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1308. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1309. this.doc.documentElement.style.removeProperty("-sf-transform");
  1310. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1311. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1312. }
  1313. async insertMAFFMetaData() {
  1314. const maffMetaData = await this.maffMetaDataPromise;
  1315. if (maffMetaData && maffMetaData.content) {
  1316. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1317. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1318. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1319. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1320. if (originalURLElement) {
  1321. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1322. }
  1323. if (archiveTimeElement) {
  1324. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1325. if (value) {
  1326. const date = new Date(value);
  1327. if (!isNaN(date.getTime())) {
  1328. this.options.saveDate = new Date(value);
  1329. }
  1330. }
  1331. }
  1332. }
  1333. }
  1334. async setDocInfo() {
  1335. const titleElement = this.doc.querySelector("title");
  1336. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1337. const authorElement = this.doc.querySelector("meta[name=author]");
  1338. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1339. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1340. const headingElement = this.doc.querySelector("h1");
  1341. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1342. this.options.info = {
  1343. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1344. lang: this.doc.documentElement.lang,
  1345. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1346. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1347. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1348. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1349. };
  1350. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1351. }
  1352. }
  1353. // ---------------
  1354. // ProcessorHelper
  1355. // ---------------
  1356. const DATA_URI_PREFIX = "data:";
  1357. const ABOUT_BLANK_URI = "about:blank";
  1358. const EMPTY_DATA_URI = "data:null;base64,";
  1359. const REGEXP_URL_HASH = /(#.+?)$/;
  1360. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1361. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1362. class ProcessorHelper {
  1363. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1364. const date = options.saveDate;
  1365. const url = util.parseURL(options.saveUrl);
  1366. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1367. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1368. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1370. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1371. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1376. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1377. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1387. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1388. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1397. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1398. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1401. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1402. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1403. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1404. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1405. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1406. for (const [name, value] of params) {
  1407. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1408. }
  1409. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1410. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1411. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1412. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1413. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1414. if (content) {
  1415. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1416. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1417. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1418. }
  1419. return template.trim();
  1420. function decode(value) {
  1421. try {
  1422. return decodeURI(value);
  1423. } catch (error) {
  1424. return value;
  1425. }
  1426. }
  1427. }
  1428. static setBackgroundImage(element, url, style) {
  1429. element.style.setProperty("background-blend-mode", "normal", "important");
  1430. element.style.setProperty("background-clip", "content-box", "important");
  1431. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1432. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1433. element.style.setProperty("background-image", url, "important");
  1434. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1435. element.style.setProperty("background-origin", "content-box", "important");
  1436. element.style.setProperty("background-repeat", "no-repeat", "important");
  1437. }
  1438. static processShortcutIcons(doc) {
  1439. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1440. if (!shortcutIcon) {
  1441. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1442. if (shortcutIcon) {
  1443. shortcutIcon.rel = "icon";
  1444. }
  1445. }
  1446. if (shortcutIcon) {
  1447. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1448. if (linkElement != shortcutIcon) {
  1449. linkElement.remove();
  1450. }
  1451. });
  1452. }
  1453. }
  1454. static removeSingleLineCssComments(stylesheet) {
  1455. const removedRules = [];
  1456. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1457. const ruleData = cssRule.data;
  1458. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1459. removedRules.push(cssRule);
  1460. }
  1461. }
  1462. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1463. }
  1464. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1465. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet);
  1466. const imports = getImportFunctions(stylesheetContent);
  1467. await Promise.all(imports.map(async cssImport => {
  1468. const match = matchImport(cssImport);
  1469. if (match) {
  1470. const regExpCssImport = getRegExp(cssImport);
  1471. let resourceURL = normalizeURL(match.resourceURL);
  1472. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1473. try {
  1474. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1475. } catch (error) {
  1476. // ignored
  1477. }
  1478. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1479. const content = await util.getContent(resourceURL, {
  1480. maxResourceSize: options.maxResourceSize,
  1481. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1482. validateTextContentType: true,
  1483. frameId: options.frameId
  1484. });
  1485. resourceURL = content.resourceURL;
  1486. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1487. let importedStylesheetContent = removeCssComments(content.data);
  1488. if (options.compressCSS) {
  1489. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1490. }
  1491. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1492. if (stylesheetContent.includes(cssImport)) {
  1493. const ancestorStyleSheets = new Set(importedStyleSheets);
  1494. ancestorStyleSheets.add(resourceURL);
  1495. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1496. workStylesheet.textContent = importedStylesheetContent;
  1497. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1498. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1499. } else {
  1500. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1501. }
  1502. }
  1503. } else {
  1504. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1505. }
  1506. } else {
  1507. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1508. }
  1509. }
  1510. }));
  1511. return stylesheetContent;
  1512. }
  1513. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet) {
  1514. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1515. urlFunctions.map(urlFunction => {
  1516. const originalResourceURL = matchURL(urlFunction);
  1517. let resourceURL = normalizeURL(originalResourceURL);
  1518. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1519. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1520. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1521. }
  1522. if (!testIgnoredPath(resourceURL)) {
  1523. if (!resourceURL || testValidPath(resourceURL)) {
  1524. let resolvedURL;
  1525. if (!originalResourceURL.startsWith("#")) {
  1526. try {
  1527. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1528. } catch (error) {
  1529. // ignored
  1530. }
  1531. }
  1532. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1533. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1534. }
  1535. } else {
  1536. let newUrlFunction;
  1537. if (originalResourceURL) {
  1538. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1539. } else {
  1540. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1541. }
  1542. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1543. }
  1544. }
  1545. });
  1546. return stylesheetContent;
  1547. }
  1548. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1549. resourceURL = normalizeURL(resourceURL);
  1550. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1551. const content = await util.getContent(resourceURL, {
  1552. maxResourceSize: options.maxResourceSize,
  1553. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1554. charset: options.charset,
  1555. frameId: options.frameId,
  1556. validateTextContentType: true
  1557. });
  1558. resourceURL = content.resourceURL;
  1559. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1560. let stylesheetContent = removeCssComments(content.data);
  1561. if (options.compressCSS) {
  1562. stylesheetContent = util.compressCSS(stylesheetContent);
  1563. }
  1564. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1565. return stylesheetContent;
  1566. }
  1567. }
  1568. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1569. const promises = [];
  1570. const removedRules = [];
  1571. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1572. const ruleData = cssRule.data;
  1573. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1574. removedRules.push(cssRule);
  1575. } else if (ruleData.block && ruleData.block.children) {
  1576. if (ruleData.type == "Rule") {
  1577. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1578. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1579. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1580. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1581. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1582. }
  1583. }
  1584. }
  1585. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1586. await Promise.all(promises);
  1587. async function processFontFaceRule(ruleData, fontURLs) {
  1588. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1589. if (declaration.type == "Declaration" && declaration.value.children) {
  1590. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1591. await Promise.all(urlFunctions.map(async urlFunction => {
  1592. const originalResourceURL = matchURL(urlFunction);
  1593. const resourceURL = normalizeURL(originalResourceURL);
  1594. if (!testIgnoredPath(resourceURL)) {
  1595. if (testValidURL(resourceURL)) {
  1596. let { content } = await batchRequest.addURL(resourceURL, true, "font");
  1597. let resourceURLs = fontURLs.get(declaration);
  1598. if (!resourceURLs) {
  1599. resourceURLs = [];
  1600. fontURLs.set(declaration, resourceURLs);
  1601. }
  1602. resourceURLs.push(resourceURL);
  1603. replaceURLs(declaration, originalResourceURL, content);
  1604. }
  1605. }
  1606. }));
  1607. }
  1608. }));
  1609. function replaceURLs(declaration, oldURL, newURL) {
  1610. declaration.value.children.forEach(token => {
  1611. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1612. token.value.value = newURL;
  1613. }
  1614. });
  1615. }
  1616. }
  1617. }
  1618. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1619. await Promise.all(declarations.map(async declaration => {
  1620. let children = declaration.value.children;
  1621. if (!children && declaration.value && declaration.value.type == "Raw") {
  1622. try {
  1623. children = cssTree.parse(declaration.value.value, { context: "value" }).children;
  1624. } catch (error) {
  1625. // ignored
  1626. }
  1627. }
  1628. if (declaration.type == "Declaration" && children) {
  1629. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1630. await Promise.all(urlFunctions.map(async urlFunction => {
  1631. const originalResourceURL = matchURL(urlFunction);
  1632. const resourceURL = normalizeURL(originalResourceURL);
  1633. if (!testIgnoredPath(resourceURL)) {
  1634. if (testValidURL(resourceURL)) {
  1635. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", true);
  1636. let variableDefined;
  1637. const tokens = [];
  1638. findURLToken(originalResourceURL, children, (token, parent, rootFunction) => {
  1639. if (!originalResourceURL.startsWith("#")) {
  1640. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1641. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1642. tokens.push({ parent, token, value });
  1643. variableDefined = true;
  1644. } else {
  1645. token.data.value.value = content;
  1646. }
  1647. }
  1648. });
  1649. if (variableDefined) {
  1650. cssVariables.set(indexResource, content);
  1651. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1652. }
  1653. }
  1654. }
  1655. }));
  1656. }
  1657. }));
  1658. function findURLToken(url, children, callback, depth = 0) {
  1659. for (let token = children.head; token; token = token.next) {
  1660. if (token.data.children) {
  1661. findURLToken(url, token.data.children, callback, depth + 1);
  1662. }
  1663. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1664. callback(token, children, depth == 0);
  1665. }
  1666. }
  1667. }
  1668. }
  1669. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1670. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1671. let resourceURL = resourceElement.getAttribute(attributeName);
  1672. if (resourceURL != null) {
  1673. resourceURL = normalizeURL(resourceURL);
  1674. let originURL = resourceElement.dataset.singleFileOriginURL;
  1675. delete resourceElement.dataset.singleFileOriginURL;
  1676. if (!testIgnoredPath(resourceURL)) {
  1677. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1678. if (testValidPath(resourceURL)) {
  1679. try {
  1680. resourceURL = util.resolveURL(resourceURL, baseURI);
  1681. } catch (error) {
  1682. // ignored
  1683. }
  1684. if (testValidURL(resourceURL)) {
  1685. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", resourceElement.tagName == "IMG" && attributeName == "src");
  1686. if (originURL) {
  1687. if (content == EMPTY_DATA_URI) {
  1688. try {
  1689. originURL = util.resolveURL(originURL, baseURI);
  1690. } catch (error) {
  1691. // ignored
  1692. }
  1693. try {
  1694. resourceURL = originURL;
  1695. content = (await util.getContent(resourceURL, {
  1696. asBinary: true,
  1697. expectedType: "image",
  1698. maxResourceSize: options.maxResourceSize,
  1699. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1700. frameId: options.windowId
  1701. })).data;
  1702. } catch (error) {
  1703. // ignored
  1704. }
  1705. }
  1706. }
  1707. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1708. resourceElement.remove();
  1709. } else {
  1710. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1711. if (!forbiddenPrefixFound) {
  1712. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1713. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1714. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1715. cssVariables.set(indexResource, content);
  1716. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1717. styles.set(resourceElement, declarationList);
  1718. } else {
  1719. resourceElement.setAttribute(attributeName, content);
  1720. }
  1721. } else {
  1722. resourceElement.setAttribute(attributeName, content);
  1723. }
  1724. }
  1725. }
  1726. }
  1727. }
  1728. }
  1729. }
  1730. }));
  1731. }
  1732. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1733. let attributeName = "xlink:href";
  1734. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1735. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1736. if (originalResourceURL == null) {
  1737. attributeName = "href";
  1738. originalResourceURL = resourceElement.getAttribute(attributeName);
  1739. }
  1740. let resourceURL = normalizeURL(originalResourceURL);
  1741. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1742. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1743. try {
  1744. resourceURL = util.resolveURL(resourceURL, baseURI);
  1745. } catch (error) {
  1746. // ignored
  1747. }
  1748. if (testValidURL(resourceURL)) {
  1749. const { content } = await batchRequest.addURL(resourceURL);
  1750. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1751. if (hashMatch && hashMatch[0]) {
  1752. let symbolElement;
  1753. try {
  1754. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1755. } catch (error) {
  1756. // ignored
  1757. }
  1758. if (symbolElement) {
  1759. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1760. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1761. }
  1762. } else {
  1763. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1764. }
  1765. }
  1766. } else if (resourceURL == options.url) {
  1767. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1768. }
  1769. }));
  1770. }
  1771. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1772. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1773. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1774. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1775. let resourceURL = normalizeURL(srcsetValue.url);
  1776. if (!testIgnoredPath(resourceURL)) {
  1777. if (testValidPath(resourceURL)) {
  1778. try {
  1779. resourceURL = util.resolveURL(resourceURL, baseURI);
  1780. } catch (error) {
  1781. // ignored
  1782. }
  1783. if (testValidURL(resourceURL)) {
  1784. const { content } = await batchRequest.addURL(resourceURL, true, "image");
  1785. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1786. if (forbiddenPrefixFound) {
  1787. return "";
  1788. }
  1789. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1790. } else {
  1791. return "";
  1792. }
  1793. } else {
  1794. return "";
  1795. }
  1796. } else {
  1797. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1798. }
  1799. }));
  1800. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1801. }));
  1802. }
  1803. static replaceImageSource(imgElement, variableName, options) {
  1804. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1805. if (attributeValue) {
  1806. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1807. if (imageData && imageData.replaceable) {
  1808. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1809. const backgroundStyle = {};
  1810. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1811. if (backgroundSize) {
  1812. backgroundStyle["background-size"] = imageData.objectFit;
  1813. }
  1814. if (imageData.objectPosition) {
  1815. backgroundStyle["background-position"] = imageData.objectPosition;
  1816. }
  1817. if (imageData.backgroundColor) {
  1818. backgroundStyle["background-color"] = imageData.backgroundColor;
  1819. }
  1820. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1821. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1822. return true;
  1823. }
  1824. }
  1825. }
  1826. }
  1827. // ----
  1828. // Util
  1829. // ----
  1830. const BLOB_URI_PREFIX = "blob:";
  1831. const HTTP_URI_PREFIX = /^https?:\/\//;
  1832. const FILE_URI_PREFIX = /^file:\/\//;
  1833. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1834. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1835. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1836. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1837. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1838. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1839. 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;
  1840. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1841. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1842. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1843. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1844. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1845. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1846. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1847. function getUpdatedResourceContent(resourceURL, content, options) {
  1848. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1849. options.updatedResources[resourceURL].retrieved = true;
  1850. return options.updatedResources[resourceURL].content;
  1851. } else {
  1852. return content.data || "";
  1853. }
  1854. }
  1855. function normalizeURL(url) {
  1856. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1857. return url;
  1858. } else {
  1859. return url.split("#")[0];
  1860. }
  1861. }
  1862. function getCSSValue(value) {
  1863. let result = "";
  1864. try {
  1865. result = cssTree.generate(value);
  1866. } catch (error) {
  1867. // ignored
  1868. }
  1869. return result;
  1870. }
  1871. function getOnEventAttributeNames(doc) {
  1872. const element = doc.createElement("div");
  1873. const attributeNames = [];
  1874. for (const propertyName in element) {
  1875. if (propertyName.startsWith("on")) {
  1876. attributeNames.push(propertyName);
  1877. }
  1878. }
  1879. attributeNames.push("onunload");
  1880. return attributeNames;
  1881. }
  1882. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1883. let maxLength;
  1884. if (template) {
  1885. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1886. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1887. if (template.match(replaceRegExp)) {
  1888. const matchedLength = template.match(replaceRegExp)[0];
  1889. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1890. if (isNaN(maxLength) || maxLength <= 0) {
  1891. maxLength = null;
  1892. }
  1893. } else {
  1894. replaceRegExp = new RegExp(regExpVariable, "g");
  1895. }
  1896. if (template.match(replaceRegExp)) {
  1897. let value = await valueGetter();
  1898. if (!dontReplaceSlash) {
  1899. value = value.replace(/\/+/g, replacementCharacter);
  1900. }
  1901. if (maxLength) {
  1902. value = await util.truncateText(value, maxLength);
  1903. }
  1904. return template.replace(replaceRegExp, value);
  1905. }
  1906. }
  1907. return template;
  1908. }
  1909. function getLastSegment(url, replacementCharacter) {
  1910. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1911. if (!lastSegment) {
  1912. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1913. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1914. }
  1915. if (!lastSegment) {
  1916. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1917. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1918. }
  1919. if (!lastSegment) {
  1920. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1921. }
  1922. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1923. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1924. lastSegment = lastSegmentMatch[1];
  1925. }
  1926. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1927. return lastSegment;
  1928. }
  1929. function getRegExp(string) {
  1930. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1931. }
  1932. function getUrlFunctions(stylesheetContent, unique) {
  1933. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1934. if (unique) {
  1935. return [...new Set(result)];
  1936. } else {
  1937. return result;
  1938. }
  1939. }
  1940. function getImportFunctions(stylesheetContent) {
  1941. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1942. }
  1943. function findShortcutIcon(shortcutIcons) {
  1944. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1945. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1946. return shortcutIcons[0];
  1947. }
  1948. function matchURL(stylesheetContent) {
  1949. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1950. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1951. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1952. return match && match[1];
  1953. }
  1954. function testIgnoredPath(resourceURL) {
  1955. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1956. }
  1957. function testValidPath(resourceURL) {
  1958. return resourceURL && !resourceURL.match(EMPTY_URL);
  1959. }
  1960. function testValidURL(resourceURL) {
  1961. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1962. }
  1963. function matchImport(stylesheetContent) {
  1964. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1965. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1966. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1967. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1968. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1969. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1970. if (match) {
  1971. const [, resourceURL, media] = match;
  1972. return { resourceURL, media };
  1973. }
  1974. }
  1975. function removeCssComments(stylesheetContent) {
  1976. try {
  1977. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  1978. } catch (error) {
  1979. let start, end;
  1980. do {
  1981. start = stylesheetContent.indexOf("/*");
  1982. end = stylesheetContent.indexOf("*/", start + 2);
  1983. if (start != -1 && end != -1) {
  1984. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1985. }
  1986. } while (start != -1 && end != -1);
  1987. return stylesheetContent;
  1988. }
  1989. }
  1990. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  1991. if (mediaQuery) {
  1992. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1993. } else {
  1994. return stylesheetContent;
  1995. }
  1996. }
  1997. function log(...args) {
  1998. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1999. }
  2000. // -----
  2001. // Stats
  2002. // -----
  2003. const STATS_DEFAULT_VALUES = {
  2004. discarded: {
  2005. "HTML bytes": 0,
  2006. "hidden elements": 0,
  2007. "HTML imports": 0,
  2008. scripts: 0,
  2009. objects: 0,
  2010. "audio sources": 0,
  2011. "video sources": 0,
  2012. frames: 0,
  2013. "CSS rules": 0,
  2014. canvas: 0,
  2015. stylesheets: 0,
  2016. resources: 0,
  2017. medias: 0
  2018. },
  2019. processed: {
  2020. "HTML bytes": 0,
  2021. "hidden elements": 0,
  2022. "HTML imports": 0,
  2023. scripts: 0,
  2024. objects: 0,
  2025. "audio sources": 0,
  2026. "video sources": 0,
  2027. frames: 0,
  2028. "CSS rules": 0,
  2029. canvas: 0,
  2030. stylesheets: 0,
  2031. resources: 0,
  2032. medias: 0
  2033. }
  2034. };
  2035. class Stats {
  2036. constructor(options) {
  2037. this.options = options;
  2038. if (options.displayStats) {
  2039. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2040. }
  2041. }
  2042. set(type, subType, value) {
  2043. if (this.options.displayStats) {
  2044. this.data[type][subType] = value;
  2045. }
  2046. }
  2047. add(type, subType, value) {
  2048. if (this.options.displayStats) {
  2049. this.data[type][subType] += value;
  2050. }
  2051. }
  2052. addAll(pageData) {
  2053. if (this.options.displayStats) {
  2054. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2055. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2056. }
  2057. }
  2058. }
  2059. return { getClass };
  2060. })();