single-file-core.js 92 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. }
  34. async run() {
  35. if (this.options.userScriptEnabled) {
  36. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  37. }
  38. this.runner = new Runner(this.options, true);
  39. await this.runner.loadPage();
  40. await this.runner.initialize();
  41. if (this.options.userScriptEnabled) {
  42. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  43. }
  44. await this.runner.run();
  45. }
  46. cancel() {
  47. this.cancelled = true;
  48. if (this.runner) {
  49. this.runner.cancel();
  50. }
  51. }
  52. getPageData() {
  53. return this.runner.getPageData();
  54. }
  55. }
  56. // -------------
  57. // ProgressEvent
  58. // -------------
  59. const PAGE_LOADING = "page-loading";
  60. const PAGE_LOADED = "page-loaded";
  61. const RESOURCES_INITIALIZING = "resource-initializing";
  62. const RESOURCES_INITIALIZED = "resources-initialized";
  63. const RESOURCE_LOADED = "resource-loaded";
  64. const PAGE_ENDED = "page-ended";
  65. const STAGE_STARTED = "stage-started";
  66. const STAGE_ENDED = "stage-ended";
  67. const STAGE_TASK_STARTED = "stage-task-started";
  68. const STAGE_TASK_ENDED = "stage-task-ended";
  69. class ProgressEvent {
  70. constructor(type, detail) {
  71. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  72. }
  73. }
  74. // ------
  75. // Runner
  76. // ------
  77. const RESOLVE_URLS_STAGE = 0;
  78. const REPLACE_DATA_STAGE = 1;
  79. const REPLACE_DOCS_STAGE = 2;
  80. const POST_PROCESS_STAGE = 3;
  81. const STAGES = [{
  82. sequential: [
  83. { action: "preProcessPage" },
  84. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  85. { action: "replaceStyleContents" },
  86. { action: "resetCharsetMeta" },
  87. { option: "saveFavicon", action: "saveFavicon" },
  88. { action: "replaceCanvasElements" },
  89. { action: "insertFonts" },
  90. { action: "insertShadowRootContents" },
  91. { action: "setInputValues" },
  92. { option: "removeScripts", action: "removeScripts" },
  93. { option: "selected", action: "removeUnselectedElements" },
  94. { option: "removeVideoSrc", action: "insertVideoPosters" },
  95. { option: "removeFrames", action: "removeFrames" },
  96. { option: "removeVideoSrc", action: "removeVideoSources" },
  97. { option: "removeAudioSrc", action: "removeAudioSources" },
  98. { action: "removeDiscardedResources" },
  99. { option: "removeHiddenElements", action: "removeHiddenElements" },
  100. { action: "resolveHrefs" },
  101. { action: "resolveStyleAttributeURLs" }
  102. ],
  103. parallel: [
  104. { action: "resolveStylesheetURLs" },
  105. { option: "!removeFrames", action: "resolveFrameURLs" },
  106. { action: "resolveHtmlImportURLs" }
  107. ]
  108. }, {
  109. sequential: [
  110. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  111. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  112. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  113. ],
  114. parallel: [
  115. { action: "processStylesheets" },
  116. { action: "processStyleAttributes" },
  117. { action: "processPageResources" },
  118. { option: "!removeScripts", action: "processScripts" }
  119. ]
  120. }, {
  121. sequential: [
  122. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  123. ],
  124. parallel: [
  125. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  126. { option: "!removeFrames", action: "processFrames" },
  127. { option: "!removeImports", action: "processHtmlImports" },
  128. ]
  129. }, {
  130. sequential: [
  131. { action: "replaceStylesheets" },
  132. { action: "replaceStyleAttributes" },
  133. { action: "insertVariables" },
  134. { option: "compressHTML", action: "compressHTML" },
  135. { action: "cleanupPage" }
  136. ],
  137. parallel: [
  138. { option: "enableMaff", action: "insertMAFFMetaData" },
  139. { action: "setDocInfo" }
  140. ]
  141. }];
  142. class Runner {
  143. constructor(options, root) {
  144. const rootDocDefined = root && options.doc;
  145. this.root = root;
  146. this.options = options;
  147. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  148. const matchResourceReferrer = this.options.url.match(/^.*\//);
  149. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  150. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  151. this.options.rootDocument = root;
  152. this.options.updatedResources = this.options.updatedResources || {};
  153. this.options.fontTests = new Map();
  154. this.batchRequest = new BatchRequest();
  155. this.processor = new Processor(options, this.batchRequest);
  156. if (rootDocDefined) {
  157. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  158. this.options.canvases = docData.canvases;
  159. this.options.fonts = docData.fonts;
  160. this.options.stylesheets = docData.stylesheets;
  161. this.options.images = docData.images;
  162. this.options.posters = docData.posters;
  163. this.options.usedFonts = docData.usedFonts;
  164. this.options.shadowRoots = docData.shadowRoots;
  165. this.options.imports = docData.imports;
  166. this.options.referrer = docData.referrer;
  167. this.markedElements = docData.markedElements;
  168. }
  169. if (this.options.saveRawPage) {
  170. this.options.removeFrames = true;
  171. }
  172. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  173. this.onprogress = options.onprogress || (() => { });
  174. }
  175. async loadPage() {
  176. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  177. await this.processor.loadPage(this.options.content);
  178. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  179. }
  180. async initialize() {
  181. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  182. await this.executeStage(RESOLVE_URLS_STAGE);
  183. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  184. if (this.root && this.options.doc) {
  185. util.postProcessDoc(this.options.doc, this.markedElements);
  186. }
  187. }
  188. cancel() {
  189. this.cancelled = true;
  190. this.batchRequest.cancel();
  191. if (this.root) {
  192. if (this.options.frames) {
  193. this.options.frames.forEach(cancelRunner);
  194. }
  195. if (this.options.imports) {
  196. this.options.imports.forEach(cancelRunner);
  197. }
  198. }
  199. function cancelRunner(resourceData) {
  200. if (resourceData.runner) {
  201. resourceData.runner.cancel();
  202. }
  203. }
  204. }
  205. async run() {
  206. if (this.root) {
  207. this.processor.initialize(this.batchRequest);
  208. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  209. }
  210. await this.batchRequest.run(detail => {
  211. detail.pageURL = this.options.url;
  212. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  213. }, this.options);
  214. await this.pendingPromises;
  215. this.options.doc = null;
  216. this.options.win = null;
  217. await this.executeStage(REPLACE_DOCS_STAGE);
  218. await this.executeStage(POST_PROCESS_STAGE);
  219. this.processor.finalize();
  220. }
  221. getDocument() {
  222. return this.processor.doc;
  223. }
  224. getStyleSheets() {
  225. return this.processor.stylesheets;
  226. }
  227. getPageData() {
  228. if (this.root) {
  229. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  230. }
  231. return this.processor.getPageData();
  232. }
  233. async executeStage(step) {
  234. if (DEBUG) {
  235. log("**** STARTED STAGE", step, "****");
  236. }
  237. const frame = !this.root;
  238. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  239. STAGES[step].sequential.forEach(task => {
  240. let startTime;
  241. if (DEBUG) {
  242. startTime = Date.now();
  243. log(" -- STARTED task =", task.action);
  244. }
  245. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  246. if (!this.cancelled) {
  247. this.executeTask(task);
  248. }
  249. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  250. if (DEBUG) {
  251. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  252. }
  253. });
  254. let parallelTasksPromise;
  255. if (STAGES[step].parallel) {
  256. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  257. let startTime;
  258. if (DEBUG) {
  259. startTime = Date.now();
  260. log(" // STARTED task =", task.action);
  261. }
  262. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  263. if (!this.cancelled) {
  264. await this.executeTask(task);
  265. }
  266. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  267. if (DEBUG) {
  268. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  269. }
  270. }));
  271. } else {
  272. parallelTasksPromise = Promise.resolve();
  273. }
  274. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  275. if (DEBUG) {
  276. log("**** ENDED STAGE", step, "****");
  277. }
  278. return parallelTasksPromise;
  279. }
  280. executeTask(task) {
  281. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  282. return this.processor[task.action]();
  283. }
  284. }
  285. }
  286. // ------------
  287. // BatchRequest
  288. // ------------
  289. class BatchRequest {
  290. constructor() {
  291. this.requests = new Map();
  292. this.duplicates = new Map();
  293. }
  294. addURL(resourceURL, asBinary, expectedType, groupDuplicates) {
  295. return new Promise((resolve, reject) => {
  296. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType]);
  297. let resourceRequests = this.requests.get(requestKey);
  298. if (!resourceRequests) {
  299. resourceRequests = [];
  300. this.requests.set(requestKey, resourceRequests);
  301. }
  302. const callbacks = { resolve, reject };
  303. resourceRequests.push(callbacks);
  304. if (groupDuplicates) {
  305. let duplicateRequests = this.duplicates.get(requestKey);
  306. if (!duplicateRequests) {
  307. duplicateRequests = [];
  308. this.duplicates.set(requestKey, duplicateRequests);
  309. }
  310. duplicateRequests.push(callbacks);
  311. }
  312. });
  313. }
  314. getMaxResources() {
  315. return this.requests.size;
  316. }
  317. run(onloadListener, options) {
  318. const resourceURLs = [...this.requests.keys()];
  319. let indexResource = 0;
  320. return Promise.all(resourceURLs.map(async requestKey => {
  321. const [resourceURL, asBinary, expectedType] = JSON.parse(requestKey);
  322. const resourceRequests = this.requests.get(requestKey);
  323. try {
  324. const currentIndexResource = indexResource;
  325. indexResource = indexResource + 1;
  326. const content = await util.getContent(resourceURL, {
  327. asBinary,
  328. expectedType,
  329. maxResourceSize: options.maxResourceSize,
  330. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  331. frameId: options.windowId,
  332. resourceReferrer: options.resourceReferrer
  333. });
  334. onloadListener({ url: resourceURL });
  335. if (!this.cancelled) {
  336. resourceRequests.forEach(callbacks => {
  337. const duplicateCallbacks = this.duplicates.get(requestKey);
  338. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  339. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  340. });
  341. }
  342. } catch (error) {
  343. indexResource = indexResource + 1;
  344. onloadListener({ url: resourceURL });
  345. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  346. }
  347. this.requests.delete(requestKey);
  348. }));
  349. }
  350. cancel() {
  351. this.cancelled = true;
  352. const resourceURLs = [...this.requests.keys()];
  353. resourceURLs.forEach(requestKey => {
  354. const resourceRequests = this.requests.get(requestKey);
  355. resourceRequests.forEach(callbacks => callbacks.reject());
  356. this.requests.delete(requestKey);
  357. });
  358. }
  359. }
  360. // ---------
  361. // Processor
  362. // ---------
  363. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  364. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  365. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  366. const SCRIPT_TAG_FOUND = /<script/gi;
  367. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  368. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  369. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  370. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  371. const UTF8_CHARSET = "utf-8";
  372. class Processor {
  373. constructor(options, batchRequest) {
  374. this.options = options;
  375. this.stats = new Stats(options);
  376. this.baseURI = normalizeURL(options.baseURI || options.url);
  377. this.batchRequest = batchRequest;
  378. this.stylesheets = new Map();
  379. this.styles = new Map();
  380. this.cssVariables = new Map();
  381. this.fontTests = options.fontTests;
  382. }
  383. initialize() {
  384. this.options.saveDate = new Date();
  385. this.options.saveUrl = this.options.url;
  386. if (this.options.enableMaff) {
  387. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url));
  388. }
  389. this.maxResources = this.batchRequest.getMaxResources();
  390. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  391. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  392. }
  393. if (!this.options.removeImports && this.options.imports) {
  394. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  395. }
  396. this.stats.set("processed", "resources", this.maxResources);
  397. }
  398. async loadPage(pageContent, charset) {
  399. let content;
  400. if (!pageContent || this.options.saveRawPage) {
  401. content = await util.getContent(this.baseURI, {
  402. maxResourceSize: this.options.maxResourceSize,
  403. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  404. charset,
  405. frameId: this.options.windowId,
  406. resourceReferrer: this.options.resourceReferrer
  407. });
  408. pageContent = content.data;
  409. }
  410. this.doc = util.parseDocContent(pageContent, this.baseURI);
  411. if (this.options.saveRawPage) {
  412. let charset;
  413. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  414. const charsetDeclaration = element.content.split(";")[1];
  415. if (charsetDeclaration && !charset) {
  416. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  417. }
  418. });
  419. if (charset && charset != content.charset) {
  420. return this.loadPage(pageContent, charset);
  421. }
  422. }
  423. this.workStyleElement = this.doc.createElement("style");
  424. this.doc.body.appendChild(this.workStyleElement);
  425. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  426. }
  427. finalize() {
  428. if (this.workStyleElement.parentNode) {
  429. this.workStyleElement.remove();
  430. }
  431. }
  432. async getPageData() {
  433. util.postProcessDoc(this.doc);
  434. const url = util.parseURL(this.baseURI);
  435. if (this.options.insertSingleFileComment) {
  436. const firstComment = this.doc.documentElement.firstChild;
  437. let infobarURL = this.options.saveUrl, infobarSaveDate = this.options.saveDate;
  438. if (firstComment.nodeType == 8 && (firstComment.textContent.includes(util.COMMENT_HEADER_LEGACY) || firstComment.textContent.includes(util.COMMENT_HEADER))) {
  439. const info = this.doc.documentElement.firstChild.textContent.split("\n");
  440. const [, , url, saveDate] = info;
  441. infobarURL = url.split("url: ")[1];
  442. infobarSaveDate = saveDate.split("saved date: ")[1];
  443. firstComment.remove();
  444. }
  445. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  446. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  447. " \n url: " + infobarURL +
  448. " \n saved date: " + infobarSaveDate +
  449. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  450. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  451. }
  452. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  453. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  454. if (!canonicalLink) {
  455. canonicalLink = this.doc.createElement("link");
  456. canonicalLink.setAttribute("rel", "canonical");
  457. this.doc.head.appendChild(canonicalLink);
  458. }
  459. if (canonicalLink && !canonicalLink.href) {
  460. canonicalLink.href = this.options.saveUrl;
  461. }
  462. }
  463. if (this.options.insertMetaNoIndex) {
  464. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  465. if (!metaElement) {
  466. metaElement = this.doc.createElement("meta");
  467. metaElement.setAttribute("name", "robots");
  468. metaElement.setAttribute("content", "noindex");
  469. this.doc.head.appendChild(metaElement);
  470. }
  471. }
  472. let size;
  473. if (this.options.displayStats) {
  474. size = util.getContentSize(this.doc.documentElement.outerHTML);
  475. }
  476. const content = util.serialize(this.doc, this.options.compressHTML);
  477. if (this.options.displayStats) {
  478. const contentSize = util.getContentSize(content);
  479. this.stats.set("processed", "HTML bytes", contentSize);
  480. this.stats.add("discarded", "HTML bytes", size - contentSize);
  481. }
  482. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  483. const replacementCharacter = this.options.filenameReplacementCharacter;
  484. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  485. if (!this.options.backgroundSave) {
  486. filename = filename.replace(/\//g, replacementCharacter);
  487. }
  488. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  489. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  490. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  491. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  492. filename = filename + "…" + extension;
  493. }
  494. if (!filename) {
  495. filename = "Unnamed page";
  496. }
  497. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  498. const pageData = {
  499. stats: this.stats.data,
  500. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  501. filename,
  502. content
  503. };
  504. if (this.options.addProof) {
  505. pageData.hash = await util.digest("SHA-256", content);
  506. }
  507. if (this.options.retrieveLinks) {
  508. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  509. }
  510. return pageData;
  511. }
  512. preProcessPage() {
  513. if (this.options.win) {
  514. 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));
  515. }
  516. if (this.options.images && !this.options.saveRawPage) {
  517. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  518. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  519. if (attributeValue) {
  520. const imageData = this.options.images[Number(attributeValue)];
  521. if (imageData) {
  522. if (this.options.removeHiddenElements && (
  523. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  524. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  525. )) {
  526. imgElement.setAttribute("src", EMPTY_IMAGE);
  527. } else {
  528. if (imageData.currentSrc) {
  529. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  530. imgElement.setAttribute("src", imageData.currentSrc);
  531. }
  532. if (this.options.loadDeferredImages) {
  533. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  534. imageData.src = imgElement.dataset.src;
  535. imgElement.setAttribute("src", imgElement.dataset.src);
  536. imgElement.removeAttribute("data-src");
  537. }
  538. }
  539. }
  540. }
  541. }
  542. });
  543. if (this.options.loadDeferredImages) {
  544. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  545. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  546. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  547. imgElement.removeAttribute("data-srcset");
  548. }
  549. });
  550. }
  551. }
  552. }
  553. replaceStyleContents() {
  554. if (this.options.stylesheets) {
  555. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  556. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  557. if (attributeValue) {
  558. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  559. if (stylesheetContent) {
  560. styleElement.textContent = stylesheetContent;
  561. }
  562. }
  563. });
  564. }
  565. }
  566. removeUnselectedElements() {
  567. removeUnmarkedElements(this.doc.body);
  568. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  569. function removeUnmarkedElements(element) {
  570. let selectedElementFound = false;
  571. Array.from(element.childNodes).forEach(node => {
  572. if (node.nodeType == 1) {
  573. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  574. selectedElementFound = selectedElementFound || isSelectedElement;
  575. if (isSelectedElement) {
  576. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  577. removeUnmarkedElements(node);
  578. } else if (selectedElementFound) {
  579. removeNode(node);
  580. } else {
  581. hideNode(node);
  582. }
  583. }
  584. });
  585. }
  586. function removeNode(node) {
  587. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  588. node.remove();
  589. } else {
  590. hideNode(node);
  591. }
  592. }
  593. function hideNode(node) {
  594. if (canHideNode(node)) {
  595. node.style.setProperty("display", "none", "important");
  596. Array.from(node.childNodes).forEach(removeNode);
  597. }
  598. }
  599. function canHideNode(node) {
  600. if (node.nodeType == 1) {
  601. const tagName = node.tagName && node.tagName.toLowerCase();
  602. return (tagName != "svg" && tagName != "style" && tagName != "link");
  603. }
  604. }
  605. }
  606. insertVideoPosters() {
  607. if (this.options.posters) {
  608. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  609. let videoElement;
  610. if (element.tagName == "VIDEO") {
  611. videoElement = element;
  612. } else {
  613. videoElement = element.parentElement;
  614. }
  615. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  616. if (attributeValue) {
  617. const posterURL = this.options.posters[Number(attributeValue)];
  618. if (!videoElement.poster && posterURL) {
  619. videoElement.setAttribute("poster", posterURL);
  620. }
  621. }
  622. });
  623. }
  624. }
  625. removeFrames() {
  626. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  627. this.stats.set("discarded", "frames", frameElements.length);
  628. this.stats.set("processed", "frames", frameElements.length);
  629. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  630. }
  631. removeScripts() {
  632. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  633. this.doc.querySelectorAll("[href]").forEach(element => {
  634. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  635. element.setAttribute("href", "");
  636. }
  637. });
  638. this.doc.querySelectorAll("[src]").forEach(element => {
  639. if (element.src && element.src.match(/^\s*javascript:/)) {
  640. element.removeAttribute("src");
  641. }
  642. });
  643. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  644. this.stats.set("discarded", "scripts", scriptElements.length);
  645. this.stats.set("processed", "scripts", scriptElements.length);
  646. scriptElements.forEach(element => element.remove());
  647. }
  648. removeVideoSources() {
  649. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  650. this.stats.set("discarded", "video sources", videoSourceElements.length);
  651. this.stats.set("processed", "video sources", videoSourceElements.length);
  652. videoSourceElements.forEach(element => {
  653. if (element.tagName == "SOURCE") {
  654. element.remove();
  655. } else {
  656. videoSourceElements.forEach(element => element.removeAttribute("src"));
  657. }
  658. });
  659. }
  660. removeAudioSources() {
  661. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  662. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  663. this.stats.set("processed", "audio sources", audioSourceElements.length);
  664. audioSourceElements.forEach(element => {
  665. if (element.tagName == "SOURCE") {
  666. element.remove();
  667. } else {
  668. audioSourceElements.forEach(element => element.removeAttribute("src"));
  669. }
  670. });
  671. }
  672. removeDiscardedResources() {
  673. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  674. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  675. 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\"])");
  676. this.stats.set("discarded", "objects", objectElements.length);
  677. this.stats.set("processed", "objects", objectElements.length);
  678. objectElements.forEach(element => element.remove());
  679. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  680. replacedAttributeValue.forEach(element => {
  681. let regExp;
  682. if (this.options.removeScripts) {
  683. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  684. } else {
  685. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  686. }
  687. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  688. if (relValue.length) {
  689. element.setAttribute("rel", relValue);
  690. } else {
  691. element.remove();
  692. }
  693. });
  694. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  695. if (this.options.removeHiddenElements) {
  696. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  697. }
  698. if (!this.options.saveFavicon) {
  699. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  700. }
  701. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  702. }
  703. resetCharsetMeta() {
  704. let charset;
  705. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  706. const charsetDeclaration = element.content.split(";")[1];
  707. if (charsetDeclaration && !charset) {
  708. charset = charsetDeclaration.split("=")[1];
  709. if (charset) {
  710. this.charset = charset.trim().toLowerCase();
  711. }
  712. }
  713. element.remove();
  714. });
  715. const metaElement = this.doc.createElement("meta");
  716. metaElement.setAttribute("charset", UTF8_CHARSET);
  717. if (this.doc.head.firstChild) {
  718. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  719. } else {
  720. this.doc.head.appendChild(metaElement);
  721. }
  722. }
  723. setInputValues() {
  724. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  725. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  726. input.setAttribute("value", value || "");
  727. });
  728. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  729. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  730. if (value == "true") {
  731. input.setAttribute("checked", "");
  732. }
  733. });
  734. this.doc.querySelectorAll("textarea").forEach(textarea => {
  735. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  736. textarea.textContent = value || "";
  737. });
  738. this.doc.querySelectorAll("select").forEach(select => {
  739. select.querySelectorAll("option").forEach(option => {
  740. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  741. if (selected) {
  742. option.setAttribute("selected", "");
  743. }
  744. });
  745. });
  746. }
  747. saveFavicon() {
  748. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  749. if (!faviconElement) {
  750. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  751. }
  752. if (!faviconElement) {
  753. faviconElement = this.doc.createElement("link");
  754. faviconElement.setAttribute("type", "image/x-icon");
  755. faviconElement.setAttribute("rel", "shortcut icon");
  756. faviconElement.setAttribute("href", "/favicon.ico");
  757. }
  758. this.doc.head.appendChild(faviconElement);
  759. }
  760. replaceCanvasElements() {
  761. if (this.options.canvases) {
  762. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  763. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  764. if (attributeValue) {
  765. const canvasData = this.options.canvases[Number(attributeValue)];
  766. if (canvasData) {
  767. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  768. this.stats.add("processed", "canvas", 1);
  769. }
  770. }
  771. });
  772. }
  773. }
  774. insertFonts() {
  775. if (this.options.fonts && this.options.fonts.length) {
  776. let stylesheetContent = "";
  777. this.options.fonts.forEach(fontData => {
  778. if (fontData["font-family"] && fontData.src) {
  779. stylesheetContent += "@font-face{";
  780. let stylesContent = "";
  781. Object.keys(fontData).forEach(fontStyle => {
  782. if (stylesContent) {
  783. stylesContent += ";";
  784. }
  785. stylesContent += fontStyle + ":" + fontData[fontStyle];
  786. });
  787. stylesheetContent += stylesContent + "}";
  788. }
  789. });
  790. if (stylesheetContent) {
  791. const styleElement = this.doc.createElement("style");
  792. styleElement.textContent = stylesheetContent;
  793. const existingStyleElement = this.doc.querySelector("style");
  794. if (existingStyleElement) {
  795. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  796. } else {
  797. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  798. }
  799. }
  800. }
  801. }
  802. removeHiddenElements() {
  803. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  804. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  805. this.stats.set("discarded", "hidden elements", removedElements.length);
  806. this.stats.set("processed", "hidden elements", removedElements.length);
  807. if (hiddenElements.length) {
  808. const styleElement = this.doc.createElement("style");
  809. styleElement.textContent = ".sf-hidden{display:none!important;}";
  810. this.doc.head.appendChild(styleElement);
  811. hiddenElements.forEach(element => {
  812. if (element.style.getPropertyValue("display") != "none") {
  813. if (element.style.getPropertyPriority("display") == "important") {
  814. element.style.setProperty("display", "none", "important");
  815. } else {
  816. element.classList.add("sf-hidden");
  817. }
  818. }
  819. });
  820. }
  821. removedElements.forEach(element => element.remove());
  822. }
  823. resolveHrefs() {
  824. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  825. const href = element.getAttribute("href").trim();
  826. if (!testIgnoredPath(href)) {
  827. let resolvedURL;
  828. try {
  829. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  830. } catch (error) {
  831. // ignored
  832. }
  833. if (resolvedURL) {
  834. const url = normalizeURL(this.options.url);
  835. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  836. resolvedURL = resolvedURL.substring(url.length);
  837. }
  838. try {
  839. element.setAttribute("href", resolvedURL);
  840. } catch (error) {
  841. // ignored
  842. }
  843. }
  844. }
  845. });
  846. }
  847. resolveStyleAttributeURLs() {
  848. this.doc.querySelectorAll("[style]").forEach(element => {
  849. let styleContent = element.getAttribute("style");
  850. if (this.options.compressCSS) {
  851. styleContent = util.compressCSS(styleContent);
  852. }
  853. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement);
  854. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  855. this.styles.set(element, declarationList);
  856. });
  857. }
  858. async resolveStylesheetURLs() {
  859. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]")).map(async element => {
  860. const options = {
  861. maxResourceSize: this.options.maxResourceSize,
  862. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  863. url: this.options.url,
  864. charset: this.charset,
  865. compressCSS: this.options.compressCSS,
  866. updatedResources: this.options.updatedResources,
  867. rootDocument: this.options.rootDocument,
  868. frameId: this.options.windowId,
  869. resourceReferrer: this.options.resourceReferrer
  870. };
  871. let mediaText;
  872. if (element.media) {
  873. mediaText = element.media.toLowerCase();
  874. }
  875. const stylesheetInfo = { mediaText };
  876. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  877. stylesheetInfo.scoped = true;
  878. }
  879. if (element.tagName == "LINK" && element.charset) {
  880. options.charset = element.charset;
  881. }
  882. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  883. }));
  884. if (this.options.rootDocument) {
  885. const newResources = Object.keys(this.options.updatedResources).filter(url => this.options.updatedResources[url].type == "stylesheet" && !this.options.updatedResources[url].retrieved).map(url => this.options.updatedResources[url]);
  886. await Promise.all(newResources.map(async resource => {
  887. resource.retrieved = true;
  888. const stylesheetInfo = {};
  889. const element = this.doc.createElement("style");
  890. this.doc.body.appendChild(element);
  891. element.textContent = resource.content;
  892. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, this.options, this.workStyleElement);
  893. }));
  894. }
  895. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  896. stylesheets.set(element, stylesheetInfo);
  897. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  898. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  899. options = Object.assign({}, options, { charset: getCharset(stylesheetContent) });
  900. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  901. }
  902. let stylesheet;
  903. try {
  904. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  905. } catch (error) {
  906. // ignored
  907. }
  908. if (stylesheet && stylesheet.children) {
  909. if (options.compressCSS) {
  910. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  911. }
  912. stylesheetInfo.stylesheet = stylesheet;
  913. } else {
  914. stylesheets.delete(element);
  915. }
  916. }
  917. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  918. let content;
  919. if (element.tagName == "LINK") {
  920. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  921. } else {
  922. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  923. }
  924. return content || "";
  925. }
  926. }
  927. async resolveFrameURLs() {
  928. if (!this.options.saveRawPage) {
  929. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  930. await Promise.all(frameElements.map(async frameElement => {
  931. if (frameElement.tagName == "OBJECT") {
  932. frameElement.setAttribute("data", "data:text/html,");
  933. } else {
  934. frameElement.removeAttribute("src");
  935. frameElement.removeAttribute("srcdoc");
  936. }
  937. Array.from(frameElement.childNodes).forEach(node => node.remove());
  938. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  939. if (this.options.frames && frameWindowId) {
  940. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  941. if (frameData) {
  942. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  943. }
  944. }
  945. }));
  946. }
  947. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  948. options.insertSingleFileComment = false;
  949. options.insertCanonicalLink = false;
  950. options.insertMetaNoIndex = false;
  951. options.saveFavicon = false;
  952. options.url = frameData.baseURI;
  953. options.windowId = frameWindowId;
  954. if (frameData.content) {
  955. options.content = frameData.content;
  956. options.canvases = frameData.canvases;
  957. options.fonts = frameData.fonts;
  958. options.stylesheets = frameData.stylesheets;
  959. options.images = frameData.images;
  960. options.posters = frameData.posters;
  961. options.usedFonts = frameData.usedFonts;
  962. options.shadowRoots = frameData.shadowRoots;
  963. options.imports = frameData.imports;
  964. frameData.runner = new Runner(options);
  965. frameData.frameElement = frameElement;
  966. await frameData.runner.loadPage();
  967. await frameData.runner.initialize();
  968. frameData.maxResources = batchRequest.getMaxResources();
  969. }
  970. }
  971. }
  972. insertShadowRootContents() {
  973. const doc = this.doc;
  974. const options = this.options;
  975. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  976. processElement(this.doc);
  977. if (this.options.removeScripts) {
  978. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  979. }
  980. const scriptElement = doc.createElement("script");
  981. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  982. 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)}}})}})()`;
  983. doc.body.appendChild(scriptElement);
  984. }
  985. function processElement(element) {
  986. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  987. shadowRootElements.forEach(element => {
  988. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  989. if (attributeValue) {
  990. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  991. if (shadowRootData) {
  992. const templateElement = doc.createElement("template");
  993. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  994. if (shadowRootData.delegatesFocus) {
  995. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  996. }
  997. if (shadowRootData.adoptedStyleSheets) {
  998. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  999. const styleElement = doc.createElement("style");
  1000. styleElement.textContent = stylesheetContent;
  1001. templateElement.appendChild(styleElement);
  1002. });
  1003. }
  1004. const shadowDoc = util.parseDocContent(shadowRootData.content);
  1005. if (shadowDoc.head) {
  1006. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  1007. if (metaCharset) {
  1008. metaCharset.remove();
  1009. }
  1010. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1011. }
  1012. if (shadowDoc.body) {
  1013. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1014. }
  1015. processElement(templateElement);
  1016. if (element.firstChild) {
  1017. element.insertBefore(templateElement, element.firstChild);
  1018. } else {
  1019. element.appendChild(templateElement);
  1020. }
  1021. }
  1022. }
  1023. });
  1024. }
  1025. }
  1026. async resolveHtmlImportURLs() {
  1027. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1028. await Promise.all(linkElements.map(async linkElement => {
  1029. const resourceURL = linkElement.href;
  1030. linkElement.removeAttribute("href");
  1031. const options = Object.create(this.options);
  1032. options.insertSingleFileComment = false;
  1033. options.insertCanonicalLink = false;
  1034. options.insertMetaNoIndex = false;
  1035. options.saveFavicon = false;
  1036. options.removeUnusedStyles = false;
  1037. options.removeAlternativeMedias = false;
  1038. options.removeUnusedFonts = false;
  1039. options.removeHiddenElements = false;
  1040. options.url = resourceURL;
  1041. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1042. if (attributeValue) {
  1043. const importData = options.imports[Number(attributeValue)];
  1044. if (importData) {
  1045. options.content = importData.content;
  1046. importData.runner = new Runner(options);
  1047. await importData.runner.loadPage();
  1048. await importData.runner.initialize();
  1049. if (!options.removeImports) {
  1050. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1051. }
  1052. importData.runner.getStyleSheets().forEach(stylesheet => {
  1053. const importedStyleElement = this.doc.createElement("style");
  1054. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1055. this.stylesheets.set(importedStyleElement, stylesheet);
  1056. });
  1057. }
  1058. }
  1059. if (options.removeImports) {
  1060. linkElement.remove();
  1061. this.stats.add("discarded", "HTML imports", 1);
  1062. }
  1063. }));
  1064. }
  1065. removeUnusedStyles() {
  1066. if (!this.mediaAllInfo) {
  1067. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1068. }
  1069. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1070. this.stats.set("processed", "CSS rules", stats.processed);
  1071. this.stats.set("discarded", "CSS rules", stats.discarded);
  1072. }
  1073. removeUnusedFonts() {
  1074. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1075. }
  1076. removeAlternativeMedias() {
  1077. const stats = util.minifyMedias(this.stylesheets);
  1078. this.stats.set("processed", "medias", stats.processed);
  1079. this.stats.set("discarded", "medias", stats.discarded);
  1080. }
  1081. async processStylesheets() {
  1082. this.options.fontURLs = new Map();
  1083. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1084. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1085. ));
  1086. }
  1087. async processStyleAttributes() {
  1088. return Promise.all([...this.styles].map(([, declarationList]) =>
  1089. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1090. ));
  1091. }
  1092. async processPageResources() {
  1093. const processAttributeArgs = [
  1094. ["link[href][rel*=\"icon\"]", "href", false, true],
  1095. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1096. ["img[src], input[src][type=image]", "src", true],
  1097. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1098. ["video[poster]", "poster"],
  1099. ["*[background]", "background"],
  1100. ["image", "xlink:href"],
  1101. ["image", "href"]
  1102. ];
  1103. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1104. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1105. );
  1106. resourcePromises = resourcePromises.concat([
  1107. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1108. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1109. ]);
  1110. if (!this.options.removeAudioSrc) {
  1111. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1112. }
  1113. if (!this.options.removeVideoSrc) {
  1114. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1115. }
  1116. await Promise.all(resourcePromises);
  1117. if (this.options.saveFavicon) {
  1118. ProcessorHelper.processShortcutIcons(this.doc);
  1119. }
  1120. }
  1121. async processScripts() {
  1122. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1123. let resourceURL;
  1124. let scriptSrc;
  1125. if (element.tagName == "SCRIPT") {
  1126. scriptSrc = element.getAttribute("src");
  1127. element.removeAttribute("src");
  1128. } else {
  1129. scriptSrc = element.getAttribute("href");
  1130. element.removeAttribute("href");
  1131. }
  1132. element.removeAttribute("integrity");
  1133. element.textContent = "";
  1134. try {
  1135. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1136. } catch (error) {
  1137. // ignored
  1138. }
  1139. if (testValidURL(resourceURL)) {
  1140. this.stats.add("processed", "scripts", 1);
  1141. const content = await util.getContent(resourceURL, {
  1142. asBinary: true,
  1143. charset: this.charset != UTF8_CHARSET && this.charset,
  1144. maxResourceSize: this.options.maxResourceSize,
  1145. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1146. frameId: this.options.windowId,
  1147. resourceReferrer: this.options.resourceReferrer
  1148. });
  1149. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1150. if (element.tagName == "SCRIPT") {
  1151. element.setAttribute("src", content.data);
  1152. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1153. element.setAttribute("async", "");
  1154. }
  1155. } else {
  1156. const scriptElement = this.doc.createElement("script");
  1157. scriptElement.setAttribute("src", content.data);
  1158. scriptElement.setAttribute("async", "");
  1159. element.parentElement.replaceChild(scriptElement, element);
  1160. }
  1161. }
  1162. }));
  1163. }
  1164. removeAlternativeImages() {
  1165. util.removeAlternativeImages(this.doc);
  1166. }
  1167. async removeAlternativeFonts() {
  1168. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1169. }
  1170. async processFrames() {
  1171. if (this.options.frames) {
  1172. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1173. await Promise.all(frameElements.map(async frameElement => {
  1174. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1175. if (frameWindowId) {
  1176. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1177. if (frameData) {
  1178. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1179. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1180. this.stats.add("processed", "frames", 1);
  1181. await frameData.runner.run();
  1182. const pageData = await frameData.runner.getPageData();
  1183. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1184. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1185. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1186. sandbox += " allow-scripts allow-same-origin";
  1187. }
  1188. frameElement.setAttribute("sandbox", sandbox);
  1189. if (frameElement.tagName == "OBJECT") {
  1190. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1191. } else {
  1192. if (frameElement.tagName == "FRAME") {
  1193. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1194. } else {
  1195. frameElement.setAttribute("srcdoc", pageData.content);
  1196. frameElement.removeAttribute("src");
  1197. }
  1198. }
  1199. this.stats.addAll(pageData);
  1200. } else {
  1201. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1202. this.stats.add("discarded", "frames", 1);
  1203. }
  1204. }
  1205. }
  1206. }));
  1207. }
  1208. }
  1209. async processHtmlImports() {
  1210. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1211. await Promise.all(linkElements.map(async linkElement => {
  1212. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1213. if (attributeValue) {
  1214. const importData = this.options.imports[Number(attributeValue)];
  1215. if (importData.runner) {
  1216. this.stats.add("processed", "HTML imports", 1);
  1217. await importData.runner.run();
  1218. const pageData = await importData.runner.getPageData();
  1219. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1220. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1221. this.stats.addAll(pageData);
  1222. } else {
  1223. this.stats.add("discarded", "HTML imports", 1);
  1224. }
  1225. }
  1226. }));
  1227. }
  1228. replaceStylesheets() {
  1229. this.doc.querySelectorAll("style").forEach(styleElement => {
  1230. const stylesheetInfo = this.stylesheets.get(styleElement);
  1231. if (stylesheetInfo) {
  1232. this.stylesheets.delete(styleElement);
  1233. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1234. styleElement.textContent = stylesheetContent;
  1235. if (stylesheetInfo.mediaText) {
  1236. styleElement.media = stylesheetInfo.mediaText;
  1237. }
  1238. } else {
  1239. styleElement.remove();
  1240. }
  1241. });
  1242. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1243. const stylesheetInfo = this.stylesheets.get(linkElement);
  1244. if (stylesheetInfo) {
  1245. this.stylesheets.delete(linkElement);
  1246. const styleElement = this.doc.createElement("style");
  1247. if (stylesheetInfo.mediaText) {
  1248. styleElement.media = stylesheetInfo.mediaText;
  1249. }
  1250. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1251. styleElement.textContent = stylesheetContent;
  1252. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1253. } else {
  1254. linkElement.remove();
  1255. }
  1256. });
  1257. }
  1258. replaceStyleAttributes() {
  1259. this.doc.querySelectorAll("[style]").forEach(element => {
  1260. const declarations = this.styles.get(element);
  1261. if (declarations) {
  1262. this.styles.delete(element);
  1263. let styleContent = cssTree.generate(declarations);
  1264. element.setAttribute("style", styleContent);
  1265. } else {
  1266. element.setAttribute("style", "");
  1267. }
  1268. });
  1269. }
  1270. insertVariables() {
  1271. if (this.cssVariables.size) {
  1272. const styleElement = this.doc.createElement("style");
  1273. const firstStyleElement = this.doc.head.querySelector("style");
  1274. if (firstStyleElement) {
  1275. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1276. } else {
  1277. this.doc.head.appendChild(styleElement);
  1278. }
  1279. let stylesheetContent = "";
  1280. this.cssVariables.forEach((content, indexResource) => {
  1281. this.cssVariables.delete(indexResource);
  1282. if (stylesheetContent) {
  1283. stylesheetContent += ";";
  1284. }
  1285. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1286. });
  1287. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1288. }
  1289. }
  1290. compressHTML() {
  1291. let size;
  1292. if (this.options.displayStats) {
  1293. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1294. }
  1295. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1296. if (this.options.displayStats) {
  1297. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1298. }
  1299. }
  1300. cleanupPage() {
  1301. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1302. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1303. if (metaCharset) {
  1304. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1305. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1306. this.doc.head.querySelector("meta[charset]").remove();
  1307. }
  1308. }
  1309. }
  1310. resetZoomLevel() {
  1311. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1312. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1313. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1314. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1315. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1316. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1317. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1318. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1319. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1320. this.doc.documentElement.style.removeProperty("-sf-transform");
  1321. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1322. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1323. }
  1324. async insertMAFFMetaData() {
  1325. const maffMetaData = await this.maffMetaDataPromise;
  1326. if (maffMetaData && maffMetaData.content) {
  1327. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1328. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1329. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1330. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1331. if (originalURLElement) {
  1332. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1333. }
  1334. if (archiveTimeElement) {
  1335. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1336. if (value) {
  1337. const date = new Date(value);
  1338. if (!isNaN(date.getTime())) {
  1339. this.options.saveDate = new Date(value);
  1340. }
  1341. }
  1342. }
  1343. }
  1344. }
  1345. async setDocInfo() {
  1346. const titleElement = this.doc.querySelector("title");
  1347. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1348. const authorElement = this.doc.querySelector("meta[name=author]");
  1349. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1350. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1351. const headingElement = this.doc.querySelector("h1");
  1352. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1353. this.options.info = {
  1354. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1355. lang: this.doc.documentElement.lang,
  1356. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1357. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1358. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1359. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1360. };
  1361. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1362. }
  1363. }
  1364. // ---------------
  1365. // ProcessorHelper
  1366. // ---------------
  1367. const DATA_URI_PREFIX = "data:";
  1368. const ABOUT_BLANK_URI = "about:blank";
  1369. const EMPTY_DATA_URI = "data:null;base64,";
  1370. const REGEXP_URL_HASH = /(#.+?)$/;
  1371. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1372. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1373. class ProcessorHelper {
  1374. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1375. const date = options.saveDate;
  1376. const url = util.parseURL(options.saveUrl);
  1377. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1387. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1388. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1389. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1390. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1391. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1392. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1393. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1397. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1398. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1401. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1402. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1403. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1404. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1405. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1406. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1407. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1408. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1409. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1410. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1411. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1412. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1413. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1414. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1415. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1416. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1417. for (const [name, value] of params) {
  1418. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1419. }
  1420. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1421. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1422. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1423. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1424. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1425. if (content) {
  1426. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1427. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1428. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1429. }
  1430. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1431. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1432. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1433. template = await evalTemplateVariable(template, "profile-name", () => options.profileName, dontReplaceSlash, options.filenameReplacementCharacter);
  1434. return template.trim();
  1435. function decode(value) {
  1436. try {
  1437. return decodeURI(value);
  1438. } catch (error) {
  1439. return value;
  1440. }
  1441. }
  1442. }
  1443. static setBackgroundImage(element, url, style) {
  1444. element.style.setProperty("background-blend-mode", "normal", "important");
  1445. element.style.setProperty("background-clip", "content-box", "important");
  1446. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1447. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1448. element.style.setProperty("background-image", url, "important");
  1449. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1450. element.style.setProperty("background-origin", "content-box", "important");
  1451. element.style.setProperty("background-repeat", "no-repeat", "important");
  1452. }
  1453. static processShortcutIcons(doc) {
  1454. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1455. if (!shortcutIcon) {
  1456. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1457. if (shortcutIcon) {
  1458. shortcutIcon.rel = "icon";
  1459. }
  1460. }
  1461. if (shortcutIcon) {
  1462. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1463. if (linkElement != shortcutIcon) {
  1464. linkElement.remove();
  1465. }
  1466. });
  1467. }
  1468. }
  1469. static removeSingleLineCssComments(stylesheet) {
  1470. const removedRules = [];
  1471. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1472. const ruleData = cssRule.data;
  1473. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1474. removedRules.push(cssRule);
  1475. }
  1476. }
  1477. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1478. }
  1479. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1480. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet);
  1481. const imports = getImportFunctions(stylesheetContent);
  1482. await Promise.all(imports.map(async cssImport => {
  1483. const match = matchImport(cssImport);
  1484. if (match) {
  1485. const regExpCssImport = getRegExp(cssImport);
  1486. let resourceURL = normalizeURL(match.resourceURL);
  1487. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1488. try {
  1489. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1490. } catch (error) {
  1491. // ignored
  1492. }
  1493. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1494. const content = await getStylesheetContent(resourceURL);
  1495. resourceURL = content.resourceURL;
  1496. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1497. let importedStylesheetContent = removeCssComments(content.data);
  1498. if (options.compressCSS) {
  1499. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1500. }
  1501. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1502. if (stylesheetContent.includes(cssImport)) {
  1503. const ancestorStyleSheets = new Set(importedStyleSheets);
  1504. ancestorStyleSheets.add(resourceURL);
  1505. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1506. workStylesheet.textContent = importedStylesheetContent;
  1507. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1508. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1509. } else {
  1510. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1511. }
  1512. }
  1513. } else {
  1514. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1515. }
  1516. } else {
  1517. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1518. }
  1519. }
  1520. }));
  1521. return stylesheetContent;
  1522. async function getStylesheetContent(resourceURL) {
  1523. const content = await util.getContent(resourceURL, {
  1524. maxResourceSize: options.maxResourceSize,
  1525. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1526. validateTextContentType: true,
  1527. frameId: options.frameId,
  1528. charset: options.charset,
  1529. resourceReferrer: options.resourceReferrer
  1530. });
  1531. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1532. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1533. return getStylesheetContent(resourceURL);
  1534. } else {
  1535. return content;
  1536. }
  1537. }
  1538. }
  1539. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet) {
  1540. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1541. urlFunctions.map(urlFunction => {
  1542. const originalResourceURL = matchURL(urlFunction);
  1543. let resourceURL = normalizeURL(originalResourceURL);
  1544. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1545. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1546. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1547. }
  1548. if (!testIgnoredPath(resourceURL)) {
  1549. if (!resourceURL || testValidPath(resourceURL)) {
  1550. let resolvedURL;
  1551. if (!originalResourceURL.startsWith("#")) {
  1552. try {
  1553. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1554. } catch (error) {
  1555. // ignored
  1556. }
  1557. }
  1558. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1559. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1560. }
  1561. } else {
  1562. let newUrlFunction;
  1563. if (originalResourceURL) {
  1564. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1565. } else {
  1566. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1567. }
  1568. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1569. }
  1570. }
  1571. });
  1572. return stylesheetContent;
  1573. }
  1574. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1575. resourceURL = normalizeURL(resourceURL);
  1576. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1577. const content = await util.getContent(resourceURL, {
  1578. maxResourceSize: options.maxResourceSize,
  1579. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1580. charset: options.charset,
  1581. frameId: options.frameId,
  1582. resourceReferrer: options.resourceReferrer,
  1583. validateTextContentType: true
  1584. });
  1585. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1586. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1587. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1588. }
  1589. resourceURL = content.resourceURL;
  1590. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1591. let stylesheetContent = removeCssComments(content.data);
  1592. if (options.compressCSS) {
  1593. stylesheetContent = util.compressCSS(stylesheetContent);
  1594. }
  1595. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1596. return stylesheetContent;
  1597. }
  1598. }
  1599. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1600. const promises = [];
  1601. const removedRules = [];
  1602. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1603. const ruleData = cssRule.data;
  1604. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1605. removedRules.push(cssRule);
  1606. } else if (ruleData.block && ruleData.block.children) {
  1607. if (ruleData.type == "Rule") {
  1608. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1609. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1610. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1611. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1612. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1613. }
  1614. }
  1615. }
  1616. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1617. await Promise.all(promises);
  1618. async function processFontFaceRule(ruleData, fontURLs) {
  1619. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1620. if (declaration.type == "Declaration" && declaration.value.children) {
  1621. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1622. await Promise.all(urlFunctions.map(async urlFunction => {
  1623. const originalResourceURL = matchURL(urlFunction);
  1624. const resourceURL = normalizeURL(originalResourceURL);
  1625. if (!testIgnoredPath(resourceURL)) {
  1626. if (testValidURL(resourceURL)) {
  1627. let { content } = await batchRequest.addURL(resourceURL, true, "font");
  1628. let resourceURLs = fontURLs.get(declaration);
  1629. if (!resourceURLs) {
  1630. resourceURLs = [];
  1631. fontURLs.set(declaration, resourceURLs);
  1632. }
  1633. resourceURLs.push(resourceURL);
  1634. replaceURLs(declaration, originalResourceURL, content);
  1635. }
  1636. }
  1637. }));
  1638. }
  1639. }));
  1640. function replaceURLs(declaration, oldURL, newURL) {
  1641. declaration.value.children.forEach(token => {
  1642. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1643. token.value.value = newURL;
  1644. }
  1645. });
  1646. }
  1647. }
  1648. }
  1649. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1650. await Promise.all(declarations.map(async declaration => {
  1651. if (declaration.value && !declaration.value.children && declaration.value.type == "Raw") {
  1652. try {
  1653. declaration.value = cssTree.parse(declaration.value.value, { context: "value" });
  1654. } catch (error) {
  1655. // ignored
  1656. }
  1657. }
  1658. if (declaration.type == "Declaration" && declaration.value.children) {
  1659. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1660. await Promise.all(urlFunctions.map(async urlFunction => {
  1661. const originalResourceURL = matchURL(urlFunction);
  1662. const resourceURL = normalizeURL(originalResourceURL);
  1663. if (!testIgnoredPath(resourceURL)) {
  1664. if (testValidURL(resourceURL)) {
  1665. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", true);
  1666. let variableDefined;
  1667. const tokens = [];
  1668. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1669. if (!originalResourceURL.startsWith("#")) {
  1670. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1671. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1672. tokens.push({ parent, token, value });
  1673. variableDefined = true;
  1674. } else {
  1675. token.data.value.value = content;
  1676. }
  1677. }
  1678. });
  1679. if (variableDefined) {
  1680. cssVariables.set(indexResource, content);
  1681. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1682. }
  1683. }
  1684. }
  1685. }));
  1686. }
  1687. }));
  1688. function findURLToken(url, children, callback, depth = 0) {
  1689. for (let token = children.head; token; token = token.next) {
  1690. if (token.data.children) {
  1691. findURLToken(url, token.data.children, callback, depth + 1);
  1692. }
  1693. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1694. callback(token, children, depth == 0);
  1695. }
  1696. }
  1697. }
  1698. }
  1699. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1700. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1701. let resourceURL = resourceElement.getAttribute(attributeName);
  1702. if (resourceURL != null) {
  1703. resourceURL = normalizeURL(resourceURL);
  1704. let originURL = resourceElement.dataset.singleFileOriginURL;
  1705. delete resourceElement.dataset.singleFileOriginURL;
  1706. if (!testIgnoredPath(resourceURL)) {
  1707. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1708. if (testValidPath(resourceURL)) {
  1709. try {
  1710. resourceURL = util.resolveURL(resourceURL, baseURI);
  1711. } catch (error) {
  1712. // ignored
  1713. }
  1714. if (testValidURL(resourceURL)) {
  1715. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, "image", resourceElement.tagName == "IMG" && attributeName == "src");
  1716. if (originURL) {
  1717. if (content == EMPTY_DATA_URI) {
  1718. try {
  1719. originURL = util.resolveURL(originURL, baseURI);
  1720. } catch (error) {
  1721. // ignored
  1722. }
  1723. try {
  1724. resourceURL = originURL;
  1725. content = (await util.getContent(resourceURL, {
  1726. asBinary: true,
  1727. expectedType: "image",
  1728. maxResourceSize: options.maxResourceSize,
  1729. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1730. frameId: options.windowId,
  1731. resourceReferrer: options.resourceReferrer
  1732. })).data;
  1733. } catch (error) {
  1734. // ignored
  1735. }
  1736. }
  1737. }
  1738. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1739. resourceElement.remove();
  1740. } else {
  1741. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1742. if (!forbiddenPrefixFound) {
  1743. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1744. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1745. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1746. cssVariables.set(indexResource, content);
  1747. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1748. styles.set(resourceElement, declarationList);
  1749. } else {
  1750. resourceElement.setAttribute(attributeName, content);
  1751. }
  1752. } else {
  1753. resourceElement.setAttribute(attributeName, content);
  1754. }
  1755. }
  1756. }
  1757. }
  1758. }
  1759. }
  1760. }
  1761. }));
  1762. }
  1763. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1764. let attributeName = "xlink:href";
  1765. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1766. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1767. if (originalResourceURL == null) {
  1768. attributeName = "href";
  1769. originalResourceURL = resourceElement.getAttribute(attributeName);
  1770. }
  1771. let resourceURL = normalizeURL(originalResourceURL);
  1772. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1773. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1774. try {
  1775. resourceURL = util.resolveURL(resourceURL, baseURI);
  1776. } catch (error) {
  1777. // ignored
  1778. }
  1779. if (testValidURL(resourceURL)) {
  1780. const { content } = await batchRequest.addURL(resourceURL);
  1781. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1782. if (hashMatch && hashMatch[0]) {
  1783. let symbolElement;
  1784. try {
  1785. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1786. } catch (error) {
  1787. // ignored
  1788. }
  1789. if (symbolElement) {
  1790. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1791. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1792. }
  1793. } else {
  1794. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1795. }
  1796. }
  1797. } else if (resourceURL == options.url) {
  1798. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1799. }
  1800. }));
  1801. }
  1802. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1803. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1804. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1805. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1806. let resourceURL = normalizeURL(srcsetValue.url);
  1807. if (!testIgnoredPath(resourceURL)) {
  1808. if (testValidPath(resourceURL)) {
  1809. try {
  1810. resourceURL = util.resolveURL(resourceURL, baseURI);
  1811. } catch (error) {
  1812. // ignored
  1813. }
  1814. if (testValidURL(resourceURL)) {
  1815. const { content } = await batchRequest.addURL(resourceURL, true, "image");
  1816. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1817. if (forbiddenPrefixFound) {
  1818. return "";
  1819. }
  1820. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1821. } else {
  1822. return "";
  1823. }
  1824. } else {
  1825. return "";
  1826. }
  1827. } else {
  1828. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1829. }
  1830. }));
  1831. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1832. }));
  1833. }
  1834. static replaceImageSource(imgElement, variableName, options) {
  1835. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1836. if (attributeValue) {
  1837. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1838. if (imageData && imageData.replaceable) {
  1839. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1840. const backgroundStyle = {};
  1841. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1842. if (backgroundSize) {
  1843. backgroundStyle["background-size"] = imageData.objectFit;
  1844. }
  1845. if (imageData.objectPosition) {
  1846. backgroundStyle["background-position"] = imageData.objectPosition;
  1847. }
  1848. if (imageData.backgroundColor) {
  1849. backgroundStyle["background-color"] = imageData.backgroundColor;
  1850. }
  1851. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1852. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1853. return true;
  1854. }
  1855. }
  1856. }
  1857. }
  1858. // ----
  1859. // Util
  1860. // ----
  1861. const BLOB_URI_PREFIX = "blob:";
  1862. const HTTP_URI_PREFIX = /^https?:\/\//;
  1863. const FILE_URI_PREFIX = /^file:\/\//;
  1864. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1865. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1866. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1867. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1868. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1869. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1870. 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;
  1871. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1872. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1873. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1874. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1875. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1876. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1877. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1878. function getUpdatedResourceContent(resourceURL, content, options) {
  1879. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1880. options.updatedResources[resourceURL].retrieved = true;
  1881. return options.updatedResources[resourceURL].content;
  1882. } else {
  1883. return content.data || "";
  1884. }
  1885. }
  1886. function normalizeURL(url) {
  1887. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1888. return url;
  1889. } else {
  1890. return url.split("#")[0];
  1891. }
  1892. }
  1893. function getCSSValue(value) {
  1894. let result = "";
  1895. try {
  1896. result = cssTree.generate(value);
  1897. } catch (error) {
  1898. // ignored
  1899. }
  1900. return result;
  1901. }
  1902. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  1903. const stylesheetCharset = getCharset(stylesheetContent);
  1904. if (stylesheetCharset) {
  1905. return stylesheetCharset == charset.toLowerCase();
  1906. } else {
  1907. return true;
  1908. }
  1909. }
  1910. function getCharset(stylesheetContent) {
  1911. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  1912. if (match && match[1]) {
  1913. return match[1].toLowerCase().trim();
  1914. }
  1915. }
  1916. function getOnEventAttributeNames(doc) {
  1917. const element = doc.createElement("div");
  1918. const attributeNames = [];
  1919. for (const propertyName in element) {
  1920. if (propertyName.startsWith("on")) {
  1921. attributeNames.push(propertyName);
  1922. }
  1923. }
  1924. attributeNames.push("onunload");
  1925. return attributeNames;
  1926. }
  1927. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1928. let maxLength;
  1929. if (template) {
  1930. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1931. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1932. if (template.match(replaceRegExp)) {
  1933. const matchedLength = template.match(replaceRegExp)[0];
  1934. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1935. if (isNaN(maxLength) || maxLength <= 0) {
  1936. maxLength = null;
  1937. }
  1938. } else {
  1939. replaceRegExp = new RegExp(regExpVariable, "g");
  1940. }
  1941. if (template.match(replaceRegExp)) {
  1942. let value = await valueGetter();
  1943. if (!dontReplaceSlash) {
  1944. value = value.replace(/\/+/g, replacementCharacter);
  1945. }
  1946. if (maxLength) {
  1947. value = await util.truncateText(value, maxLength);
  1948. }
  1949. return template.replace(replaceRegExp, value);
  1950. }
  1951. }
  1952. return template;
  1953. }
  1954. function getLastSegment(url, replacementCharacter) {
  1955. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1956. if (!lastSegment) {
  1957. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1958. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1959. }
  1960. if (!lastSegment) {
  1961. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1962. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1963. }
  1964. if (!lastSegment) {
  1965. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1966. }
  1967. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1968. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1969. lastSegment = lastSegmentMatch[1];
  1970. }
  1971. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1972. return lastSegment;
  1973. }
  1974. function getRegExp(string) {
  1975. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1976. }
  1977. function getUrlFunctions(stylesheetContent, unique) {
  1978. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1979. if (unique) {
  1980. return [...new Set(result)];
  1981. } else {
  1982. return result;
  1983. }
  1984. }
  1985. function getImportFunctions(stylesheetContent) {
  1986. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1987. }
  1988. function findShortcutIcon(shortcutIcons) {
  1989. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1990. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1991. return shortcutIcons[0];
  1992. }
  1993. function matchURL(stylesheetContent) {
  1994. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1995. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1996. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1997. return match && match[1];
  1998. }
  1999. function testIgnoredPath(resourceURL) {
  2000. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  2001. }
  2002. function testValidPath(resourceURL) {
  2003. return resourceURL && !resourceURL.match(EMPTY_URL);
  2004. }
  2005. function testValidURL(resourceURL) {
  2006. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  2007. }
  2008. function matchImport(stylesheetContent) {
  2009. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  2010. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  2011. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2012. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2013. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2014. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2015. if (match) {
  2016. const [, resourceURL, media] = match;
  2017. return { resourceURL, media };
  2018. }
  2019. }
  2020. function removeCssComments(stylesheetContent) {
  2021. try {
  2022. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2023. } catch (error) {
  2024. let start, end;
  2025. do {
  2026. start = stylesheetContent.indexOf("/*");
  2027. end = stylesheetContent.indexOf("*/", start + 2);
  2028. if (start != -1 && end != -1) {
  2029. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2030. }
  2031. } while (start != -1 && end != -1);
  2032. return stylesheetContent;
  2033. }
  2034. }
  2035. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2036. if (mediaQuery) {
  2037. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2038. } else {
  2039. return stylesheetContent;
  2040. }
  2041. }
  2042. function log(...args) {
  2043. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2044. }
  2045. // -----
  2046. // Stats
  2047. // -----
  2048. const STATS_DEFAULT_VALUES = {
  2049. discarded: {
  2050. "HTML bytes": 0,
  2051. "hidden elements": 0,
  2052. "HTML imports": 0,
  2053. scripts: 0,
  2054. objects: 0,
  2055. "audio sources": 0,
  2056. "video sources": 0,
  2057. frames: 0,
  2058. "CSS rules": 0,
  2059. canvas: 0,
  2060. stylesheets: 0,
  2061. resources: 0,
  2062. medias: 0
  2063. },
  2064. processed: {
  2065. "HTML bytes": 0,
  2066. "hidden elements": 0,
  2067. "HTML imports": 0,
  2068. scripts: 0,
  2069. objects: 0,
  2070. "audio sources": 0,
  2071. "video sources": 0,
  2072. frames: 0,
  2073. "CSS rules": 0,
  2074. canvas: 0,
  2075. stylesheets: 0,
  2076. resources: 0,
  2077. medias: 0
  2078. }
  2079. };
  2080. class Stats {
  2081. constructor(options) {
  2082. this.options = options;
  2083. if (options.displayStats) {
  2084. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2085. }
  2086. }
  2087. set(type, subType, value) {
  2088. if (this.options.displayStats) {
  2089. this.data[type][subType] = value;
  2090. }
  2091. }
  2092. add(type, subType, value) {
  2093. if (this.options.displayStats) {
  2094. this.data[type][subType] += value;
  2095. }
  2096. }
  2097. addAll(pageData) {
  2098. if (this.options.displayStats) {
  2099. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2100. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2101. }
  2102. }
  2103. }
  2104. return { getClass };
  2105. })();