single-file-core.js 88 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149
  1. /*
  2. * Copyright 2010-2020 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * The code in this file is free software: you can redistribute it and/or
  8. * modify it under the terms of the GNU Affero General Public License
  9. * (GNU AGPL) as published by the Free Software Foundation, either version 3
  10. * of the License, or (at your option) any later version.
  11. *
  12. * The code in this file is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero
  15. * General Public License for more details.
  16. *
  17. * As additional permission under GNU AGPL version 3 section 7, you may
  18. * distribute UNMODIFIED VERSIONS OF THIS file without the copy of the GNU
  19. * AGPL normally required by section 4, provided you include this license
  20. * notice and a URL through which recipients can access the Corresponding
  21. * Source.
  22. */
  23. this.singlefile.lib.core = this.singlefile.lib.core || (() => {
  24. const DEBUG = false;
  25. let util, cssTree, sessionId = 0;
  26. function getClass(...args) {
  27. [util, cssTree] = args;
  28. return SingleFileClass;
  29. }
  30. class SingleFileClass {
  31. constructor(options) {
  32. this.options = options;
  33. if (options.sessionId === undefined) {
  34. options.sessionId = sessionId;
  35. sessionId++;
  36. }
  37. }
  38. async run() {
  39. if (this.options.userScriptEnabled) {
  40. await util.waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  41. }
  42. this.runner = new Runner(this.options, true);
  43. await this.runner.loadPage();
  44. await this.runner.initialize();
  45. if (this.options.userScriptEnabled) {
  46. await util.waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  47. }
  48. await this.runner.run();
  49. }
  50. cancel() {
  51. this.cancelled = true;
  52. if (this.runner) {
  53. this.runner.cancel();
  54. }
  55. }
  56. getPageData() {
  57. return this.runner.getPageData();
  58. }
  59. }
  60. // -------------
  61. // ProgressEvent
  62. // -------------
  63. const PAGE_LOADING = "page-loading";
  64. const PAGE_LOADED = "page-loaded";
  65. const RESOURCES_INITIALIZING = "resource-initializing";
  66. const RESOURCES_INITIALIZED = "resources-initialized";
  67. const RESOURCE_LOADED = "resource-loaded";
  68. const PAGE_ENDED = "page-ended";
  69. const STAGE_STARTED = "stage-started";
  70. const STAGE_ENDED = "stage-ended";
  71. const STAGE_TASK_STARTED = "stage-task-started";
  72. const STAGE_TASK_ENDED = "stage-task-ended";
  73. class ProgressEvent {
  74. constructor(type, detail) {
  75. return { type, detail, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED, STAGE_STARTED, STAGE_ENDED, STAGE_TASK_STARTED, STAGE_TASK_ENDED };
  76. }
  77. }
  78. // ------
  79. // Runner
  80. // ------
  81. const RESOLVE_URLS_STAGE = 0;
  82. const REPLACE_DATA_STAGE = 1;
  83. const REPLACE_DOCS_STAGE = 2;
  84. const POST_PROCESS_STAGE = 3;
  85. const STAGES = [{
  86. sequential: [
  87. { action: "preProcessPage" },
  88. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  89. { action: "replaceStyleContents" },
  90. { action: "resetCharsetMeta" },
  91. { option: "saveFavicon", action: "saveFavicon" },
  92. { action: "replaceCanvasElements" },
  93. { action: "insertFonts" },
  94. { action: "insertShadowRootContents" },
  95. { action: "setInputValues" },
  96. { option: "removeScripts", action: "removeScripts" },
  97. { option: "selected", action: "removeUnselectedElements" },
  98. { option: "removeVideoSrc", action: "insertVideoPosters" },
  99. { option: "removeFrames", action: "removeFrames" },
  100. { option: "removeVideoSrc", action: "removeVideoSources" },
  101. { option: "removeAudioSrc", action: "removeAudioSources" },
  102. { action: "removeDiscardedResources" },
  103. { option: "removeHiddenElements", action: "removeHiddenElements" },
  104. { action: "resolveHrefs" },
  105. { action: "resolveStyleAttributeURLs" }
  106. ],
  107. parallel: [
  108. { action: "resolveStylesheetURLs" },
  109. { option: "!removeFrames", action: "resolveFrameURLs" },
  110. { action: "resolveHtmlImportURLs" }
  111. ]
  112. }, {
  113. sequential: [
  114. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  115. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  116. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  117. ],
  118. parallel: [
  119. { action: "processStylesheets" },
  120. { action: "processStyleAttributes" },
  121. { action: "processPageResources" },
  122. { option: "!removeScripts", action: "processScripts" }
  123. ]
  124. }, {
  125. sequential: [
  126. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  127. ],
  128. parallel: [
  129. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  130. { option: "!removeFrames", action: "processFrames" },
  131. { option: "!removeImports", action: "processHtmlImports" },
  132. ]
  133. }, {
  134. sequential: [
  135. { action: "replaceStylesheets" },
  136. { action: "replaceStyleAttributes" },
  137. { action: "insertVariables" },
  138. { option: "compressHTML", action: "compressHTML" },
  139. { action: "cleanupPage" }
  140. ],
  141. parallel: [
  142. { option: "enableMaff", action: "insertMAFFMetaData" },
  143. { action: "setDocInfo" }
  144. ]
  145. }];
  146. class Runner {
  147. constructor(options, root) {
  148. const rootDocDefined = root && options.doc;
  149. this.root = root;
  150. this.options = options;
  151. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  152. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  153. this.options.rootDocument = root;
  154. this.options.updatedResources = this.options.updatedResources || {};
  155. this.batchRequest = new BatchRequest();
  156. this.processor = new Processor(options, this.batchRequest);
  157. if (rootDocDefined) {
  158. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  159. this.options.canvases = docData.canvases;
  160. this.options.fonts = docData.fonts;
  161. this.options.stylesheets = docData.stylesheets;
  162. this.options.images = docData.images;
  163. this.options.posters = docData.posters;
  164. this.options.usedFonts = docData.usedFonts;
  165. this.options.shadowRoots = docData.shadowRoots;
  166. this.options.imports = docData.imports;
  167. this.options.referrer = docData.referrer;
  168. this.markedElements = docData.markedElements;
  169. }
  170. if (this.options.saveRawPage) {
  171. this.options.removeFrames = true;
  172. }
  173. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  174. this.onprogress = options.onprogress || (() => { });
  175. }
  176. async loadPage() {
  177. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  178. await this.processor.loadPage(this.options.content);
  179. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  180. }
  181. async initialize() {
  182. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  183. await this.executeStage(RESOLVE_URLS_STAGE);
  184. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  185. if (this.root && this.options.doc) {
  186. util.postProcessDoc(this.options.doc, this.markedElements);
  187. }
  188. }
  189. cancel() {
  190. this.cancelled = true;
  191. this.batchRequest.cancel();
  192. if (this.root) {
  193. if (this.options.frames) {
  194. this.options.frames.forEach(cancelRunner);
  195. }
  196. if (this.options.imports) {
  197. this.options.imports.forEach(cancelRunner);
  198. }
  199. }
  200. function cancelRunner(resourceData) {
  201. if (resourceData.runner) {
  202. resourceData.runner.cancel();
  203. }
  204. }
  205. }
  206. async run() {
  207. if (this.root) {
  208. this.processor.initialize(this.batchRequest);
  209. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  210. }
  211. await this.batchRequest.run(detail => {
  212. detail.pageURL = this.options.url;
  213. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  214. }, this.options);
  215. await this.pendingPromises;
  216. this.options.doc = null;
  217. this.options.win = null;
  218. await this.executeStage(REPLACE_DOCS_STAGE);
  219. await this.executeStage(POST_PROCESS_STAGE);
  220. this.processor.finalize();
  221. }
  222. getDocument() {
  223. return this.processor.doc;
  224. }
  225. getStyleSheets() {
  226. return this.processor.stylesheets;
  227. }
  228. getPageData() {
  229. if (this.root) {
  230. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  231. }
  232. return this.processor.getPageData();
  233. }
  234. async executeStage(step) {
  235. if (DEBUG) {
  236. log("**** STARTED STAGE", step, "****");
  237. }
  238. const frame = !this.root;
  239. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  240. STAGES[step].sequential.forEach(task => {
  241. let startTime;
  242. if (DEBUG) {
  243. startTime = Date.now();
  244. log(" -- STARTED task =", task.action);
  245. }
  246. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  247. if (!this.cancelled) {
  248. this.executeTask(task);
  249. }
  250. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  251. if (DEBUG) {
  252. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  253. }
  254. });
  255. let parallelTasksPromise;
  256. if (STAGES[step].parallel) {
  257. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  258. let startTime;
  259. if (DEBUG) {
  260. startTime = Date.now();
  261. log(" // STARTED task =", task.action);
  262. }
  263. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  264. if (!this.cancelled) {
  265. await this.executeTask(task);
  266. }
  267. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  268. if (DEBUG) {
  269. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  270. }
  271. }));
  272. } else {
  273. parallelTasksPromise = Promise.resolve();
  274. }
  275. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  276. if (DEBUG) {
  277. log("**** ENDED STAGE", step, "****");
  278. }
  279. return parallelTasksPromise;
  280. }
  281. executeTask(task) {
  282. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  283. return this.processor[task.action]();
  284. }
  285. }
  286. }
  287. // ------------
  288. // BatchRequest
  289. // ------------
  290. class BatchRequest {
  291. constructor() {
  292. this.requests = new Map();
  293. this.duplicates = new Map();
  294. }
  295. addURL(resourceURL, asBinary, groupDuplicates) {
  296. return new Promise((resolve, reject) => {
  297. const requestKey = JSON.stringify([resourceURL, asBinary]);
  298. let resourceRequests = this.requests.get(requestKey);
  299. if (!resourceRequests) {
  300. resourceRequests = [];
  301. this.requests.set(requestKey, resourceRequests);
  302. }
  303. const callbacks = { resolve, reject };
  304. resourceRequests.push(callbacks);
  305. if (groupDuplicates) {
  306. let duplicateRequests = this.duplicates.get(requestKey);
  307. if (!duplicateRequests) {
  308. duplicateRequests = [];
  309. this.duplicates.set(requestKey, duplicateRequests);
  310. }
  311. duplicateRequests.push(callbacks);
  312. }
  313. });
  314. }
  315. getMaxResources() {
  316. return this.requests.size;
  317. }
  318. run(onloadListener, options) {
  319. const resourceURLs = [...this.requests.keys()];
  320. let indexResource = 0;
  321. return Promise.all(resourceURLs.map(async requestKey => {
  322. const [resourceURL, asBinary] = JSON.parse(requestKey);
  323. const resourceRequests = this.requests.get(requestKey);
  324. try {
  325. const currentIndexResource = indexResource;
  326. indexResource = indexResource + 1;
  327. const content = await util.getContent(resourceURL, {
  328. asBinary,
  329. maxResourceSize: options.maxResourceSize,
  330. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  331. frameId: options.windowId
  332. });
  333. onloadListener({ url: resourceURL });
  334. if (!this.cancelled) {
  335. resourceRequests.forEach(callbacks => {
  336. const duplicateCallbacks = this.duplicates.get(requestKey);
  337. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  338. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  339. });
  340. }
  341. } catch (error) {
  342. indexResource = indexResource + 1;
  343. onloadListener({ url: resourceURL });
  344. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  345. }
  346. this.requests.delete(requestKey);
  347. }));
  348. }
  349. cancel() {
  350. this.cancelled = true;
  351. const resourceURLs = [...this.requests.keys()];
  352. resourceURLs.forEach(requestKey => {
  353. const resourceRequests = this.requests.get(requestKey);
  354. resourceRequests.forEach(callbacks => callbacks.reject());
  355. this.requests.delete(requestKey);
  356. });
  357. }
  358. }
  359. // ---------
  360. // Processor
  361. // ---------
  362. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  363. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  364. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  365. const SCRIPT_TAG_FOUND = /<script/gi;
  366. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  367. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  368. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  369. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  370. const UTF8_CHARSET = "utf-8";
  371. const SINGLE_FILE_UI_ELEMENT_CLASS = "single-file-ui-element";
  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. }
  382. initialize() {
  383. this.options.saveDate = new Date();
  384. this.options.saveUrl = this.options.url;
  385. if (this.options.enableMaff) {
  386. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  387. }
  388. this.maxResources = this.batchRequest.getMaxResources();
  389. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  390. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  391. }
  392. if (!this.options.removeImports && this.options.imports) {
  393. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  394. }
  395. this.stats.set("processed", "resources", this.maxResources);
  396. }
  397. async loadPage(pageContent, charset) {
  398. let content;
  399. if (!pageContent || this.options.saveRawPage) {
  400. content = await util.getContent(this.baseURI, {
  401. maxResourceSize: this.options.maxResourceSize,
  402. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  403. charset,
  404. frameId: this.options.windowId
  405. });
  406. pageContent = content.data;
  407. }
  408. this.doc = util.parseDocContent(pageContent, this.baseURI);
  409. if (this.options.saveRawPage) {
  410. let charset;
  411. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  412. const charsetDeclaration = element.content.split(";")[1];
  413. if (charsetDeclaration && !charset) {
  414. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  415. }
  416. });
  417. if (charset && charset != content.charset) {
  418. return this.loadPage(pageContent, charset);
  419. }
  420. }
  421. this.workStyleElement = this.doc.createElement("style");
  422. this.doc.body.appendChild(this.workStyleElement);
  423. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  424. }
  425. finalize() {
  426. if (this.workStyleElement.parentNode) {
  427. this.workStyleElement.remove();
  428. }
  429. }
  430. async getPageData() {
  431. util.postProcessDoc(this.doc);
  432. const url = util.parseURL(this.baseURI);
  433. if (this.options.insertSingleFileComment) {
  434. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  435. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  436. " \n url: " + this.options.saveUrl +
  437. " \n saved date: " + this.options.saveDate +
  438. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  439. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  440. }
  441. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  442. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  443. if (!canonicalLink) {
  444. canonicalLink = this.doc.createElement("link");
  445. canonicalLink.setAttribute("rel", "canonical");
  446. this.doc.head.appendChild(canonicalLink);
  447. }
  448. if (canonicalLink && !canonicalLink.href) {
  449. canonicalLink.href = this.options.saveUrl;
  450. }
  451. }
  452. let size;
  453. if (this.options.displayStats) {
  454. size = util.getContentSize(this.doc.documentElement.outerHTML);
  455. }
  456. const content = util.serialize(this.doc, this.options.compressHTML);
  457. if (this.options.displayStats) {
  458. const contentSize = util.getContentSize(content);
  459. this.stats.set("processed", "HTML bytes", contentSize);
  460. this.stats.add("discarded", "HTML bytes", size - contentSize);
  461. }
  462. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  463. const replacementCharacter = this.options.filenameReplacementCharacter;
  464. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  465. if (!this.options.backgroundSave) {
  466. filename = filename.replace(/\//g, replacementCharacter);
  467. }
  468. if (!this.options.saveToGDrive && util.getContentSize(filename) > this.options.filenameMaxLength) {
  469. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  470. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  471. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  472. filename = filename + "…" + extension;
  473. }
  474. if (!filename) {
  475. filename = "Unnamed page";
  476. }
  477. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  478. const pageData = {
  479. stats: this.stats.data,
  480. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  481. filename,
  482. content
  483. };
  484. if (this.options.addProof) {
  485. pageData.hash = await util.digest("SHA-256", content);
  486. }
  487. if (this.options.retrieveLinks) {
  488. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  489. }
  490. return pageData;
  491. }
  492. preProcessPage() {
  493. if (this.options.win) {
  494. 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));
  495. }
  496. if (this.options.images && !this.options.saveRawPage) {
  497. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  498. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  499. if (attributeValue) {
  500. const imageData = this.options.images[Number(attributeValue)];
  501. if (imageData) {
  502. if (this.options.removeHiddenElements && (
  503. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  504. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  505. )) {
  506. imgElement.setAttribute("src", EMPTY_IMAGE);
  507. } else {
  508. if (imageData.currentSrc) {
  509. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  510. imgElement.setAttribute("src", imageData.currentSrc);
  511. }
  512. if (this.options.loadDeferredImages) {
  513. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  514. imageData.src = imgElement.dataset.src;
  515. imgElement.setAttribute("src", imgElement.dataset.src);
  516. imgElement.removeAttribute("data-src");
  517. }
  518. }
  519. }
  520. }
  521. }
  522. });
  523. if (this.options.loadDeferredImages) {
  524. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  525. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  526. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  527. imgElement.removeAttribute("data-srcset");
  528. }
  529. });
  530. }
  531. }
  532. }
  533. replaceStyleContents() {
  534. if (this.options.stylesheets) {
  535. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  536. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  537. if (attributeValue) {
  538. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  539. if (stylesheetContent) {
  540. styleElement.textContent = stylesheetContent;
  541. }
  542. }
  543. });
  544. }
  545. }
  546. removeUnselectedElements() {
  547. removeUnmarkedElements(this.doc.body);
  548. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  549. function removeUnmarkedElements(element) {
  550. let selectedElementFound = false;
  551. Array.from(element.childNodes).forEach(node => {
  552. if (node.nodeType == 1) {
  553. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  554. selectedElementFound = selectedElementFound || isSelectedElement;
  555. if (isSelectedElement) {
  556. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  557. removeUnmarkedElements(node);
  558. } else if (selectedElementFound) {
  559. removeNode(node);
  560. } else {
  561. hideNode(node);
  562. }
  563. }
  564. });
  565. }
  566. function removeNode(node) {
  567. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  568. node.remove();
  569. } else {
  570. hideNode(node);
  571. }
  572. }
  573. function hideNode(node) {
  574. if (canHideNode(node)) {
  575. node.style.setProperty("display", "none", "important");
  576. Array.from(node.childNodes).forEach(removeNode);
  577. }
  578. }
  579. function canHideNode(node) {
  580. if (node.nodeType == 1) {
  581. const tagName = node.tagName && node.tagName.toLowerCase();
  582. return (tagName != "svg" && tagName != "style" && tagName != "link");
  583. }
  584. }
  585. }
  586. insertVideoPosters() {
  587. if (this.options.posters) {
  588. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  589. let videoElement;
  590. if (element.tagName == "VIDEO") {
  591. videoElement = element;
  592. } else {
  593. videoElement = element.parentElement;
  594. }
  595. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  596. if (attributeValue) {
  597. const posterURL = this.options.posters[Number(attributeValue)];
  598. if (!videoElement.poster && posterURL) {
  599. videoElement.setAttribute("poster", posterURL);
  600. }
  601. }
  602. });
  603. }
  604. }
  605. removeFrames() {
  606. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  607. this.stats.set("discarded", "frames", frameElements.length);
  608. this.stats.set("processed", "frames", frameElements.length);
  609. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  610. }
  611. removeScripts() {
  612. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  613. this.doc.querySelectorAll("[href]").forEach(element => {
  614. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  615. element.setAttribute("href", "");
  616. }
  617. });
  618. this.doc.querySelectorAll("[src]").forEach(element => {
  619. if (element.src && element.src.match(/^\s*javascript:/)) {
  620. element.removeAttribute("src");
  621. }
  622. });
  623. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  624. this.stats.set("discarded", "scripts", scriptElements.length);
  625. this.stats.set("processed", "scripts", scriptElements.length);
  626. scriptElements.forEach(element => element.remove());
  627. }
  628. removeVideoSources() {
  629. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  630. this.stats.set("discarded", "video sources", videoSourceElements.length);
  631. this.stats.set("processed", "video sources", videoSourceElements.length);
  632. videoSourceElements.forEach(element => {
  633. if (element.tagName == "SOURCE") {
  634. element.remove();
  635. } else {
  636. videoSourceElements.forEach(element => element.removeAttribute("src"));
  637. }
  638. });
  639. }
  640. removeAudioSources() {
  641. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  642. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  643. this.stats.set("processed", "audio sources", audioSourceElements.length);
  644. audioSourceElements.forEach(element => {
  645. if (element.tagName == "SOURCE") {
  646. element.remove();
  647. } else {
  648. audioSourceElements.forEach(element => element.removeAttribute("src"));
  649. }
  650. });
  651. }
  652. removeDiscardedResources() {
  653. this.doc.querySelectorAll("." + SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  654. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  655. 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\"])");
  656. this.stats.set("discarded", "objects", objectElements.length);
  657. this.stats.set("processed", "objects", objectElements.length);
  658. objectElements.forEach(element => element.remove());
  659. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  660. replacedAttributeValue.forEach(element => {
  661. let regExp;
  662. if (this.options.removeScripts) {
  663. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch)/g;
  664. } else {
  665. regExp = /(preconnect|prerender|dns-prefetch|prefetch)/g;
  666. }
  667. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  668. if (relValue.length) {
  669. element.setAttribute("rel", relValue);
  670. } else {
  671. element.remove();
  672. }
  673. });
  674. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  675. if (this.options.compressHTML) {
  676. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  677. }
  678. if (!this.options.saveFavicon) {
  679. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  680. }
  681. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  682. }
  683. resetCharsetMeta() {
  684. let charset;
  685. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  686. const charsetDeclaration = element.content.split(";")[1];
  687. if (charsetDeclaration && !charset) {
  688. charset = charsetDeclaration.split("=")[1];
  689. if (charset) {
  690. this.charset = charset.trim().toLowerCase();
  691. }
  692. }
  693. element.remove();
  694. });
  695. const metaElement = this.doc.createElement("meta");
  696. metaElement.setAttribute("charset", UTF8_CHARSET);
  697. if (this.doc.head.firstChild) {
  698. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  699. } else {
  700. this.doc.head.appendChild(metaElement);
  701. }
  702. }
  703. setInputValues() {
  704. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  705. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  706. input.setAttribute("value", value || "");
  707. });
  708. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  709. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  710. if (value == "true") {
  711. input.setAttribute("checked", "");
  712. }
  713. });
  714. this.doc.querySelectorAll("textarea").forEach(textarea => {
  715. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  716. textarea.textContent = value || "";
  717. });
  718. this.doc.querySelectorAll("select").forEach(select => {
  719. select.querySelectorAll("option").forEach(option => {
  720. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  721. if (selected) {
  722. option.setAttribute("selected", "");
  723. }
  724. });
  725. });
  726. }
  727. saveFavicon() {
  728. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  729. if (!faviconElement) {
  730. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  731. }
  732. if (!faviconElement) {
  733. faviconElement = this.doc.createElement("link");
  734. faviconElement.setAttribute("type", "image/x-icon");
  735. faviconElement.setAttribute("rel", "shortcut icon");
  736. faviconElement.setAttribute("href", "/favicon.ico");
  737. }
  738. this.doc.head.appendChild(faviconElement);
  739. }
  740. replaceCanvasElements() {
  741. if (this.options.canvases) {
  742. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  743. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  744. if (attributeValue) {
  745. const canvasData = this.options.canvases[Number(attributeValue)];
  746. if (canvasData) {
  747. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  748. this.stats.add("processed", "canvas", 1);
  749. }
  750. }
  751. });
  752. }
  753. }
  754. insertFonts() {
  755. if (this.options.fonts && this.options.fonts.length) {
  756. let stylesheetContent = "";
  757. this.options.fonts.forEach(fontData => {
  758. if (fontData["font-family"] && fontData.src) {
  759. stylesheetContent += "@font-face{";
  760. let stylesContent = "";
  761. Object.keys(fontData).forEach(fontStyle => {
  762. if (stylesContent) {
  763. stylesContent += ";";
  764. }
  765. stylesContent += fontStyle + ":" + fontData[fontStyle];
  766. });
  767. stylesheetContent += stylesContent + "}";
  768. }
  769. });
  770. if (stylesheetContent) {
  771. const styleElement = this.doc.createElement("style");
  772. styleElement.textContent = stylesheetContent;
  773. const existingStyleElement = this.doc.querySelector("style");
  774. if (existingStyleElement) {
  775. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  776. } else {
  777. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  778. }
  779. }
  780. }
  781. }
  782. removeHiddenElements() {
  783. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  784. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  785. this.stats.set("discarded", "hidden elements", removedElements.length);
  786. this.stats.set("processed", "hidden elements", removedElements.length);
  787. if (hiddenElements.length) {
  788. const styleElement = this.doc.createElement("style");
  789. styleElement.textContent = ".sf-hidden{display:none!important;}";
  790. this.doc.head.appendChild(styleElement);
  791. hiddenElements.forEach(element => {
  792. if (element.style.display != "none") {
  793. element.classList.add("sf-hidden");
  794. }
  795. });
  796. }
  797. removedElements.forEach(element => element.remove());
  798. }
  799. resolveHrefs() {
  800. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  801. const href = element.getAttribute("href").trim();
  802. if (!testIgnoredPath(href)) {
  803. let resolvedURL;
  804. try {
  805. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  806. } catch (error) {
  807. // ignored
  808. }
  809. if (resolvedURL) {
  810. const url = normalizeURL(this.options.url);
  811. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  812. resolvedURL = resolvedURL.substring(url.length);
  813. }
  814. try {
  815. element.setAttribute("href", resolvedURL);
  816. } catch (error) {
  817. // ignored
  818. }
  819. }
  820. }
  821. });
  822. }
  823. resolveStyleAttributeURLs() {
  824. this.doc.querySelectorAll("[style]").forEach(element => {
  825. let styleContent = element.getAttribute("style");
  826. if (this.options.compressCSS) {
  827. styleContent = util.compressCSS(styleContent);
  828. }
  829. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  830. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  831. this.styles.set(element, declarationList);
  832. });
  833. }
  834. async resolveStylesheetURLs() {
  835. const options = {
  836. maxResourceSize: this.options.maxResourceSize,
  837. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  838. url: this.options.url,
  839. charset: this.charset,
  840. compressCSS: this.options.compressCSS,
  841. updatedResources: this.options.updatedResources,
  842. rootDocument: this.options.rootDocument,
  843. frameId: this.options.windowId
  844. };
  845. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  846. .map(async element => {
  847. let mediaText;
  848. if (element.media) {
  849. mediaText = element.media.toLowerCase();
  850. }
  851. const stylesheetInfo = { mediaText };
  852. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  853. stylesheetInfo.scoped = true;
  854. }
  855. if (element.tagName.toLowerCase() == "link") {
  856. if (element.charset) {
  857. options.charset = element.charset;
  858. }
  859. }
  860. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  861. }));
  862. if (options.rootDocument) {
  863. const newResources = Object.keys(options.updatedResources).filter(url => options.updatedResources[url].type == "stylesheet" && !options.updatedResources[url].retrieved).map(url => options.updatedResources[url]);
  864. await Promise.all(newResources.map(async resource => {
  865. resource.retrieved = true;
  866. const stylesheetInfo = {};
  867. const element = this.doc.createElement("style");
  868. this.doc.body.appendChild(element);
  869. element.textContent = resource.content;
  870. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  871. }));
  872. }
  873. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  874. stylesheets.set(element, stylesheetInfo);
  875. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  876. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  877. if (match && match[1] && match[1] != options.charset) {
  878. options.charset = match[1];
  879. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  880. }
  881. let stylesheet;
  882. try {
  883. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  884. } catch (error) {
  885. // ignored
  886. }
  887. if (stylesheet && stylesheet.children) {
  888. if (options.compressCSS) {
  889. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  890. }
  891. stylesheetInfo.stylesheet = stylesheet;
  892. } else {
  893. stylesheets.delete(element);
  894. }
  895. }
  896. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  897. let content;
  898. if (element.tagName.toLowerCase() == "link") {
  899. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  900. } else {
  901. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  902. }
  903. return content || "";
  904. }
  905. }
  906. async resolveFrameURLs() {
  907. if (!this.options.saveRawPage) {
  908. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  909. await Promise.all(frameElements.map(async frameElement => {
  910. if (frameElement.tagName == "OBJECT") {
  911. frameElement.setAttribute("data", "data:text/html,");
  912. } else {
  913. frameElement.removeAttribute("src");
  914. frameElement.removeAttribute("srcdoc");
  915. }
  916. Array.from(frameElement.childNodes).forEach(node => node.remove());
  917. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  918. if (this.options.frames && frameWindowId) {
  919. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  920. if (frameData) {
  921. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  922. }
  923. }
  924. }));
  925. }
  926. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  927. options.insertSingleFileComment = false;
  928. options.insertCanonicalLink = false;
  929. options.saveFavicon = false;
  930. options.url = frameData.baseURI;
  931. options.windowId = frameWindowId;
  932. if (frameData.content) {
  933. options.content = frameData.content;
  934. options.canvases = frameData.canvases;
  935. options.fonts = frameData.fonts;
  936. options.stylesheets = frameData.stylesheets;
  937. options.images = frameData.images;
  938. options.posters = frameData.posters;
  939. options.usedFonts = frameData.usedFonts;
  940. options.shadowRoots = frameData.shadowRoots;
  941. options.imports = frameData.imports;
  942. frameData.runner = new Runner(options);
  943. frameData.frameElement = frameElement;
  944. await frameData.runner.loadPage();
  945. await frameData.runner.initialize();
  946. frameData.maxResources = batchRequest.getMaxResources();
  947. }
  948. }
  949. }
  950. insertShadowRootContents() {
  951. const doc = this.doc;
  952. const options = this.options;
  953. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  954. processElement(this.doc);
  955. if (this.options.removeScripts) {
  956. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  957. }
  958. const scriptElement = doc.createElement("script");
  959. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  960. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{if (!element.parentElement.shadowRoot) {const shadowRoot=element.parentElement.attachShadow({mode:element.getAttribute("${SHADOW_MODE_ATTRIBUTE_NAME}"),delegatesFocus:Boolean(element.getAttribute("${SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME}"))});shadowRoot.innerHTML=element.innerHTML;element.remove();processNode(shadowRoot)}})}})()`;
  961. doc.body.appendChild(scriptElement);
  962. }
  963. function processElement(element) {
  964. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  965. shadowRootElements.forEach(element => {
  966. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  967. if (attributeValue) {
  968. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  969. if (shadowRootData) {
  970. const templateElement = doc.createElement("template");
  971. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  972. if (shadowRootData.delegatesFocus) {
  973. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME);
  974. }
  975. if (shadowRootData.adoptedStyleSheets) {
  976. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  977. const styleElement = doc.createElement("style");
  978. styleElement.textContent = stylesheetContent;
  979. templateElement.appendChild(styleElement);
  980. });
  981. }
  982. const shadowDoc = util.parseDocContent(shadowRootData.content);
  983. if (shadowDoc.head) {
  984. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  985. if (metaCharset) {
  986. metaCharset.remove();
  987. }
  988. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  989. }
  990. if (shadowDoc.body) {
  991. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  992. }
  993. processElement(templateElement);
  994. if (element.firstChild) {
  995. element.insertBefore(templateElement, element.firstChild);
  996. } else {
  997. element.appendChild(templateElement);
  998. }
  999. }
  1000. }
  1001. });
  1002. }
  1003. }
  1004. async resolveHtmlImportURLs() {
  1005. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1006. await Promise.all(linkElements.map(async linkElement => {
  1007. const resourceURL = linkElement.href;
  1008. linkElement.removeAttribute("href");
  1009. const options = Object.create(this.options);
  1010. options.insertSingleFileComment = false;
  1011. options.insertCanonicalLink = false;
  1012. options.saveFavicon = false;
  1013. options.removeUnusedStyles = false;
  1014. options.removeAlternativeMedias = false;
  1015. options.removeUnusedFonts = false;
  1016. options.removeHiddenElements = false;
  1017. options.url = resourceURL;
  1018. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1019. if (attributeValue) {
  1020. const importData = options.imports[Number(attributeValue)];
  1021. if (importData) {
  1022. options.content = importData.content;
  1023. importData.runner = new Runner(options);
  1024. await importData.runner.loadPage();
  1025. await importData.runner.initialize();
  1026. if (!options.removeImports) {
  1027. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1028. }
  1029. importData.runner.getStyleSheets().forEach(stylesheet => {
  1030. const importedStyleElement = this.doc.createElement("style");
  1031. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1032. this.stylesheets.set(importedStyleElement, stylesheet);
  1033. });
  1034. }
  1035. }
  1036. if (options.removeImports) {
  1037. linkElement.remove();
  1038. this.stats.add("discarded", "HTML imports", 1);
  1039. }
  1040. }));
  1041. }
  1042. removeUnusedStyles() {
  1043. if (!this.mediaAllInfo) {
  1044. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1045. }
  1046. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1047. this.stats.set("processed", "CSS rules", stats.processed);
  1048. this.stats.set("discarded", "CSS rules", stats.discarded);
  1049. }
  1050. removeUnusedFonts() {
  1051. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1052. }
  1053. removeAlternativeMedias() {
  1054. const stats = util.minifyMedias(this.stylesheets);
  1055. this.stats.set("processed", "medias", stats.processed);
  1056. this.stats.set("discarded", "medias", stats.discarded);
  1057. }
  1058. async processStylesheets() {
  1059. this.options.fontURLs = new Map();
  1060. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1061. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1062. ));
  1063. }
  1064. async processStyleAttributes() {
  1065. return Promise.all([...this.styles].map(([, declarationList]) =>
  1066. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1067. ));
  1068. }
  1069. async processPageResources() {
  1070. const processAttributeArgs = [
  1071. ["link[href][rel*=\"icon\"]", "href", false, true],
  1072. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1073. ["img[src], input[src][type=image]", "src", true],
  1074. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1075. ["video[poster]", "poster"],
  1076. ["*[background]", "background"],
  1077. ["image", "xlink:href"],
  1078. ["image", "href"]
  1079. ];
  1080. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1081. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1082. );
  1083. resourcePromises = resourcePromises.concat([
  1084. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  1085. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1086. ]);
  1087. if (!this.options.removeAudioSrc) {
  1088. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1089. }
  1090. if (!this.options.removeVideoSrc) {
  1091. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1092. }
  1093. await Promise.all(resourcePromises);
  1094. if (this.options.saveFavicon) {
  1095. ProcessorHelper.processShortcutIcons(this.doc);
  1096. }
  1097. }
  1098. async processScripts() {
  1099. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1100. let resourceURL;
  1101. let scriptSrc;
  1102. if (element.tagName == "SCRIPT") {
  1103. scriptSrc = element.getAttribute("src");
  1104. element.removeAttribute("src");
  1105. } else {
  1106. scriptSrc = element.getAttribute("href");
  1107. element.removeAttribute("href");
  1108. }
  1109. element.removeAttribute("integrity");
  1110. element.textContent = "";
  1111. try {
  1112. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1113. } catch (error) {
  1114. // ignored
  1115. }
  1116. if (testValidURL(resourceURL)) {
  1117. this.stats.add("processed", "scripts", 1);
  1118. const content = await util.getContent(resourceURL, {
  1119. asBinary: true,
  1120. charset: this.charset != UTF8_CHARSET && this.charset,
  1121. maxResourceSize: this.options.maxResourceSize,
  1122. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1123. frameId: this.options.windowId
  1124. });
  1125. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1126. if (element.tagName == "SCRIPT") {
  1127. element.setAttribute("src", content.data);
  1128. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1129. element.setAttribute("async", "");
  1130. }
  1131. } else {
  1132. const scriptElement = this.doc.createElement("script");
  1133. scriptElement.setAttribute("src", content.data);
  1134. scriptElement.setAttribute("async", "");
  1135. element.parentElement.replaceChild(scriptElement, element);
  1136. }
  1137. }
  1138. }));
  1139. }
  1140. removeAlternativeImages() {
  1141. util.removeAlternativeImages(this.doc);
  1142. }
  1143. async removeAlternativeFonts() {
  1144. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs);
  1145. }
  1146. async processFrames() {
  1147. if (this.options.frames) {
  1148. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1149. await Promise.all(frameElements.map(async frameElement => {
  1150. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1151. if (frameWindowId) {
  1152. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1153. if (frameData) {
  1154. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1155. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1156. this.stats.add("processed", "frames", 1);
  1157. await frameData.runner.run();
  1158. const pageData = await frameData.runner.getPageData();
  1159. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1160. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1161. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1162. sandbox += " allow-scripts allow-same-origin";
  1163. }
  1164. frameElement.setAttribute("sandbox", sandbox);
  1165. if (frameElement.tagName == "OBJECT") {
  1166. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1167. } else {
  1168. if (frameElement.tagName == "FRAME") {
  1169. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1170. } else {
  1171. frameElement.setAttribute("srcdoc", pageData.content);
  1172. frameElement.removeAttribute("src");
  1173. }
  1174. }
  1175. this.stats.addAll(pageData);
  1176. } else {
  1177. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1178. this.stats.add("discarded", "frames", 1);
  1179. }
  1180. }
  1181. }
  1182. }));
  1183. }
  1184. }
  1185. async processHtmlImports() {
  1186. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1187. await Promise.all(linkElements.map(async linkElement => {
  1188. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1189. if (attributeValue) {
  1190. const importData = this.options.imports[Number(attributeValue)];
  1191. if (importData.runner) {
  1192. this.stats.add("processed", "HTML imports", 1);
  1193. await importData.runner.run();
  1194. const pageData = await importData.runner.getPageData();
  1195. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1196. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1197. this.stats.addAll(pageData);
  1198. } else {
  1199. this.stats.add("discarded", "HTML imports", 1);
  1200. }
  1201. }
  1202. }));
  1203. }
  1204. replaceStylesheets() {
  1205. this.doc.querySelectorAll("style").forEach(styleElement => {
  1206. const stylesheetInfo = this.stylesheets.get(styleElement);
  1207. if (stylesheetInfo) {
  1208. this.stylesheets.delete(styleElement);
  1209. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1210. styleElement.textContent = stylesheetContent;
  1211. if (stylesheetInfo.mediaText) {
  1212. styleElement.media = stylesheetInfo.mediaText;
  1213. }
  1214. } else {
  1215. styleElement.remove();
  1216. }
  1217. });
  1218. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1219. const stylesheetInfo = this.stylesheets.get(linkElement);
  1220. if (stylesheetInfo) {
  1221. this.stylesheets.delete(linkElement);
  1222. const styleElement = this.doc.createElement("style");
  1223. if (stylesheetInfo.mediaText) {
  1224. styleElement.media = stylesheetInfo.mediaText;
  1225. }
  1226. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1227. styleElement.textContent = stylesheetContent;
  1228. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1229. } else {
  1230. linkElement.remove();
  1231. }
  1232. });
  1233. }
  1234. replaceStyleAttributes() {
  1235. this.doc.querySelectorAll("[style]").forEach(element => {
  1236. const declarations = this.styles.get(element);
  1237. if (declarations) {
  1238. this.styles.delete(element);
  1239. let styleContent = cssTree.generate(declarations);
  1240. element.setAttribute("style", styleContent);
  1241. } else {
  1242. element.setAttribute("style", "");
  1243. }
  1244. });
  1245. }
  1246. insertVariables() {
  1247. if (this.cssVariables.size) {
  1248. const styleElement = this.doc.createElement("style");
  1249. const firstStyleElement = this.doc.head.querySelector("style");
  1250. if (firstStyleElement) {
  1251. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1252. } else {
  1253. this.doc.head.appendChild(styleElement);
  1254. }
  1255. let stylesheetContent = "";
  1256. this.cssVariables.forEach((content, indexResource) => {
  1257. this.cssVariables.delete(indexResource);
  1258. if (stylesheetContent) {
  1259. stylesheetContent += ";";
  1260. }
  1261. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1262. });
  1263. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1264. }
  1265. }
  1266. compressHTML() {
  1267. let size;
  1268. if (this.options.displayStats) {
  1269. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1270. }
  1271. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1272. if (this.options.displayStats) {
  1273. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1274. }
  1275. }
  1276. cleanupPage() {
  1277. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1278. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1279. if (metaCharset) {
  1280. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1281. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1282. this.doc.head.querySelector("meta[charset]").remove();
  1283. }
  1284. }
  1285. }
  1286. resetZoomLevel() {
  1287. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1288. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1289. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1290. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1291. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1292. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1293. this.doc.documentElement.style.removeProperty("-sf-transform");
  1294. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1295. }
  1296. async insertMAFFMetaData() {
  1297. const maffMetaData = await this.maffMetaDataPromise;
  1298. if (maffMetaData && maffMetaData.content) {
  1299. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1300. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1301. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1302. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1303. if (originalURLElement) {
  1304. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1305. }
  1306. if (archiveTimeElement) {
  1307. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1308. if (value) {
  1309. const date = new Date(value);
  1310. if (!isNaN(date.getTime())) {
  1311. this.options.saveDate = new Date(value);
  1312. }
  1313. }
  1314. }
  1315. }
  1316. }
  1317. async setDocInfo() {
  1318. const titleElement = this.doc.querySelector("title");
  1319. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1320. const authorElement = this.doc.querySelector("meta[name=author]");
  1321. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1322. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1323. const headingElement = this.doc.querySelector("h1");
  1324. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1325. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1326. this.options.info = {
  1327. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1328. lang: this.doc.documentElement.lang,
  1329. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1330. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1331. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1332. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1333. };
  1334. }
  1335. }
  1336. // ---------------
  1337. // ProcessorHelper
  1338. // ---------------
  1339. const DATA_URI_PREFIX = "data:";
  1340. const ABOUT_BLANK_URI = "about:blank";
  1341. const EMPTY_DATA_URI = "data:null;base64,";
  1342. const REGEXP_URL_HASH = /(#.+?)$/;
  1343. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1344. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1345. class ProcessorHelper {
  1346. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1347. const date = options.saveDate;
  1348. const url = util.parseURL(options.saveUrl);
  1349. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1350. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1351. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1352. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1353. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1354. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1355. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1356. template = await evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1357. template = await evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1358. template = await evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1359. template = await evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1360. template = await evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1361. template = await evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1362. template = await evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1363. template = await evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1364. template = await evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1365. template = await evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1366. template = await evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1367. template = await evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1368. template = await evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1369. template = await evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1370. template = await evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1371. template = await evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1372. template = await evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1373. template = await evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1374. template = await evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1375. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1376. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1377. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1378. template = await evalTemplateVariable(template, "url-href", () => decode(url.href) || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1379. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1380. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1381. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", dontReplaceSlash, options.filenameReplacementCharacter);
  1382. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1383. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1384. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1385. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1386. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1387. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1388. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1389. for (const [name, value] of params) {
  1390. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1391. }
  1392. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1393. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1394. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1395. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1396. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1397. if (content) {
  1398. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1399. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1400. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1401. }
  1402. return template.trim();
  1403. function decode(value) {
  1404. try {
  1405. return decodeURI(value);
  1406. } catch (error) {
  1407. return value;
  1408. }
  1409. }
  1410. }
  1411. static setBackgroundImage(element, url, style) {
  1412. element.style.setProperty("background-blend-mode", "normal", "important");
  1413. element.style.setProperty("background-clip", "content-box", "important");
  1414. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1415. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1416. element.style.setProperty("background-image", url, "important");
  1417. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1418. element.style.setProperty("background-origin", "content-box", "important");
  1419. element.style.setProperty("background-repeat", "no-repeat", "important");
  1420. }
  1421. static processShortcutIcons(doc) {
  1422. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1423. if (!shortcutIcon) {
  1424. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1425. if (shortcutIcon) {
  1426. shortcutIcon.rel = "icon";
  1427. }
  1428. }
  1429. if (shortcutIcon) {
  1430. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1431. if (linkElement != shortcutIcon) {
  1432. linkElement.remove();
  1433. }
  1434. });
  1435. }
  1436. }
  1437. static removeSingleLineCssComments(stylesheet) {
  1438. const removedRules = [];
  1439. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1440. const ruleData = cssRule.data;
  1441. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1442. removedRules.push(cssRule);
  1443. }
  1444. }
  1445. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1446. }
  1447. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1448. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1449. const imports = getImportFunctions(stylesheetContent);
  1450. await Promise.all(imports.map(async cssImport => {
  1451. const match = matchImport(cssImport);
  1452. if (match) {
  1453. const regExpCssImport = getRegExp(cssImport);
  1454. let resourceURL = normalizeURL(match.resourceURL);
  1455. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1456. try {
  1457. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1458. } catch (error) {
  1459. // ignored
  1460. }
  1461. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1462. const content = await util.getContent(resourceURL, {
  1463. maxResourceSize: options.maxResourceSize,
  1464. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1465. validateTextContentType: true,
  1466. frameId: options.frameId
  1467. });
  1468. resourceURL = content.resourceURL;
  1469. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1470. let importedStylesheetContent = removeCssComments(content.data);
  1471. if (options.compressCSS) {
  1472. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1473. }
  1474. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1475. if (stylesheetContent.includes(cssImport)) {
  1476. const ancestorStyleSheets = new Set(importedStyleSheets);
  1477. ancestorStyleSheets.add(resourceURL);
  1478. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1479. workStylesheet.textContent = importedStylesheetContent;
  1480. if (workStylesheet.sheet.cssRules.length) {
  1481. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1482. } else {
  1483. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1484. }
  1485. }
  1486. } else {
  1487. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1488. }
  1489. } else {
  1490. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1491. }
  1492. }
  1493. }));
  1494. return stylesheetContent;
  1495. }
  1496. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1497. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1498. urlFunctions.map(urlFunction => {
  1499. const originalResourceURL = matchURL(urlFunction);
  1500. const resourceURL = normalizeURL(originalResourceURL);
  1501. if (!testIgnoredPath(resourceURL)) {
  1502. if (!resourceURL || testValidPath(resourceURL)) {
  1503. let resolvedURL;
  1504. if (!originalResourceURL.startsWith("#")) {
  1505. try {
  1506. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1507. } catch (error) {
  1508. // ignored
  1509. }
  1510. }
  1511. if (testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1512. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1513. }
  1514. } else {
  1515. let newUrlFunction;
  1516. if (originalResourceURL) {
  1517. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1518. } else {
  1519. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1520. }
  1521. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1522. }
  1523. }
  1524. });
  1525. return stylesheetContent;
  1526. }
  1527. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1528. resourceURL = normalizeURL(resourceURL);
  1529. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1530. const content = await util.getContent(resourceURL, {
  1531. maxResourceSize: options.maxResourceSize,
  1532. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1533. charset: options.charset,
  1534. frameId: options.frameId,
  1535. validateTextContentType: true
  1536. });
  1537. resourceURL = content.resourceURL;
  1538. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1539. let stylesheetContent = removeCssComments(content.data);
  1540. if (options.compressCSS) {
  1541. stylesheetContent = util.compressCSS(stylesheetContent);
  1542. }
  1543. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1544. return stylesheetContent;
  1545. }
  1546. }
  1547. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1548. const promises = [];
  1549. const removedRules = [];
  1550. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1551. const ruleData = cssRule.data;
  1552. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1553. removedRules.push(cssRule);
  1554. } else if (ruleData.block && ruleData.block.children) {
  1555. if (ruleData.type == "Rule") {
  1556. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1557. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1558. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1559. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1560. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1561. }
  1562. }
  1563. }
  1564. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1565. await Promise.all(promises);
  1566. async function processFontFaceRule(ruleData, fontURLs) {
  1567. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1568. if (declaration.type == "Declaration" && declaration.value.children) {
  1569. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1570. await Promise.all(urlFunctions.map(async urlFunction => {
  1571. const originalResourceURL = matchURL(urlFunction);
  1572. const resourceURL = normalizeURL(originalResourceURL);
  1573. if (!testIgnoredPath(resourceURL)) {
  1574. if (testValidURL(resourceURL)) {
  1575. let { content } = await batchRequest.addURL(resourceURL, true);
  1576. let resourceURLs = fontURLs.get(declaration);
  1577. if (!resourceURLs) {
  1578. resourceURLs = [];
  1579. fontURLs.set(declaration, resourceURLs);
  1580. }
  1581. resourceURLs.push(resourceURL);
  1582. replaceURLs(declaration, originalResourceURL, content);
  1583. }
  1584. }
  1585. }));
  1586. }
  1587. }));
  1588. function replaceURLs(declaration, oldURL, newURL) {
  1589. declaration.value.children.forEach(token => {
  1590. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1591. token.value.value = newURL;
  1592. }
  1593. });
  1594. }
  1595. }
  1596. }
  1597. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1598. await Promise.all(declarations.map(async declaration => {
  1599. if (declaration.type == "Declaration" && declaration.value.children) {
  1600. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1601. await Promise.all(urlFunctions.map(async urlFunction => {
  1602. const originalResourceURL = matchURL(urlFunction);
  1603. const resourceURL = normalizeURL(originalResourceURL);
  1604. if (!testIgnoredPath(resourceURL)) {
  1605. if (testValidURL(resourceURL)) {
  1606. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1607. let variableDefined;
  1608. const tokens = [];
  1609. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1610. if (!originalResourceURL.startsWith("#")) {
  1611. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1612. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1613. tokens.push({ parent, token, value });
  1614. variableDefined = true;
  1615. } else {
  1616. token.data.value.value = content;
  1617. }
  1618. }
  1619. });
  1620. if (variableDefined) {
  1621. cssVariables.set(indexResource, content);
  1622. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1623. }
  1624. }
  1625. }
  1626. }));
  1627. }
  1628. }));
  1629. function findURLToken(url, children, callback, depth = 0) {
  1630. for (let token = children.head; token; token = token.next) {
  1631. if (token.data.children) {
  1632. findURLToken(url, token.data.children, callback, depth + 1);
  1633. }
  1634. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1635. callback(token, children, depth == 0);
  1636. }
  1637. }
  1638. }
  1639. }
  1640. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1641. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1642. let resourceURL = resourceElement.getAttribute(attributeName);
  1643. if (resourceURL != null) {
  1644. resourceURL = normalizeURL(resourceURL);
  1645. if (!testIgnoredPath(resourceURL)) {
  1646. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1647. if (testValidPath(resourceURL)) {
  1648. try {
  1649. resourceURL = util.resolveURL(resourceURL, baseURI);
  1650. } catch (error) {
  1651. // ignored
  1652. }
  1653. if (testValidURL(resourceURL)) {
  1654. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1655. if (resourceElement.dataset.singleFileOriginURL) {
  1656. let originURL = resourceElement.dataset.singleFileOriginURL;
  1657. delete resourceElement.dataset.singleFileOriginURL;
  1658. if (content == EMPTY_DATA_URI) {
  1659. try {
  1660. originURL = util.resolveURL(originURL, baseURI);
  1661. } catch (error) {
  1662. // ignored
  1663. }
  1664. try {
  1665. resourceURL = originURL;
  1666. content = (await util.getContent(resourceURL, {
  1667. asBinary: true,
  1668. maxResourceSize: options.maxResourceSize,
  1669. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1670. frameId: options.windowId
  1671. })).data;
  1672. } catch (error) {
  1673. // ignored
  1674. }
  1675. }
  1676. }
  1677. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1678. resourceElement.remove();
  1679. } else {
  1680. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1681. if (!forbiddenPrefixFound) {
  1682. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1683. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1684. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1685. cssVariables.set(indexResource, content);
  1686. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1687. styles.set(resourceElement, declarationList);
  1688. } else {
  1689. resourceElement.setAttribute(attributeName, content);
  1690. }
  1691. } else {
  1692. resourceElement.setAttribute(attributeName, content);
  1693. }
  1694. }
  1695. }
  1696. }
  1697. }
  1698. }
  1699. }
  1700. }));
  1701. }
  1702. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1703. let attributeName = "xlink:href";
  1704. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1705. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1706. if (originalResourceURL == null) {
  1707. attributeName = "href";
  1708. originalResourceURL = resourceElement.getAttribute(attributeName);
  1709. }
  1710. let resourceURL = normalizeURL(originalResourceURL);
  1711. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1712. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1713. try {
  1714. resourceURL = util.resolveURL(resourceURL, baseURI);
  1715. } catch (error) {
  1716. // ignored
  1717. }
  1718. if (testValidURL(resourceURL)) {
  1719. const { content } = await batchRequest.addURL(resourceURL);
  1720. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1721. if (hashMatch && hashMatch[0]) {
  1722. let symbolElement;
  1723. try {
  1724. symbolElement = util.parseSVGContent(content).querySelector(hashMatch[0]);
  1725. } catch (error) {
  1726. // ignored
  1727. }
  1728. if (symbolElement) {
  1729. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1730. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1731. }
  1732. } else {
  1733. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1734. }
  1735. }
  1736. } else if (resourceURL == options.url) {
  1737. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1738. }
  1739. }));
  1740. }
  1741. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1742. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1743. const srcset = util.parseSrcset(resourceElement.getAttribute(attributeName));
  1744. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1745. let resourceURL = normalizeURL(srcsetValue.url);
  1746. if (!testIgnoredPath(resourceURL)) {
  1747. if (testValidPath(resourceURL)) {
  1748. try {
  1749. resourceURL = util.resolveURL(resourceURL, baseURI);
  1750. } catch (error) {
  1751. // ignored
  1752. }
  1753. if (testValidURL(resourceURL)) {
  1754. const { content } = await batchRequest.addURL(resourceURL, true);
  1755. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1756. if (forbiddenPrefixFound) {
  1757. return "";
  1758. }
  1759. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1760. } else {
  1761. return "";
  1762. }
  1763. } else {
  1764. return "";
  1765. }
  1766. } else {
  1767. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1768. }
  1769. }));
  1770. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1771. }));
  1772. }
  1773. static replaceImageSource(imgElement, variableName, options) {
  1774. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1775. if (attributeValue) {
  1776. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1777. if (imageData && imageData.replaceable) {
  1778. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1779. const backgroundStyle = {};
  1780. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1781. if (backgroundSize) {
  1782. backgroundStyle["background-size"] = imageData.objectFit;
  1783. }
  1784. if (imageData.objectPosition) {
  1785. backgroundStyle["background-position"] = imageData.objectPosition;
  1786. }
  1787. if (imageData.backgroundColor) {
  1788. backgroundStyle["background-color"] = imageData.backgroundColor;
  1789. }
  1790. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1791. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1792. return true;
  1793. }
  1794. }
  1795. }
  1796. }
  1797. // ----
  1798. // Util
  1799. // ----
  1800. const BLOB_URI_PREFIX = "blob:";
  1801. const HTTP_URI_PREFIX = /^https?:\/\//;
  1802. const FILE_URI_PREFIX = /^file:\/\//;
  1803. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1804. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1805. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1806. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1807. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1808. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1809. 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;
  1810. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1811. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1812. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1813. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1814. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1815. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1816. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1817. function getUpdatedResourceContent(resourceURL, content, options) {
  1818. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1819. options.updatedResources[resourceURL].retrieved = true;
  1820. return options.updatedResources[resourceURL].content;
  1821. } else {
  1822. return content.data || "";
  1823. }
  1824. }
  1825. function normalizeURL(url) {
  1826. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1827. return url;
  1828. } else {
  1829. return url.split("#")[0];
  1830. }
  1831. }
  1832. function getCSSValue(value) {
  1833. let result = "";
  1834. try {
  1835. result = cssTree.generate(value);
  1836. } catch (error) {
  1837. // ignored
  1838. }
  1839. return result;
  1840. }
  1841. function getOnEventAttributeNames(doc) {
  1842. const element = doc.createElement("div");
  1843. const attributeNames = [];
  1844. for (const propertyName in element) {
  1845. if (propertyName.startsWith("on")) {
  1846. attributeNames.push(propertyName);
  1847. }
  1848. }
  1849. attributeNames.push("onunload");
  1850. return attributeNames;
  1851. }
  1852. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  1853. let maxLength;
  1854. if (template) {
  1855. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  1856. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  1857. if (template.match(replaceRegExp)) {
  1858. const matchedLength = template.match(replaceRegExp)[0];
  1859. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  1860. if (isNaN(maxLength) || maxLength <= 0) {
  1861. maxLength = null;
  1862. }
  1863. } else {
  1864. replaceRegExp = new RegExp(regExpVariable, "g");
  1865. }
  1866. if (template.match(replaceRegExp)) {
  1867. let value = await valueGetter();
  1868. if (!dontReplaceSlash) {
  1869. value = value.replace(/\/+/g, replacementCharacter);
  1870. }
  1871. if (maxLength) {
  1872. value = await util.truncateText(value, maxLength);
  1873. }
  1874. return template.replace(replaceRegExp, value);
  1875. }
  1876. }
  1877. return template;
  1878. }
  1879. function getLastSegment(url, replacementCharacter) {
  1880. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1881. if (!lastSegment) {
  1882. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1883. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1884. }
  1885. if (!lastSegment) {
  1886. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1887. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1888. }
  1889. if (!lastSegment) {
  1890. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  1891. }
  1892. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1893. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1894. lastSegment = lastSegmentMatch[1];
  1895. }
  1896. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1897. return lastSegment;
  1898. }
  1899. function getRegExp(string) {
  1900. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1901. }
  1902. function getUrlFunctions(stylesheetContent, unique) {
  1903. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  1904. if (unique) {
  1905. return [...new Set(result)];
  1906. } else {
  1907. return result;
  1908. }
  1909. }
  1910. function getImportFunctions(stylesheetContent) {
  1911. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1912. }
  1913. function findShortcutIcon(shortcutIcons) {
  1914. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1915. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1916. return shortcutIcons[0];
  1917. }
  1918. function matchURL(stylesheetContent) {
  1919. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1920. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1921. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1922. return match && match[1];
  1923. }
  1924. function testIgnoredPath(resourceURL) {
  1925. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1926. }
  1927. function testValidPath(resourceURL) {
  1928. return resourceURL && !resourceURL.match(EMPTY_URL);
  1929. }
  1930. function testValidURL(resourceURL) {
  1931. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1932. }
  1933. function matchImport(stylesheetContent) {
  1934. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1935. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1936. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1937. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1938. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1939. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1940. if (match) {
  1941. const [, resourceURL, media] = match;
  1942. return { resourceURL, media };
  1943. }
  1944. }
  1945. function removeCssComments(stylesheetContent) {
  1946. try {
  1947. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  1948. } catch (error) {
  1949. let start, end;
  1950. do {
  1951. start = stylesheetContent.indexOf("/*");
  1952. end = stylesheetContent.indexOf("*/", start + 2);
  1953. if (start != -1 && end != -1) {
  1954. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1955. }
  1956. } while (start != -1 && end != -1);
  1957. return stylesheetContent;
  1958. }
  1959. }
  1960. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  1961. if (mediaQuery) {
  1962. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1963. } else {
  1964. return stylesheetContent;
  1965. }
  1966. }
  1967. function log(...args) {
  1968. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1969. }
  1970. // -----
  1971. // Stats
  1972. // -----
  1973. const STATS_DEFAULT_VALUES = {
  1974. discarded: {
  1975. "HTML bytes": 0,
  1976. "hidden elements": 0,
  1977. "HTML imports": 0,
  1978. scripts: 0,
  1979. objects: 0,
  1980. "audio sources": 0,
  1981. "video sources": 0,
  1982. frames: 0,
  1983. "CSS rules": 0,
  1984. canvas: 0,
  1985. stylesheets: 0,
  1986. resources: 0,
  1987. medias: 0
  1988. },
  1989. processed: {
  1990. "HTML bytes": 0,
  1991. "hidden elements": 0,
  1992. "HTML imports": 0,
  1993. scripts: 0,
  1994. objects: 0,
  1995. "audio sources": 0,
  1996. "video sources": 0,
  1997. frames: 0,
  1998. "CSS rules": 0,
  1999. canvas: 0,
  2000. stylesheets: 0,
  2001. resources: 0,
  2002. medias: 0
  2003. }
  2004. };
  2005. class Stats {
  2006. constructor(options) {
  2007. this.options = options;
  2008. if (options.displayStats) {
  2009. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2010. }
  2011. }
  2012. set(type, subType, value) {
  2013. if (this.options.displayStats) {
  2014. this.data[type][subType] = value;
  2015. }
  2016. }
  2017. add(type, subType, value) {
  2018. if (this.options.displayStats) {
  2019. this.data[type][subType] += value;
  2020. }
  2021. }
  2022. addAll(pageData) {
  2023. if (this.options.displayStats) {
  2024. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2025. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2026. }
  2027. }
  2028. }
  2029. return { getClass };
  2030. })();