single-file-core.js 73 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765
  1. /*
  2. * Copyright 2010-2019 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.SingleFileCore = this.SingleFileCore || (() => {
  24. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  25. const DEBUG = false;
  26. let docUtil, cssTree, sessionId = 0;
  27. function getClass(...args) {
  28. [docUtil, cssTree] = args;
  29. return SingleFileClass;
  30. }
  31. class SingleFileClass {
  32. constructor(options) {
  33. this.options = options;
  34. options.insertFaviconLink = true;
  35. if (options.sessionId === undefined) {
  36. options.sessionId = sessionId;
  37. sessionId++;
  38. }
  39. }
  40. async run() {
  41. this.runner = new Runner(this.options, true);
  42. await this.runner.loadPage();
  43. await this.runner.initialize();
  44. await this.runner.run();
  45. }
  46. async getPageData() {
  47. return this.runner.getPageData();
  48. }
  49. }
  50. SingleFileClass.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  51. // -------------
  52. // ProgressEvent
  53. // -------------
  54. const PAGE_LOADING = "page-loading";
  55. const PAGE_LOADED = "page-loaded";
  56. const RESOURCES_INITIALIZING = "resource-initializing";
  57. const RESOURCES_INITIALIZED = "resources-initialized";
  58. const RESOURCE_LOADED = "resource-loaded";
  59. const PAGE_ENDED = "page-ended";
  60. const STAGE_STARTED = "stage-started";
  61. const STAGE_ENDED = "stage-ended";
  62. const STAGE_TASK_STARTED = "stage-task-started";
  63. const STAGE_TASK_ENDED = "stage-task-ended";
  64. class ProgressEvent {
  65. constructor(type, detail) {
  66. 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 };
  67. }
  68. }
  69. // ------
  70. // Runner
  71. // ------
  72. const RESOLVE_URLS_STAGE = 0;
  73. const REPLACE_DATA_STAGE = 1;
  74. const REPLACE_DOCS_STAGE = 2;
  75. const POST_PROCESS_STAGE = 3;
  76. const STAGES = [{
  77. sequential: [
  78. { action: "preProcessPage" },
  79. { action: "insertShadowRootContents" },
  80. { action: "replaceStyleContents" },
  81. { option: "selected", action: "removeUnselectedElements" },
  82. { option: "removeVideoSrc", action: "insertVideoPosters" },
  83. { option: "removeFrames", action: "removeFrames" },
  84. { option: "removeImports", action: "removeImports" },
  85. { option: "removeScripts", action: "removeScripts" },
  86. { option: "removeVideoSrc", action: "removeVideoSources" },
  87. { option: "removeAudioSrc", action: "removeAudioSources" },
  88. { action: "removeDiscardedResources" },
  89. { action: "resetCharsetMeta" },
  90. { action: "setInputValues" },
  91. { option: "insertFaviconLink", action: "insertFaviconLink" },
  92. { action: "replaceCanvasElements" },
  93. { action: "insertFonts" },
  94. { option: "removeHiddenElements", action: "removeHiddenElements" },
  95. { action: "resolveHrefs" },
  96. { action: "resolveStyleAttributeURLs" }
  97. ],
  98. parallel: [
  99. { action: "resolveStylesheetURLs" },
  100. { option: "!removeFrames", action: "resolveFrameURLs" },
  101. { option: "!removeImports", action: "resolveHtmlImportURLs" }
  102. ]
  103. }, {
  104. sequential: [
  105. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  106. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  107. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  108. ],
  109. parallel: [
  110. { action: "processStylesheets" },
  111. { action: "processStyleAttributes" },
  112. { action: "processPageResources" },
  113. { option: "!removeScripts", action: "processScripts" }
  114. ]
  115. }, {
  116. sequential: [
  117. { option: "removeAlternativeImages", action: "removeAlternativeImages" },
  118. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" }
  119. ],
  120. parallel: [
  121. { option: "!removeFrames", action: "processFrames" },
  122. { option: "!removeImports", action: "processHtmlImports" },
  123. ]
  124. }, {
  125. sequential: [
  126. { action: "replaceStylesheets" },
  127. { action: "replaceStyleAttributes" },
  128. { action: "insertVariables" },
  129. { option: "compressHTML", action: "compressHTML" }
  130. ]
  131. }];
  132. class Runner {
  133. constructor(options, root) {
  134. this.root = root;
  135. this.options = options;
  136. this.options.url = this.options.url || (this.options.doc && this.options.doc.location.href);
  137. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  138. this.batchRequest = new BatchRequest();
  139. this.processor = new Processor(options, this.batchRequest);
  140. if (this.options.doc) {
  141. const docData = docUtil.preProcessDoc(this.options.doc, this.options.win, this.options);
  142. this.options.canvasData = docData.canvasData;
  143. this.options.fontsData = docData.fontsData;
  144. this.options.stylesheetContents = docData.stylesheetContents;
  145. this.options.imageData = docData.imageData;
  146. this.options.postersData = docData.postersData;
  147. this.options.usedFonts = docData.usedFonts;
  148. this.options.shadowRootContents = docData.shadowRootContents;
  149. this.options.referrer = docData.referrer;
  150. }
  151. if (this.options.saveRawPage) {
  152. this.options.removeFrames = true;
  153. }
  154. this.options.content = this.options.content || (this.options.doc ? docUtil.serialize(this.options.doc, false) : null);
  155. this.onprogress = options.onprogress || (() => { });
  156. }
  157. async loadPage() {
  158. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  159. await this.processor.loadPage(this.options.content);
  160. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  161. }
  162. async initialize() {
  163. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  164. await this.executeStage(RESOLVE_URLS_STAGE);
  165. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  166. if (this.options.doc) {
  167. docUtil.postProcessDoc(this.options.doc, this.options);
  168. }
  169. this.options.doc = null;
  170. this.options.win = null;
  171. }
  172. async run() {
  173. if (this.root) {
  174. this.processor.initialize(this.batchRequest);
  175. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  176. }
  177. await this.batchRequest.run(detail => {
  178. detail.pageURL = this.options.url;
  179. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  180. }, this.options);
  181. await this.pendingPromises;
  182. await this.executeStage(REPLACE_DOCS_STAGE);
  183. await this.executeStage(POST_PROCESS_STAGE);
  184. await this.processor.finalize();
  185. }
  186. async getPageData() {
  187. if (this.root) {
  188. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  189. }
  190. return this.processor.getPageData();
  191. }
  192. async executeStage(step) {
  193. if (DEBUG) {
  194. log("**** STARTED STAGE", step, "****");
  195. }
  196. const frame = !this.root;
  197. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  198. STAGES[step].sequential.forEach(task => {
  199. let startTime;
  200. if (DEBUG) {
  201. startTime = Date.now();
  202. log(" -- STARTED task =", task.action);
  203. }
  204. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  205. this.executeTask(task);
  206. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  207. if (DEBUG) {
  208. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  209. }
  210. });
  211. let parallelTasksPromise;
  212. if (STAGES[step].parallel) {
  213. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  214. let startTime;
  215. if (DEBUG) {
  216. startTime = Date.now();
  217. log(" // STARTED task =", task.action);
  218. }
  219. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  220. await this.executeTask(task);
  221. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  222. if (DEBUG) {
  223. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  224. }
  225. }));
  226. } else {
  227. parallelTasksPromise = Promise.resolve();
  228. }
  229. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  230. if (DEBUG) {
  231. log("**** ENDED STAGE", step, "****");
  232. }
  233. return parallelTasksPromise;
  234. }
  235. executeTask(task) {
  236. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  237. return this.processor[task.action]();
  238. }
  239. }
  240. }
  241. // ------------
  242. // BatchRequest
  243. // ------------
  244. class BatchRequest {
  245. constructor() {
  246. this.requests = new Map();
  247. this.duplicates = new Map();
  248. }
  249. async addURL(resourceURL, asDataURI, groupDuplicates) {
  250. return new Promise((resolve, reject) => {
  251. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  252. let resourceRequests = this.requests.get(requestKey);
  253. if (!resourceRequests) {
  254. resourceRequests = [];
  255. this.requests.set(requestKey, resourceRequests);
  256. }
  257. const callbacks = { resolve, reject };
  258. resourceRequests.push(callbacks);
  259. if (groupDuplicates) {
  260. let duplicateRequests = this.duplicates.get(requestKey);
  261. if (!duplicateRequests) {
  262. duplicateRequests = [];
  263. this.duplicates.set(requestKey, duplicateRequests);
  264. }
  265. duplicateRequests.push(callbacks);
  266. }
  267. });
  268. }
  269. getMaxResources() {
  270. return this.requests.size;
  271. }
  272. async run(onloadListener, options) {
  273. const resourceURLs = Array.from(this.requests.keys());
  274. let indexResource = 0;
  275. return Promise.all(resourceURLs.map(async requestKey => {
  276. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  277. const resourceRequests = this.requests.get(requestKey);
  278. this.requests.delete(requestKey);
  279. try {
  280. const content = await docUtil.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  281. indexResource = indexResource + 1;
  282. onloadListener({ index: indexResource, url: resourceURL });
  283. resourceRequests.forEach(callbacks => {
  284. const duplicateCallbacks = this.duplicates.get(requestKey);
  285. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  286. callbacks.resolve({ content: content.data, indexResource, duplicate });
  287. });
  288. } catch (error) {
  289. indexResource = indexResource + 1;
  290. onloadListener({ index: indexResource, url: resourceURL });
  291. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  292. }
  293. }));
  294. }
  295. }
  296. // ---------
  297. // Processor
  298. // ---------
  299. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  300. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  301. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  302. const SCRIPT_TAG_FOUND = /<script/gi;
  303. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  304. class Processor {
  305. constructor(options, batchRequest) {
  306. this.options = options;
  307. this.stats = new Stats(options);
  308. this.baseURI = Util.normalizeURL(options.baseURI || options.url);
  309. this.batchRequest = batchRequest;
  310. this.stylesheets = new Map();
  311. this.styles = new Map();
  312. this.cssVariables = new Map();
  313. }
  314. initialize() {
  315. this.options.saveDate = new Date();
  316. this.options.saveUrl = this.options.url;
  317. if (this.options.enableMaff) {
  318. this.maffMetaDataPromise = this.batchRequest.addURL(docUtil.resolveURL("index.rdf", this.options.baseURI || this.options.url), false);
  319. }
  320. this.maxResources = this.batchRequest.getMaxResources();
  321. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.framesData) {
  322. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  323. }
  324. this.stats.set("processed", "resources", this.maxResources);
  325. }
  326. async loadPage(pageContent, charset) {
  327. let content;
  328. if (!pageContent || this.options.saveRawPage) {
  329. content = await docUtil.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, charset });
  330. pageContent = content.data;
  331. }
  332. this.doc = docUtil.parseDocContent(pageContent, this.baseURI);
  333. if (this.options.saveRawPage) {
  334. let charset;
  335. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  336. const charsetDeclaration = element.content.split(";")[1];
  337. if (charsetDeclaration && !charset) {
  338. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  339. }
  340. });
  341. if (charset && charset != content.charset) {
  342. return this.loadPage(pageContent, charset);
  343. }
  344. }
  345. this.workStyleElement = this.doc.createElement("style");
  346. this.doc.body.appendChild(this.workStyleElement);
  347. this.onEventAttributeNames = Util.getOnEventAttributeNames(this.doc);
  348. }
  349. async finalize() {
  350. if (this.workStyleElement.parentNode) {
  351. this.workStyleElement.remove();
  352. }
  353. const metaCharset = this.doc.head.querySelector("meta[charset]");
  354. if (metaCharset) {
  355. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  356. }
  357. this.doc.querySelectorAll("base").forEach(element => element.remove());
  358. if (this.doc.head.querySelectorAll("*").length == 1 && metaCharset && this.doc.body.childNodes.length == 0) {
  359. this.doc.head.querySelector("meta[charset]").remove();
  360. }
  361. if (this.maffMetaDataPromise) {
  362. const maffMetaData = await this.maffMetaDataPromise;
  363. if (maffMetaData && maffMetaData.content) {
  364. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  365. const maffDoc = docUtil.parseXMLContent(maffMetaData.content);
  366. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  367. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  368. if (originalURLElement) {
  369. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  370. }
  371. if (archiveTimeElement) {
  372. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  373. if (value) {
  374. const date = new Date(value);
  375. if (!isNaN(date.getTime())) {
  376. this.options.saveDate = new Date(value);
  377. }
  378. }
  379. }
  380. }
  381. }
  382. const titleElement = this.doc.querySelector("title");
  383. const descriptionElement = this.doc.querySelector("meta[name=description]");
  384. const authorElement = this.doc.querySelector("meta[name=author]");
  385. const creatorElement = this.doc.querySelector("meta[name=creator]");
  386. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  387. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  388. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  389. this.options.info = {
  390. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  391. lang: this.doc.documentElement.lang,
  392. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  393. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  394. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : ""
  395. };
  396. }
  397. async getPageData() {
  398. docUtil.postProcessDoc(this.doc, this.options);
  399. const url = docUtil.parseURL(this.baseURI);
  400. if (this.options.insertSingleFileComment) {
  401. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  402. const commentNode = this.doc.createComment("\n Page saved with SingleFile" +
  403. " \n url: " + this.options.saveUrl +
  404. " \n saved date: " + this.options.saveDate +
  405. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  406. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  407. }
  408. let size;
  409. if (this.options.displayStats) {
  410. size = docUtil.getContentSize(this.doc.documentElement.outerHTML);
  411. }
  412. const content = docUtil.serialize(this.doc, this.options.compressHTML);
  413. if (this.options.displayStats) {
  414. const contentSize = docUtil.getContentSize(content);
  415. this.stats.set("processed", "HTML bytes", contentSize);
  416. this.stats.add("discarded", "HTML bytes", size - contentSize);
  417. }
  418. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  419. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  420. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  421. if (!this.options.backgroundSave) {
  422. filename = filename.replace(/\//g, "_");
  423. }
  424. if (docUtil.getContentSize(filename) > 192) {
  425. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  426. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  427. filename = await docUtil.truncateContent(filename, 192 - extension.length);
  428. filename = filename + "…" + extension;
  429. }
  430. if (!filename) {
  431. filename = "Unnamed page";
  432. }
  433. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  434. return {
  435. stats: this.stats.data,
  436. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  437. filename,
  438. content
  439. };
  440. }
  441. preProcessPage() {
  442. if (this.options.win) {
  443. 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));
  444. }
  445. if (this.options.imageData && !this.options.saveRawPage) {
  446. const dataAttributeName = docUtil.IMAGE_ATTRIBUTE_NAME;
  447. this.doc.querySelectorAll("img").forEach(imgElement => {
  448. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  449. if (this.options.removeHiddenElements && imgData.size && !imgData.size.pxWidth && !imgData.size.pxHeight) {
  450. imgElement.setAttribute("src", EMPTY_IMAGE);
  451. } else if (imgData.currentSrc) {
  452. imgElement.setAttribute("src", imgData.currentSrc);
  453. }
  454. });
  455. if (this.options.loadDeferredImages) {
  456. this.doc.querySelectorAll("img[data-src]").forEach(imgElement => {
  457. const imgData = this.options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  458. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  459. imgData.src = imgElement.dataset.src;
  460. imgElement.setAttribute("src", imgElement.dataset.src);
  461. imgElement.removeAttribute("data-src");
  462. }
  463. });
  464. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  465. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  466. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  467. imgElement.removeAttribute("data-srcset");
  468. }
  469. });
  470. }
  471. }
  472. }
  473. insertShadowRootContents() {
  474. if (this.options.shadowRootContents) {
  475. this.doc.querySelectorAll("[" + docUtil.SHADOW_ROOT_ATTRIBUTE_NAME + "]").forEach((element, elementIndex) => {
  476. const elementInfo = this.options.shadowRootContents[elementIndex];
  477. if (elementInfo) {
  478. const frameElement = this.doc.createElement("iframe");
  479. frameElement.setAttribute("style", "all:initial!important;border:0!important;width:100%!important;height:" + elementInfo.height + "px!important");
  480. const windowId = "shadow-" + this.options.framesData.length;
  481. frameElement.setAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME, windowId);
  482. this.options.framesData.push({ windowId, content: elementInfo.content, baseURI: this.baseURI });
  483. element.appendChild(frameElement);
  484. }
  485. });
  486. }
  487. }
  488. replaceStyleContents() {
  489. if (this.options.stylesheetContents) {
  490. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  491. if (this.options.stylesheetContents[styleIndex]) {
  492. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  493. }
  494. });
  495. }
  496. }
  497. removeUnselectedElements() {
  498. removeUnmarkedElements(this.doc.body);
  499. this.doc.body.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  500. function removeUnmarkedElements(element) {
  501. let selectedElementFound = false;
  502. Array.from(element.childNodes).forEach(node => {
  503. if (node.nodeType == 1) {
  504. const isSelectedElement = node.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  505. selectedElementFound = selectedElementFound || isSelectedElement;
  506. if (isSelectedElement) {
  507. node.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  508. removeUnmarkedElements(node);
  509. } else if (selectedElementFound) {
  510. removeNode(node);
  511. } else {
  512. node.style.setProperty("display", "none", "important");
  513. Array.from(node.childNodes).forEach(removeNode);
  514. }
  515. }
  516. });
  517. }
  518. function removeNode(node) {
  519. if (node.nodeType == 1) {
  520. const tagName = node.tagName && node.tagName.toLowerCase();
  521. if (tagName != "svg" && tagName != "style" && tagName != "link") {
  522. node.remove();
  523. }
  524. }
  525. }
  526. }
  527. insertVideoPosters() {
  528. if (this.options.postersData) {
  529. this.doc.querySelectorAll("video[src], video > source[src]").forEach((element, videoIndex) => {
  530. let videoElement;
  531. if (element.tagName == "VIDEO") {
  532. videoElement = element;
  533. } else {
  534. videoElement = element.parentElement;
  535. }
  536. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  537. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  538. }
  539. });
  540. }
  541. }
  542. removeFrames() {
  543. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  544. this.stats.set("discarded", "frames", frameElements.length);
  545. this.stats.set("processed", "frames", frameElements.length);
  546. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  547. }
  548. removeImports() {
  549. const importElements = this.doc.querySelectorAll("link[rel=import]");
  550. this.stats.set("discarded", "HTML imports", importElements.length);
  551. this.stats.set("processed", "HTML imports", importElements.length);
  552. importElements.forEach(element => element.remove());
  553. }
  554. removeScripts() {
  555. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  556. this.doc.querySelectorAll("[href]").forEach(element => {
  557. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  558. element.setAttribute("href", "");
  559. }
  560. });
  561. this.doc.querySelectorAll("[src]").forEach(element => {
  562. if (element.src && element.src.match(/^\s*javascript:/)) {
  563. element.removeAttribute("src");
  564. }
  565. });
  566. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  567. this.stats.set("discarded", "scripts", scriptElements.length);
  568. this.stats.set("processed", "scripts", scriptElements.length);
  569. scriptElements.forEach(element => element.remove());
  570. }
  571. removeVideoSources() {
  572. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  573. this.stats.set("discarded", "video sources", videoSourceElements.length);
  574. this.stats.set("processed", "video sources", videoSourceElements.length);
  575. videoSourceElements.forEach(element => {
  576. if (element.tagName == "SOURCE") {
  577. element.remove();
  578. } else {
  579. videoSourceElements.forEach(element => element.removeAttribute("src"));
  580. }
  581. });
  582. }
  583. removeAudioSources() {
  584. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  585. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  586. this.stats.set("processed", "audio sources", audioSourceElements.length);
  587. audioSourceElements.forEach(element => {
  588. if (element.tagName == "SOURCE") {
  589. element.remove();
  590. } else {
  591. audioSourceElements.forEach(element => element.removeAttribute("src"));
  592. }
  593. });
  594. }
  595. removeDiscardedResources() {
  596. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask, singlefile-logs-window").forEach(element => element.remove());
  597. const objectElements = this.doc.querySelectorAll("applet, meta[http-equiv=refresh], meta[disabled-http-equiv], object[data]:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed[src]:not([src*=\".svg\"]):not([src*=\".pdf\"])");
  598. this.stats.set("discarded", "objects", objectElements.length);
  599. this.stats.set("processed", "objects", objectElements.length);
  600. objectElements.forEach(element => element.remove());
  601. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  602. replacedAttributeValue.forEach(element => {
  603. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  604. if (relValue.length) {
  605. element.setAttribute("rel", relValue);
  606. } else {
  607. element.remove();
  608. }
  609. });
  610. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title]").forEach(element => element.remove());
  611. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  612. if (this.options.compressHTML) {
  613. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  614. }
  615. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  616. }
  617. resetCharsetMeta() {
  618. let charset;
  619. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  620. const charsetDeclaration = element.content.split(";")[1];
  621. if (charsetDeclaration && !charset) {
  622. charset = charsetDeclaration.split("=")[1];
  623. if (charset) {
  624. this.charset = charset.trim().toLowerCase();
  625. }
  626. }
  627. element.remove();
  628. });
  629. const metaElement = this.doc.createElement("meta");
  630. metaElement.setAttribute("charset", "utf-8");
  631. if (this.doc.head.firstChild) {
  632. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  633. } else {
  634. this.doc.head.appendChild(metaElement);
  635. }
  636. }
  637. setInputValues() {
  638. this.doc.querySelectorAll("input").forEach(input => {
  639. const value = input.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  640. input.setAttribute("value", value || "");
  641. });
  642. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  643. const value = input.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  644. if (value == "true") {
  645. input.setAttribute("checked", "");
  646. }
  647. });
  648. this.doc.querySelectorAll("textarea").forEach(textarea => {
  649. const value = textarea.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME);
  650. textarea.textContent = value || "";
  651. });
  652. this.doc.querySelectorAll("select").forEach(select => {
  653. select.querySelectorAll("option").forEach(option => {
  654. const selected = option.getAttribute(docUtil.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  655. if (selected) {
  656. option.setAttribute("selected", "");
  657. }
  658. });
  659. });
  660. }
  661. insertFaviconLink() {
  662. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  663. if (!faviconElement) {
  664. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  665. }
  666. if (!faviconElement) {
  667. faviconElement = this.doc.createElement("link");
  668. faviconElement.setAttribute("type", "image/x-icon");
  669. faviconElement.setAttribute("rel", "shortcut icon");
  670. faviconElement.setAttribute("href", "/favicon.ico");
  671. }
  672. this.doc.head.appendChild(faviconElement);
  673. }
  674. replaceCanvasElements() {
  675. if (this.options.canvasData) {
  676. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  677. const canvasData = this.options.canvasData[indexCanvasElement];
  678. if (canvasData) {
  679. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  680. this.stats.add("processed", "canvas", 1);
  681. }
  682. });
  683. }
  684. }
  685. insertFonts() {
  686. if (this.options.fontsData && this.options.fontsData.length) {
  687. let stylesheetContent = "";
  688. this.options.fontsData.forEach(fontStyles => {
  689. if (fontStyles["font-family"] && fontStyles.src) {
  690. stylesheetContent += "@font-face{";
  691. let stylesContent = "";
  692. Object.keys(fontStyles).forEach(fontStyle => {
  693. if (stylesContent) {
  694. stylesContent += ";";
  695. }
  696. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  697. });
  698. stylesheetContent += stylesContent + "}";
  699. }
  700. });
  701. if (stylesheetContent) {
  702. const styleElement = this.doc.createElement("style");
  703. styleElement.textContent = stylesheetContent;
  704. const existingStyleElement = this.doc.querySelector("style");
  705. if (existingStyleElement) {
  706. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  707. } else {
  708. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  709. }
  710. }
  711. }
  712. }
  713. removeHiddenElements() {
  714. const hiddenElements = this.doc.querySelectorAll("[" + docUtil.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  715. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  716. this.stats.set("processed", "hidden elements", hiddenElements.length);
  717. hiddenElements.forEach(element => element.remove());
  718. }
  719. resolveHrefs() {
  720. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  721. const href = element.getAttribute("href").trim();
  722. if (!Util.testIgnoredPath(href) && !href.startsWith("#")) {
  723. try {
  724. element.setAttribute("href", docUtil.resolveURL(href, this.options.baseURI || this.options.url));
  725. } catch (error) {
  726. // ignored
  727. }
  728. }
  729. });
  730. }
  731. resolveStyleAttributeURLs() {
  732. this.doc.querySelectorAll("[style]").forEach(element => {
  733. let styleContent = element.getAttribute("style");
  734. if (this.options.compressCSS) {
  735. styleContent = docUtil.compressCSS(styleContent);
  736. }
  737. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI);
  738. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  739. this.styles.set(element, declarationList);
  740. });
  741. }
  742. async resolveStylesheetURLs() {
  743. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]"))
  744. .map(async element => {
  745. let mediaText;
  746. if (element.media) {
  747. mediaText = element.media.toLowerCase();
  748. }
  749. this.stylesheets.set(element, { mediaText });
  750. const options = { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled, url: this.options.url, charset: this.charset, compressCSS: this.options.compressCSS };
  751. let stylesheetContent;
  752. if (element.tagName.toLowerCase() == "link") {
  753. if (element.charset) {
  754. options.charset = element.charset;
  755. }
  756. stylesheetContent = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, this.baseURI, options, this.workStyleElement);
  757. } else {
  758. stylesheetContent = await ProcessorHelper.resolveImportURLs(element.textContent, this.baseURI, options, this.workStyleElement);
  759. }
  760. let stylesheet;
  761. try {
  762. stylesheet = cssTree.parse(stylesheetContent);
  763. } catch (error) {
  764. // ignored
  765. }
  766. if (stylesheet && stylesheet.children) {
  767. if (this.options.compressCSS) {
  768. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  769. }
  770. this.stylesheets.get(element).stylesheet = stylesheet;
  771. } else {
  772. this.stylesheets.delete(element);
  773. }
  774. }));
  775. }
  776. async resolveFrameURLs() {
  777. if (!this.options.saveRawPage && this.options.framesData) {
  778. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  779. await Promise.all(frameElements.map(async frameElement => {
  780. if (frameElement.tagName == "OBJECT") {
  781. frameElement.setAttribute("data", "data:text/html,");
  782. } else {
  783. frameElement.removeAttribute("src");
  784. frameElement.removeAttribute("srcdoc");
  785. }
  786. const frameWindowId = frameElement.getAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  787. if (frameWindowId) {
  788. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  789. if (frameData) {
  790. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  791. }
  792. }
  793. }));
  794. }
  795. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  796. options.insertSingleFileComment = false;
  797. options.insertFaviconLink = false;
  798. options.doc = null;
  799. options.win = null;
  800. options.url = frameData.baseURI;
  801. options.windowId = frameWindowId;
  802. if (frameData.content) {
  803. options.content = frameData.content;
  804. options.postersData = frameData.postersData;
  805. options.canvasData = frameData.canvasData;
  806. options.stylesheetContents = frameData.stylesheetContents;
  807. options.fontsData = frameData.fontsData;
  808. options.imageData = frameData.imageData;
  809. options.usedFonts = frameData.usedFonts;
  810. options.shadowRootContents = frameData.shadowRootContents;
  811. frameData.processor = new Runner(options);
  812. frameData.frameElement = frameElement;
  813. await frameData.processor.loadPage();
  814. await frameData.processor.initialize();
  815. frameData.maxResources = batchRequest.getMaxResources();
  816. }
  817. }
  818. }
  819. async resolveHtmlImportURLs() {
  820. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  821. if (!this.relImportProcessors) {
  822. this.relImportProcessors = new Map();
  823. }
  824. await Promise.all(linkElements.map(async linkElement => {
  825. const resourceURL = linkElement.href;
  826. linkElement.removeAttribute("href");
  827. const options = Object.create(this.options);
  828. options.insertSingleFileComment = false;
  829. options.insertFaviconLink = false;
  830. options.doc = null;
  831. options.win = null;
  832. options.url = resourceURL;
  833. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  834. const processor = new Runner(options);
  835. this.relImportProcessors.set(linkElement, processor);
  836. await processor.loadPage();
  837. return processor.initialize();
  838. }
  839. }));
  840. }
  841. removeUnusedStyles() {
  842. if (!this.mediaAllInfo) {
  843. this.mediaAllInfo = docUtil.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  844. }
  845. const stats = docUtil.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  846. this.stats.set("processed", "CSS rules", stats.processed);
  847. this.stats.set("discarded", "CSS rules", stats.discarded);
  848. }
  849. removeUnusedFonts() {
  850. docUtil.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  851. }
  852. removeAlternativeMedias() {
  853. const stats = docUtil.minifyMedias(this.stylesheets);
  854. this.stats.set("processed", "medias", stats.processed);
  855. this.stats.set("discarded", "medias", stats.discarded);
  856. }
  857. async processStylesheets() {
  858. await Promise.all(Array.from(this.stylesheets).map(([, stylesheetInfo]) =>
  859. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  860. ));
  861. }
  862. async processStyleAttributes() {
  863. return Promise.all(Array.from(this.styles).map(([, declarationList]) =>
  864. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  865. ));
  866. }
  867. async processPageResources() {
  868. const processAttributeArgs = [
  869. ["link[href][rel*=\"icon\"]", "href", false, true],
  870. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  871. ["img[src], input[src][type=image]", "src", true],
  872. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  873. ["video[poster]", "poster"],
  874. ["*[background]", "background"],
  875. ["image", "xlink:href"]
  876. ];
  877. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  878. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  879. );
  880. resourcePromises = resourcePromises.concat([
  881. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.options, this.batchRequest),
  882. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  883. ]);
  884. if (!this.options.removeAudioSrc) {
  885. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  886. }
  887. if (!this.options.removeVideoSrc) {
  888. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  889. }
  890. await Promise.all(resourcePromises);
  891. ProcessorHelper.processShortcutIcons(this.doc);
  892. }
  893. async processScripts() {
  894. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  895. let resourceURL;
  896. const scriptSrc = scriptElement.getAttribute("src");
  897. scriptElement.removeAttribute("src");
  898. scriptElement.textContent = "";
  899. try {
  900. resourceURL = docUtil.resolveURL(scriptSrc, this.baseURI);
  901. } catch (error) {
  902. // ignored
  903. }
  904. if (Util.testValidURL(resourceURL)) {
  905. this.stats.add("processed", "scripts", 1);
  906. const content = await docUtil.getContent(resourceURL, { asDataURI: true, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  907. scriptElement.setAttribute("src", content.data);
  908. }
  909. }));
  910. }
  911. removeAlternativeImages() {
  912. docUtil.removeAlternativeImages(this.doc);
  913. }
  914. removeAlternativeFonts() {
  915. docUtil.removeAlternativeFonts(this.doc, this.stylesheets);
  916. }
  917. async processFrames() {
  918. if (this.options.framesData) {
  919. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  920. await Promise.all(frameElements.map(async frameElement => {
  921. const frameWindowId = frameElement.getAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  922. if (frameWindowId) {
  923. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  924. if (frameData) {
  925. this.options.framesData = this.options.framesData.filter(frame => frame.windowId != frameWindowId);
  926. if (frameData.processor) {
  927. this.stats.add("processed", "frames", 1);
  928. await frameData.processor.run();
  929. const pageData = await frameData.processor.getPageData();
  930. frameElement.removeAttribute(docUtil.WIN_ID_ATTRIBUTE_NAME);
  931. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  932. frameElement.setAttribute("sandbox", "allow-popups allow-scripts allow-same-origin");
  933. } else {
  934. frameElement.setAttribute("sandbox", "allow-popups");
  935. }
  936. if (frameElement.tagName == "OBJECT") {
  937. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  938. } else {
  939. if (frameElement.tagName == "FRAME") {
  940. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/#/g, "%23"));
  941. } else {
  942. frameElement.setAttribute("srcdoc", pageData.content);
  943. frameElement.removeAttribute("src");
  944. }
  945. }
  946. this.stats.addAll(pageData);
  947. } else {
  948. this.stats.add("discarded", "frames", 1);
  949. }
  950. }
  951. }
  952. }));
  953. }
  954. }
  955. async processHtmlImports() {
  956. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  957. await Promise.all(linkElements.map(async linkElement => {
  958. const processor = this.relImportProcessors.get(linkElement);
  959. if (processor) {
  960. this.stats.add("processed", "HTML imports", 1);
  961. this.relImportProcessors.delete(linkElement);
  962. const pageData = await processor.getPageData();
  963. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  964. this.stats.addAll(pageData);
  965. } else {
  966. this.stats.add("discarded", "HTML imports", 1);
  967. }
  968. }));
  969. }
  970. replaceStylesheets() {
  971. this.doc.querySelectorAll("style").forEach(styleElement => {
  972. const stylesheetInfo = this.stylesheets.get(styleElement);
  973. if (stylesheetInfo) {
  974. this.stylesheets.delete(styleElement);
  975. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  976. styleElement.textContent = stylesheetContent;
  977. if (stylesheetInfo.mediaText) {
  978. styleElement.media = stylesheetInfo.mediaText;
  979. }
  980. } else {
  981. styleElement.remove();
  982. }
  983. });
  984. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  985. const stylesheetInfo = this.stylesheets.get(linkElement);
  986. if (stylesheetInfo) {
  987. this.stylesheets.delete(linkElement);
  988. const styleElement = this.doc.createElement("style");
  989. if (stylesheetInfo.mediaText) {
  990. styleElement.media = stylesheetInfo.mediaText;
  991. }
  992. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  993. styleElement.textContent = stylesheetContent;
  994. linkElement.parentElement.replaceChild(styleElement, linkElement);
  995. } else {
  996. linkElement.remove();
  997. }
  998. });
  999. }
  1000. replaceStyleAttributes() {
  1001. this.doc.querySelectorAll("[style]").forEach(async element => {
  1002. const declarations = this.styles.get(element);
  1003. if (declarations) {
  1004. this.styles.delete(element);
  1005. let styleContent = cssTree.generate(declarations);
  1006. element.setAttribute("style", styleContent);
  1007. } else {
  1008. element.setAttribute("style", "");
  1009. }
  1010. });
  1011. }
  1012. insertVariables() {
  1013. if (this.cssVariables.size) {
  1014. const styleElement = this.doc.createElement("style");
  1015. const firstStyleElement = this.doc.head.querySelector("style");
  1016. if (firstStyleElement) {
  1017. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1018. } else {
  1019. this.doc.head.appendChild(styleElement);
  1020. }
  1021. let stylesheetContent = "";
  1022. this.cssVariables.forEach((content, indexResource) => {
  1023. this.cssVariables.delete(indexResource);
  1024. if (stylesheetContent) {
  1025. stylesheetContent += ";";
  1026. }
  1027. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}:url("${content}")`;
  1028. });
  1029. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1030. }
  1031. }
  1032. compressHTML() {
  1033. let size;
  1034. if (this.options.displayStats) {
  1035. size = docUtil.getContentSize(this.doc.documentElement.outerHTML);
  1036. }
  1037. docUtil.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: docUtil.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1038. if (this.options.displayStats) {
  1039. this.stats.add("discarded", "HTML bytes", size - docUtil.getContentSize(this.doc.documentElement.outerHTML));
  1040. }
  1041. }
  1042. }
  1043. // ---------------
  1044. // ProcessorHelper
  1045. // ---------------
  1046. const DATA_URI_PREFIX = "data:";
  1047. const ABOUT_BLANK_URI = "about:blank";
  1048. const EMPTY_DATA_URI = "data:base64,";
  1049. const REGEXP_AMP = /&/g;
  1050. const REGEXP_NBSP = /\u00a0/g;
  1051. const REGEXP_START_TAG = /</g;
  1052. const REGEXP_END_TAG = />/g;
  1053. const REGEXP_URL_HASH = /(#.+?)$/;
  1054. const PREFIX_DATA_URI_VND = "data:application/vnd.";
  1055. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1056. class ProcessorHelper {
  1057. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1058. const date = options.saveDate;
  1059. const url = docUtil.parseURL(options.saveUrl);
  1060. template = await Util.evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash);
  1061. template = await Util.evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash);
  1062. template = await Util.evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash);
  1063. template = await Util.evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash);
  1064. template = await Util.evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash);
  1065. template = await Util.evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash);
  1066. template = await Util.evalTemplateVariable(template, "datetime-iso", () => date.toISOString(), dontReplaceSlash);
  1067. template = await Util.evalTemplateVariable(template, "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash);
  1068. template = await Util.evalTemplateVariable(template, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash);
  1069. template = await Util.evalTemplateVariable(template, "date-locale", () => date.toLocaleDateString(), dontReplaceSlash);
  1070. template = await Util.evalTemplateVariable(template, "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash);
  1071. template = await Util.evalTemplateVariable(template, "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash);
  1072. template = await Util.evalTemplateVariable(template, "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1073. template = await Util.evalTemplateVariable(template, "year-locale", () => String(date.getFullYear()), dontReplaceSlash);
  1074. template = await Util.evalTemplateVariable(template, "datetime-locale", () => date.toLocaleString(), dontReplaceSlash);
  1075. template = await Util.evalTemplateVariable(template, "datetime-utc", () => date.toUTCString(), dontReplaceSlash);
  1076. template = await Util.evalTemplateVariable(template, "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash);
  1077. template = await Util.evalTemplateVariable(template, "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash);
  1078. template = await Util.evalTemplateVariable(template, "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash);
  1079. template = await Util.evalTemplateVariable(template, "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash);
  1080. template = await Util.evalTemplateVariable(template, "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash);
  1081. template = await Util.evalTemplateVariable(template, "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash);
  1082. template = await Util.evalTemplateVariable(template, "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash);
  1083. template = await Util.evalTemplateVariable(template, "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash);
  1084. template = await Util.evalTemplateVariable(template, "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash);
  1085. template = await Util.evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash);
  1086. template = await Util.evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash);
  1087. template = await Util.evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash);
  1088. template = await Util.evalTemplateVariable(template, "url-href", () => decodeURI(url.href) || "No href", dontReplaceSlash);
  1089. template = await Util.evalTemplateVariable(template, "url-referrer", () => decodeURI(options.referrer) || "No referrer", dontReplaceSlash);
  1090. template = await Util.evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash);
  1091. template = await Util.evalTemplateVariable(template, "url-pathname", () => decodeURI(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash);
  1092. template = await Util.evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash);
  1093. template = await Util.evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash);
  1094. template = await Util.evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash);
  1095. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1096. for (const [name, value] of params) {
  1097. template = await Util.evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash);
  1098. }
  1099. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1100. template = await Util.evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash);
  1101. template = await Util.evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash);
  1102. template = await Util.evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash);
  1103. template = await Util.evalTemplateVariable(template, "url-last-segment", () => decodeURI(Util.getLastSegment(url)) || "No last segment", dontReplaceSlash);
  1104. if (content) {
  1105. template = await Util.evalTemplateVariable(template, "digest-sha-256", async () => docUtil.digest("SHA-256", content), dontReplaceSlash);
  1106. template = await Util.evalTemplateVariable(template, "digest-sha-384", async () => docUtil.digest("SHA-384", content), dontReplaceSlash);
  1107. template = await Util.evalTemplateVariable(template, "digest-sha-512", async () => docUtil.digest("SHA-512", content), dontReplaceSlash);
  1108. }
  1109. return template.trim();
  1110. }
  1111. static setBackgroundImage(element, url, style) {
  1112. element.style.setProperty("background-blend-mode", "normal", "important");
  1113. element.style.setProperty("background-clip", "content-box", "important");
  1114. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1115. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1116. element.style.setProperty("background-image", url, "important");
  1117. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1118. element.style.setProperty("background-origin", "content-box", "important");
  1119. element.style.setProperty("background-repeat", "no-repeat", "important");
  1120. }
  1121. static processShortcutIcons(doc) {
  1122. let shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1123. if (!shortcutIcon) {
  1124. shortcutIcon = Util.findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1125. if (shortcutIcon) {
  1126. shortcutIcon.rel = "icon";
  1127. }
  1128. }
  1129. if (shortcutIcon) {
  1130. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1131. if (linkElement != shortcutIcon) {
  1132. linkElement.remove();
  1133. }
  1134. });
  1135. }
  1136. }
  1137. static removeSingleLineCssComments(stylesheet) {
  1138. const removedRules = [];
  1139. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1140. const ruleData = cssRule.data;
  1141. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1142. removedRules.push(cssRule);
  1143. }
  1144. }
  1145. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1146. }
  1147. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet) {
  1148. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  1149. const imports = Util.getImportFunctions(stylesheetContent);
  1150. await Promise.all(imports.map(async cssImport => {
  1151. const match = Util.matchImport(cssImport);
  1152. if (match) {
  1153. let resourceURL = Util.normalizeURL(match.resourceURL);
  1154. if (!Util.testIgnoredPath(resourceURL) && Util.testValidPath(resourceURL)) {
  1155. try {
  1156. resourceURL = docUtil.resolveURL(match.resourceURL, baseURI);
  1157. } catch (error) {
  1158. // ignored
  1159. }
  1160. if (Util.testValidURL(resourceURL)) {
  1161. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, validateTextContentType: true };
  1162. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1163. resourceURL = content.resourceURL;
  1164. let importedStylesheetContent = Util.removeCssComments(content.data);
  1165. if (options.compressCSS) {
  1166. importedStylesheetContent = docUtil.compressCSS(importedStylesheetContent);
  1167. }
  1168. importedStylesheetContent = Util.wrapMediaQuery(importedStylesheetContent, match.media);
  1169. if (stylesheetContent.includes(cssImport)) {
  1170. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet);
  1171. workStylesheet.textContent = importedStylesheetContent;
  1172. if (workStylesheet.sheet.cssRules.length) {
  1173. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), importedStylesheetContent);
  1174. } else {
  1175. stylesheetContent = stylesheetContent.replace(Util.getRegExp(cssImport), "");
  1176. }
  1177. }
  1178. }
  1179. }
  1180. }
  1181. }));
  1182. return stylesheetContent;
  1183. }
  1184. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  1185. const urlFunctions = Util.getUrlFunctions(stylesheetContent);
  1186. urlFunctions.map(urlFunction => {
  1187. const originalResourceURL = Util.matchURL(urlFunction);
  1188. const resourceURL = Util.normalizeURL(originalResourceURL);
  1189. if (!Util.testIgnoredPath(resourceURL)) {
  1190. if (!resourceURL || Util.testValidPath(resourceURL)) {
  1191. let resolvedURL;
  1192. try {
  1193. resolvedURL = docUtil.resolveURL(resourceURL, baseURI);
  1194. } catch (error) {
  1195. // ignored
  1196. }
  1197. if (Util.testValidURL(resolvedURL) && resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1198. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  1199. }
  1200. } else {
  1201. let newUrlFunction;
  1202. if (originalResourceURL) {
  1203. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1204. } else {
  1205. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1206. }
  1207. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), newUrlFunction);
  1208. }
  1209. } else {
  1210. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  1211. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  1212. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  1213. stylesheetContent = stylesheetContent.replace(Util.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  1214. }
  1215. }
  1216. }
  1217. });
  1218. return stylesheetContent;
  1219. }
  1220. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1221. resourceURL = Util.normalizeURL(resourceURL);
  1222. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1223. const downloadOptions = { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled, charset: options.charset };
  1224. const content = await docUtil.getContent(resourceURL, downloadOptions);
  1225. resourceURL = content.resourceURL;
  1226. let stylesheetContent = Util.removeCssComments(content.data);
  1227. if (options.compressCSS) {
  1228. stylesheetContent = docUtil.compressCSS(stylesheetContent);
  1229. }
  1230. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1231. return stylesheetContent;
  1232. }
  1233. }
  1234. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1235. const promises = [];
  1236. const removedRules = [];
  1237. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1238. const ruleData = cssRule.data;
  1239. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1240. removedRules.push(cssRule);
  1241. } else if (ruleData.block && ruleData.block.children) {
  1242. if (ruleData.type == "Rule") {
  1243. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1244. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1245. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1246. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1247. promises.push(processFontFaceRule(ruleData));
  1248. }
  1249. }
  1250. }
  1251. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1252. await Promise.all(promises);
  1253. async function processFontFaceRule(ruleData) {
  1254. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1255. if (declaration.type == "Declaration" && declaration.value.children) {
  1256. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1257. await Promise.all(urlFunctions.map(async urlFunction => {
  1258. const originalResourceURL = Util.matchURL(urlFunction);
  1259. const resourceURL = Util.normalizeURL(originalResourceURL);
  1260. if (!Util.testIgnoredPath(resourceURL)) {
  1261. if (Util.testValidURL(resourceURL)) {
  1262. let { content } = await batchRequest.addURL(resourceURL, true);
  1263. if (content != EMPTY_DATA_URI && !content.startsWith(PREFIX_DATA_URI_VND) && !content.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1264. const validResource = await docUtil.validFont(urlFunction);
  1265. if (!validResource) {
  1266. content = EMPTY_DATA_URI;
  1267. }
  1268. }
  1269. replaceURLs(declaration, originalResourceURL, content);
  1270. }
  1271. } else if (resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(PREFIX_DATA_URI_VND) && !resourceURL.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  1272. const validResource = await docUtil.validFont(urlFunction);
  1273. if (!validResource) {
  1274. replaceURLs(declaration, originalResourceURL, EMPTY_DATA_URI);
  1275. }
  1276. }
  1277. }));
  1278. }
  1279. }));
  1280. function replaceURLs(declaration, oldURL, newURL) {
  1281. declaration.value.children.forEach(token => {
  1282. if (token.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.value)) == oldURL) {
  1283. token.value.value = newURL;
  1284. }
  1285. });
  1286. }
  1287. }
  1288. }
  1289. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1290. await Promise.all(declarations.map(async declaration => {
  1291. if (declaration.type == "Declaration" && declaration.value.children) {
  1292. const urlFunctions = Util.getUrlFunctions(Util.getCSSValue(declaration.value));
  1293. await Promise.all(urlFunctions.map(async urlFunction => {
  1294. const originalResourceURL = Util.matchURL(urlFunction);
  1295. const resourceURL = Util.normalizeURL(originalResourceURL);
  1296. if (!Util.testIgnoredPath(resourceURL)) {
  1297. if (Util.testValidURL(resourceURL)) {
  1298. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, true);
  1299. let variableDefined;
  1300. const tokens = [];
  1301. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1302. if (duplicate && options.groupDuplicateImages && rootFunction) {
  1303. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1304. tokens.push({ parent, token, value });
  1305. variableDefined = true;
  1306. } else {
  1307. token.data.value.value = content;
  1308. }
  1309. });
  1310. if (variableDefined) {
  1311. cssVariables.set(indexResource, content);
  1312. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1313. }
  1314. }
  1315. }
  1316. }));
  1317. }
  1318. }));
  1319. function findURLToken(url, children, callback, depth = 0) {
  1320. for (let token = children.head; token; token = token.next) {
  1321. if (token.data.children) {
  1322. findURLToken(url, token.data.children, callback, depth + 1);
  1323. }
  1324. if (token.data.type == "Url" && docUtil.removeQuotes(Util.getCSSValue(token.data.value)) == url) {
  1325. callback(token, children, depth == 0);
  1326. }
  1327. }
  1328. }
  1329. }
  1330. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1331. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1332. let resourceURL = resourceElement.getAttribute(attributeName);
  1333. resourceURL = Util.normalizeURL(resourceURL);
  1334. if (!Util.testIgnoredPath(resourceURL)) {
  1335. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1336. if (Util.testValidPath(resourceURL)) {
  1337. try {
  1338. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1339. } catch (error) {
  1340. // ignored
  1341. }
  1342. if (Util.testValidURL(resourceURL)) {
  1343. const { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL, true, resourceElement.tagName == "IMG" && attributeName == "src");
  1344. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1345. resourceElement.remove();
  1346. } else {
  1347. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1348. if (!forbiddenPrefixFound) {
  1349. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1350. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG) {
  1351. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1352. cssVariables.set(indexResource, content);
  1353. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1354. styles.set(resourceElement, declarationList);
  1355. } else {
  1356. resourceElement.setAttribute(attributeName, content);
  1357. }
  1358. } else {
  1359. resourceElement.setAttribute(attributeName, content);
  1360. }
  1361. }
  1362. }
  1363. }
  1364. }
  1365. }
  1366. }));
  1367. }
  1368. static async processXLinks(resourceElements, baseURI, options, batchRequest) {
  1369. const attributeName = "xlink:href";
  1370. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1371. const originalResourceURL = resourceElement.getAttribute(attributeName);
  1372. let resourceURL = Util.normalizeURL(originalResourceURL);
  1373. if (Util.testValidPath(resourceURL) && !Util.testIgnoredPath(resourceURL)) {
  1374. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1375. try {
  1376. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1377. } catch (error) {
  1378. // ignored
  1379. }
  1380. if (Util.testValidURL(resourceURL)) {
  1381. try {
  1382. const { content } = await batchRequest.addURL(resourceURL);
  1383. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1384. if (hashMatch && hashMatch[0]) {
  1385. const svgDoc = docUtil.parseSVGContent(content);
  1386. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1387. if (symbolElement) {
  1388. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1389. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1390. }
  1391. } else {
  1392. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1393. }
  1394. } catch (error) {
  1395. /* ignored */
  1396. }
  1397. }
  1398. } else if (resourceURL == options.url) {
  1399. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1400. }
  1401. }));
  1402. }
  1403. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1404. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1405. const srcset = docUtil.parseSrcset(resourceElement.getAttribute(attributeName));
  1406. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1407. let resourceURL = Util.normalizeURL(srcsetValue.url);
  1408. if (!Util.testIgnoredPath(resourceURL)) {
  1409. if (Util.testValidPath(resourceURL)) {
  1410. try {
  1411. resourceURL = docUtil.resolveURL(resourceURL, baseURI);
  1412. } catch (error) {
  1413. // ignored
  1414. }
  1415. if (Util.testValidURL(resourceURL)) {
  1416. const { content } = await batchRequest.addURL(resourceURL, true);
  1417. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1418. if (forbiddenPrefixFound) {
  1419. return "";
  1420. }
  1421. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1422. } else {
  1423. return "";
  1424. }
  1425. } else {
  1426. return "";
  1427. }
  1428. } else {
  1429. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1430. }
  1431. }));
  1432. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1433. }));
  1434. }
  1435. static replaceImageSource(imgElement, variableName, options) {
  1436. const dataAttributeName = docUtil.IMAGE_ATTRIBUTE_NAME;
  1437. if (imgElement.getAttribute(dataAttributeName) != null) {
  1438. const imgData = options.imageData[Number(imgElement.getAttribute(dataAttributeName))];
  1439. if (imgData.replaceable) {
  1440. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imgData.size.pxWidth}" height="${imgData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1441. const backgroundStyle = {};
  1442. const backgroundSize = (imgData.objectFit == "content" || imgData.objectFit == "cover") && imgData.objectFit;
  1443. if (backgroundSize) {
  1444. backgroundStyle["background-size"] = imgData.objectFit;
  1445. }
  1446. if (imgData.objectPosition) {
  1447. backgroundStyle["background-position"] = imgData.objectPosition;
  1448. }
  1449. if (imgData.backgroundColor) {
  1450. backgroundStyle["background-color"] = imgData.backgroundColor;
  1451. }
  1452. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1453. imgElement.removeAttribute(dataAttributeName);
  1454. return true;
  1455. }
  1456. }
  1457. }
  1458. }
  1459. // ----
  1460. // Util
  1461. // ----
  1462. const BLOB_URI_PREFIX = "blob:";
  1463. const HTTP_URI_PREFIX = /^https?:\/\//;
  1464. const FILE_URI_PREFIX = /^file:\/\//;
  1465. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1466. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1467. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1468. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1469. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1470. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1471. 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;
  1472. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1473. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1474. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1475. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1476. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1477. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1478. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1479. class Util {
  1480. static normalizeURL(url) {
  1481. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  1482. return url;
  1483. } else {
  1484. return url.split("#")[0];
  1485. }
  1486. }
  1487. static getCSSValue(value) {
  1488. let result = "";
  1489. try {
  1490. result = cssTree.generate(value);
  1491. } catch (error) {
  1492. // ignored
  1493. }
  1494. return result;
  1495. }
  1496. static getOnEventAttributeNames(doc) {
  1497. const element = doc.createElement("div");
  1498. const attributeNames = [];
  1499. for (const propertyName in element) {
  1500. if (propertyName.startsWith("on")) {
  1501. attributeNames.push(propertyName);
  1502. }
  1503. }
  1504. return attributeNames;
  1505. }
  1506. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1507. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1508. if (template && template.match(replaceRegExp)) {
  1509. let value = await valueGetter();
  1510. if (!dontReplaceSlash) {
  1511. value = value.replace(/\/+/g, "_");
  1512. }
  1513. return template.replace(replaceRegExp, value);
  1514. }
  1515. return template;
  1516. }
  1517. static getLastSegment(url) {
  1518. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1519. if (!lastSegment) {
  1520. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1521. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1522. }
  1523. if (!lastSegment) {
  1524. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1525. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1526. }
  1527. if (!lastSegment) {
  1528. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1529. }
  1530. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  1531. if (lastSegmentMatch && lastSegmentMatch[1]) {
  1532. lastSegment = lastSegmentMatch[1];
  1533. }
  1534. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1535. return lastSegment;
  1536. }
  1537. static getRegExp(string) {
  1538. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1539. }
  1540. static getUrlFunctions(stylesheetContent) {
  1541. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1542. }
  1543. static getImportFunctions(stylesheetContent) {
  1544. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1545. }
  1546. static findShortcutIcon(shortcutIcons) {
  1547. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  1548. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  1549. return shortcutIcons[0];
  1550. }
  1551. static matchURL(stylesheetContent) {
  1552. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1553. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1554. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1555. return match && match[1];
  1556. }
  1557. static testIgnoredPath(resourceURL) {
  1558. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  1559. }
  1560. static testValidPath(resourceURL) {
  1561. return resourceURL && !resourceURL.match(EMPTY_URL);
  1562. }
  1563. static testValidURL(resourceURL) {
  1564. return Util.testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  1565. }
  1566. static matchImport(stylesheetContent) {
  1567. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1568. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1569. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1570. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1571. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1572. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1573. if (match) {
  1574. const [, resourceURL, media] = match;
  1575. return { resourceURL, media };
  1576. }
  1577. }
  1578. static removeCssComments(stylesheetContent) {
  1579. let start, end;
  1580. do {
  1581. start = stylesheetContent.indexOf("/*");
  1582. end = stylesheetContent.indexOf("*/", start + 2);
  1583. if (start != -1 && end != -1) {
  1584. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1585. }
  1586. } while (start != -1 && end != -1);
  1587. return stylesheetContent;
  1588. }
  1589. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1590. if (mediaQuery) {
  1591. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1592. } else {
  1593. return stylesheetContent;
  1594. }
  1595. }
  1596. }
  1597. function log(...args) {
  1598. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1599. }
  1600. // -----
  1601. // Stats
  1602. // -----
  1603. const STATS_DEFAULT_VALUES = {
  1604. discarded: {
  1605. "HTML bytes": 0,
  1606. "hidden elements": 0,
  1607. "HTML imports": 0,
  1608. scripts: 0,
  1609. objects: 0,
  1610. "audio sources": 0,
  1611. "video sources": 0,
  1612. frames: 0,
  1613. "CSS rules": 0,
  1614. canvas: 0,
  1615. stylesheets: 0,
  1616. resources: 0,
  1617. medias: 0
  1618. },
  1619. processed: {
  1620. "HTML bytes": 0,
  1621. "hidden elements": 0,
  1622. "HTML imports": 0,
  1623. scripts: 0,
  1624. objects: 0,
  1625. "audio sources": 0,
  1626. "video sources": 0,
  1627. frames: 0,
  1628. "CSS rules": 0,
  1629. canvas: 0,
  1630. stylesheets: 0,
  1631. resources: 0,
  1632. medias: 0
  1633. }
  1634. };
  1635. class Stats {
  1636. constructor(options) {
  1637. this.options = options;
  1638. if (options.displayStats) {
  1639. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1640. }
  1641. }
  1642. set(type, subType, value) {
  1643. if (this.options.displayStats) {
  1644. this.data[type][subType] = value;
  1645. }
  1646. }
  1647. add(type, subType, value) {
  1648. if (this.options.displayStats) {
  1649. this.data[type][subType] += value;
  1650. }
  1651. }
  1652. addAll(pageData) {
  1653. if (this.options.displayStats) {
  1654. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1655. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1656. }
  1657. }
  1658. }
  1659. return { getClass };
  1660. })();