single-file-core.js 96 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364
  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. /* global globalThis */
  24. const DEBUG = false;
  25. const Set = globalThis.Set;
  26. const Map = globalThis.Map;
  27. let util, cssTree;
  28. function getClass(...args) {
  29. [util, cssTree] = args;
  30. return SingleFileClass;
  31. }
  32. class SingleFileClass {
  33. constructor(options) {
  34. this.options = options;
  35. }
  36. async run() {
  37. const waitForUserScript = globalThis._singleFile_waitForUserScript;
  38. if (this.options.userScriptEnabled && waitForUserScript) {
  39. await waitForUserScript(util.ON_BEFORE_CAPTURE_EVENT_NAME);
  40. }
  41. this.runner = new Runner(this.options, true);
  42. await this.runner.loadPage();
  43. await this.runner.initialize();
  44. if (this.options.userScriptEnabled && waitForUserScript) {
  45. await waitForUserScript(util.ON_AFTER_CAPTURE_EVENT_NAME);
  46. }
  47. await this.runner.run();
  48. }
  49. cancel() {
  50. this.cancelled = true;
  51. if (this.runner) {
  52. this.runner.cancel();
  53. }
  54. }
  55. getPageData() {
  56. return this.runner.getPageData();
  57. }
  58. }
  59. // -------------
  60. // ProgressEvent
  61. // -------------
  62. const PAGE_LOADING = "page-loading";
  63. const PAGE_LOADED = "page-loaded";
  64. const RESOURCES_INITIALIZING = "resource-initializing";
  65. const RESOURCES_INITIALIZED = "resources-initialized";
  66. const RESOURCE_LOADED = "resource-loaded";
  67. const PAGE_ENDED = "page-ended";
  68. const STAGE_STARTED = "stage-started";
  69. const STAGE_ENDED = "stage-ended";
  70. const STAGE_TASK_STARTED = "stage-task-started";
  71. const STAGE_TASK_ENDED = "stage-task-ended";
  72. class ProgressEvent {
  73. constructor(type, detail) {
  74. 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 };
  75. }
  76. }
  77. // ------
  78. // Runner
  79. // ------
  80. const RESOLVE_URLS_STAGE = 0;
  81. const REPLACE_DATA_STAGE = 1;
  82. const REPLACE_DOCS_STAGE = 2;
  83. const POST_PROCESS_STAGE = 3;
  84. const STAGES = [{
  85. sequential: [
  86. { action: "preProcessPage" },
  87. { option: "loadDeferredImagesKeepZoomLevel", action: "resetZoomLevel" },
  88. { action: "replaceStyleContents" },
  89. { action: "resetCharsetMeta" },
  90. { option: "saveFavicon", action: "saveFavicon" },
  91. { action: "replaceCanvasElements" },
  92. { action: "insertFonts" },
  93. { action: "insertShadowRootContents" },
  94. { action: "setInputValues" },
  95. { option: "removeScripts", action: "removeScripts" },
  96. { option: "selected", action: "removeUnselectedElements" },
  97. { option: "removeVideoSrc", action: "insertVideoPosters" },
  98. { option: "removeFrames", action: "removeFrames" },
  99. { option: "removeVideoSrc", action: "removeVideoSources" },
  100. { option: "removeAudioSrc", action: "removeAudioSources" },
  101. { action: "removeDiscardedResources" },
  102. { option: "removeHiddenElements", action: "removeHiddenElements" },
  103. { action: "resolveHrefs" },
  104. { action: "resolveStyleAttributeURLs" }
  105. ],
  106. parallel: [
  107. { action: "resolveStylesheetURLs" },
  108. { option: "!removeFrames", action: "resolveFrameURLs" },
  109. { action: "resolveHtmlImportURLs" }
  110. ]
  111. }, {
  112. sequential: [
  113. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  114. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  115. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  116. ],
  117. parallel: [
  118. { action: "processStylesheets" },
  119. { action: "processStyleAttributes" },
  120. { action: "processPageResources" },
  121. { option: "!removeScripts", action: "processScripts" }
  122. ]
  123. }, {
  124. sequential: [
  125. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  126. ],
  127. parallel: [
  128. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  129. { option: "!removeFrames", action: "processFrames" },
  130. { option: "!removeImports", action: "processHtmlImports" },
  131. ]
  132. }, {
  133. sequential: [
  134. { action: "replaceStylesheets" },
  135. { action: "replaceStyleAttributes" },
  136. { action: "insertVariables" },
  137. { option: "compressHTML", action: "compressHTML" },
  138. { action: "cleanupPage" }
  139. ],
  140. parallel: [
  141. { option: "enableMaff", action: "insertMAFFMetaData" },
  142. { action: "setDocInfo" }
  143. ]
  144. }];
  145. class Runner {
  146. constructor(options, root) {
  147. const rootDocDefined = root && options.doc;
  148. this.root = root;
  149. this.options = options;
  150. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  151. const matchResourceReferrer = this.options.url.match(/^.*\//);
  152. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  153. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  154. this.options.rootDocument = root;
  155. this.options.updatedResources = this.options.updatedResources || {};
  156. this.options.fontTests = new Map();
  157. this.batchRequest = new BatchRequest();
  158. this.processor = new Processor(options, this.batchRequest);
  159. if (rootDocDefined) {
  160. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  161. this.options.canvases = docData.canvases;
  162. this.options.fonts = docData.fonts;
  163. this.options.stylesheets = docData.stylesheets;
  164. this.options.images = docData.images;
  165. this.options.posters = docData.posters;
  166. this.options.usedFonts = docData.usedFonts;
  167. this.options.shadowRoots = docData.shadowRoots;
  168. this.options.imports = docData.imports;
  169. this.options.referrer = docData.referrer;
  170. this.markedElements = docData.markedElements;
  171. }
  172. if (this.options.saveRawPage) {
  173. this.options.removeFrames = true;
  174. }
  175. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  176. this.onprogress = options.onprogress || (() => { });
  177. }
  178. async loadPage() {
  179. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  180. await this.processor.loadPage(this.options.content);
  181. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  182. }
  183. async initialize() {
  184. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  185. await this.executeStage(RESOLVE_URLS_STAGE);
  186. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  187. if (this.root && this.options.doc) {
  188. util.postProcessDoc(this.options.doc, this.markedElements);
  189. }
  190. }
  191. cancel() {
  192. this.cancelled = true;
  193. this.batchRequest.cancel();
  194. if (this.root) {
  195. if (this.options.frames) {
  196. this.options.frames.forEach(cancelRunner);
  197. }
  198. if (this.options.imports) {
  199. this.options.imports.forEach(cancelRunner);
  200. }
  201. }
  202. function cancelRunner(resourceData) {
  203. if (resourceData.runner) {
  204. resourceData.runner.cancel();
  205. }
  206. }
  207. }
  208. async run() {
  209. if (this.root) {
  210. this.processor.initialize(this.batchRequest);
  211. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  212. }
  213. await this.batchRequest.run(detail => {
  214. detail.pageURL = this.options.url;
  215. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  216. }, this.options);
  217. await this.pendingPromises;
  218. this.options.doc = null;
  219. this.options.win = null;
  220. await this.executeStage(REPLACE_DOCS_STAGE);
  221. await this.executeStage(POST_PROCESS_STAGE);
  222. this.processor.finalize();
  223. }
  224. getDocument() {
  225. return this.processor.doc;
  226. }
  227. getStyleSheets() {
  228. return this.processor.stylesheets;
  229. }
  230. getPageData() {
  231. if (this.root) {
  232. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  233. }
  234. return this.processor.getPageData();
  235. }
  236. async executeStage(step) {
  237. if (DEBUG) {
  238. log("**** STARTED STAGE", step, "****");
  239. }
  240. const frame = !this.root;
  241. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  242. STAGES[step].sequential.forEach(task => {
  243. let startTime;
  244. if (DEBUG) {
  245. startTime = Date.now();
  246. log(" -- STARTED task =", task.action);
  247. }
  248. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  249. if (!this.cancelled) {
  250. this.executeTask(task);
  251. }
  252. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  253. if (DEBUG) {
  254. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  255. }
  256. });
  257. let parallelTasksPromise;
  258. if (STAGES[step].parallel) {
  259. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  260. let startTime;
  261. if (DEBUG) {
  262. startTime = Date.now();
  263. log(" // STARTED task =", task.action);
  264. }
  265. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  266. if (!this.cancelled) {
  267. await this.executeTask(task);
  268. }
  269. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  270. if (DEBUG) {
  271. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  272. }
  273. }));
  274. } else {
  275. parallelTasksPromise = Promise.resolve();
  276. }
  277. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  278. if (DEBUG) {
  279. log("**** ENDED STAGE", step, "****");
  280. }
  281. return parallelTasksPromise;
  282. }
  283. executeTask(task) {
  284. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  285. return this.processor[task.action]();
  286. }
  287. }
  288. }
  289. // ------------
  290. // BatchRequest
  291. // ------------
  292. class BatchRequest {
  293. constructor() {
  294. this.requests = new Map();
  295. this.duplicates = new Map();
  296. }
  297. addURL(resourceURL, { asBinary, expectedType, groupDuplicates, baseURI, blockMixedContent } = {}) {
  298. return new Promise((resolve, reject) => {
  299. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType, baseURI, blockMixedContent]);
  300. let resourceRequests = this.requests.get(requestKey);
  301. if (!resourceRequests) {
  302. resourceRequests = [];
  303. this.requests.set(requestKey, resourceRequests);
  304. }
  305. const callbacks = { resolve, reject };
  306. resourceRequests.push(callbacks);
  307. if (groupDuplicates) {
  308. let duplicateRequests = this.duplicates.get(requestKey);
  309. if (!duplicateRequests) {
  310. duplicateRequests = [];
  311. this.duplicates.set(requestKey, duplicateRequests);
  312. }
  313. duplicateRequests.push(callbacks);
  314. }
  315. });
  316. }
  317. getMaxResources() {
  318. return this.requests.size;
  319. }
  320. run(onloadListener, options) {
  321. const resourceURLs = [...this.requests.keys()];
  322. let indexResource = 0;
  323. return Promise.all(resourceURLs.map(async requestKey => {
  324. const [resourceURL, asBinary, expectedType, baseURI, blockMixedContent] = JSON.parse(requestKey);
  325. const resourceRequests = this.requests.get(requestKey);
  326. try {
  327. const currentIndexResource = indexResource;
  328. indexResource = indexResource + 1;
  329. const content = await util.getContent(resourceURL, {
  330. asBinary,
  331. expectedType,
  332. maxResourceSize: options.maxResourceSize,
  333. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  334. frameId: options.windowId,
  335. resourceReferrer: options.resourceReferrer,
  336. baseURI,
  337. blockMixedContent,
  338. acceptHeaders: options.acceptHeaders
  339. });
  340. onloadListener({ url: resourceURL });
  341. if (!this.cancelled) {
  342. resourceRequests.forEach(callbacks => {
  343. const duplicateCallbacks = this.duplicates.get(requestKey);
  344. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  345. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  346. });
  347. }
  348. } catch (error) {
  349. indexResource = indexResource + 1;
  350. onloadListener({ url: resourceURL });
  351. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  352. }
  353. this.requests.delete(requestKey);
  354. }));
  355. }
  356. cancel() {
  357. this.cancelled = true;
  358. const resourceURLs = [...this.requests.keys()];
  359. resourceURLs.forEach(requestKey => {
  360. const resourceRequests = this.requests.get(requestKey);
  361. resourceRequests.forEach(callbacks => callbacks.reject());
  362. this.requests.delete(requestKey);
  363. });
  364. }
  365. }
  366. // ---------
  367. // Processor
  368. // ---------
  369. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  370. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  371. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  372. const SCRIPT_TAG_FOUND = /<script/gi;
  373. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  374. const CANVAS_TAG_FOUND = /<canvas/gi;
  375. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  376. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  377. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  378. const UTF8_CHARSET = "utf-8";
  379. class Processor {
  380. constructor(options, batchRequest) {
  381. this.options = options;
  382. this.stats = new Stats(options);
  383. this.baseURI = normalizeURL(options.baseURI || options.url);
  384. this.batchRequest = batchRequest;
  385. this.stylesheets = new Map();
  386. this.styles = new Map();
  387. this.cssVariables = new Map();
  388. this.fontTests = options.fontTests;
  389. }
  390. initialize() {
  391. this.options.saveDate = new Date();
  392. this.options.saveUrl = this.options.url;
  393. if (this.options.enableMaff) {
  394. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), { expectedType: "document" });
  395. }
  396. this.maxResources = this.batchRequest.getMaxResources();
  397. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  398. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  399. }
  400. if (!this.options.removeImports && this.options.imports) {
  401. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  402. }
  403. this.stats.set("processed", "resources", this.maxResources);
  404. }
  405. async loadPage(pageContent, charset) {
  406. let content;
  407. if (!pageContent || this.options.saveRawPage) {
  408. content = await util.getContent(this.baseURI, {
  409. maxResourceSize: this.options.maxResourceSize,
  410. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  411. charset,
  412. frameId: this.options.windowId,
  413. resourceReferrer: this.options.resourceReferrer,
  414. expectedType: "document",
  415. acceptHeaders: this.options.acceptHeaders
  416. });
  417. pageContent = content.data;
  418. }
  419. this.doc = util.parseDocContent(pageContent, this.baseURI);
  420. if (this.options.saveRawPage) {
  421. let charset;
  422. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  423. const charsetDeclaration = element.content.split(";")[1];
  424. if (charsetDeclaration && !charset) {
  425. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  426. }
  427. });
  428. if (charset && content.charset && charset.toLowerCase() != content.charset.toLowerCase()) {
  429. return this.loadPage(pageContent, charset);
  430. }
  431. }
  432. this.workStyleElement = this.doc.createElement("style");
  433. this.doc.body.appendChild(this.workStyleElement);
  434. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  435. }
  436. finalize() {
  437. if (this.workStyleElement.parentNode) {
  438. this.workStyleElement.remove();
  439. }
  440. }
  441. async getPageData() {
  442. util.postProcessDoc(this.doc);
  443. const url = util.parseURL(this.baseURI);
  444. if (this.options.insertSingleFileComment) {
  445. const firstComment = this.doc.documentElement.firstChild;
  446. let infobarURL = this.options.saveUrl, infobarSaveDate = this.options.saveDate;
  447. if (firstComment.nodeType == 8 && (firstComment.textContent.includes(util.COMMENT_HEADER_LEGACY) || firstComment.textContent.includes(util.COMMENT_HEADER))) {
  448. const info = this.doc.documentElement.firstChild.textContent.split("\n");
  449. const [, , url, saveDate] = info;
  450. infobarURL = url.split("url: ")[1];
  451. infobarSaveDate = saveDate.split("saved date: ")[1];
  452. firstComment.remove();
  453. }
  454. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  455. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  456. " \n url: " + infobarURL +
  457. " \n saved date: " + infobarSaveDate +
  458. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  459. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  460. }
  461. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  462. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  463. if (!canonicalLink) {
  464. canonicalLink = this.doc.createElement("link");
  465. canonicalLink.setAttribute("rel", "canonical");
  466. this.doc.head.appendChild(canonicalLink);
  467. }
  468. if (canonicalLink && !canonicalLink.href) {
  469. canonicalLink.href = this.options.saveUrl;
  470. }
  471. }
  472. if (this.options.insertMetaCSP) {
  473. const metaTag = this.doc.createElement("meta");
  474. metaTag.httpEquiv = "content-security-policy";
  475. metaTag.content = "default-src 'none'; font-src 'self' data:; img-src 'self' data:; style-src 'unsafe-inline'; media-src 'self' data:; script-src 'unsafe-inline' data:;";
  476. this.doc.head.appendChild(metaTag);
  477. }
  478. if (this.options.insertMetaNoIndex) {
  479. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  480. if (!metaElement) {
  481. metaElement = this.doc.createElement("meta");
  482. metaElement.setAttribute("name", "robots");
  483. metaElement.setAttribute("content", "noindex");
  484. this.doc.head.appendChild(metaElement);
  485. }
  486. }
  487. let size;
  488. if (this.options.displayStats) {
  489. size = util.getContentSize(this.doc.documentElement.outerHTML);
  490. }
  491. const content = util.serialize(this.doc, this.options.compressHTML);
  492. if (this.options.displayStats) {
  493. const contentSize = util.getContentSize(content);
  494. this.stats.set("processed", "HTML bytes", contentSize);
  495. this.stats.add("discarded", "HTML bytes", size - contentSize);
  496. }
  497. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  498. const replacementCharacter = this.options.filenameReplacementCharacter;
  499. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  500. if (!this.options.backgroundSave) {
  501. filename = filename.replace(/\//g, replacementCharacter);
  502. }
  503. if (!this.options.saveToGDrive && !this.options.saveToGitHub && !this.options.saveWithCompanion && util.getContentSize(filename) > this.options.filenameMaxLength) {
  504. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  505. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  506. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  507. filename = filename + "…" + extension;
  508. }
  509. if (!filename) {
  510. filename = "Unnamed page";
  511. }
  512. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  513. const pageData = {
  514. stats: this.stats.data,
  515. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  516. filename,
  517. content
  518. };
  519. if (this.options.addProof) {
  520. pageData.hash = await util.digest("SHA-256", content);
  521. }
  522. if (this.options.retrieveLinks) {
  523. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  524. }
  525. return pageData;
  526. }
  527. preProcessPage() {
  528. if (this.options.win) {
  529. 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));
  530. }
  531. if (this.options.images && !this.options.saveRawPage) {
  532. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  533. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  534. if (attributeValue) {
  535. const imageData = this.options.images[Number(attributeValue)];
  536. if (imageData) {
  537. if (this.options.removeHiddenElements && (
  538. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  539. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  540. )) {
  541. imgElement.setAttribute("src", EMPTY_IMAGE);
  542. } else {
  543. if (imageData.currentSrc) {
  544. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  545. imgElement.setAttribute("src", imageData.currentSrc);
  546. }
  547. if (this.options.loadDeferredImages) {
  548. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  549. imageData.src = imgElement.dataset.src;
  550. imgElement.setAttribute("src", imgElement.dataset.src);
  551. imgElement.removeAttribute("data-src");
  552. }
  553. }
  554. }
  555. }
  556. }
  557. });
  558. if (this.options.loadDeferredImages) {
  559. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  560. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  561. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  562. imgElement.removeAttribute("data-srcset");
  563. }
  564. });
  565. }
  566. }
  567. }
  568. replaceStyleContents() {
  569. if (this.options.stylesheets) {
  570. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  571. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  572. if (attributeValue) {
  573. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  574. if (stylesheetContent) {
  575. styleElement.textContent = stylesheetContent;
  576. }
  577. }
  578. });
  579. }
  580. }
  581. removeUnselectedElements() {
  582. removeUnmarkedElements(this.doc.body);
  583. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  584. function removeUnmarkedElements(element) {
  585. let selectedElementFound = false;
  586. Array.from(element.childNodes).forEach(node => {
  587. if (node.nodeType == 1) {
  588. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  589. selectedElementFound = selectedElementFound || isSelectedElement;
  590. if (isSelectedElement) {
  591. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  592. removeUnmarkedElements(node);
  593. } else if (selectedElementFound) {
  594. removeNode(node);
  595. } else {
  596. hideNode(node);
  597. }
  598. }
  599. });
  600. }
  601. function removeNode(node) {
  602. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  603. node.remove();
  604. } else {
  605. hideNode(node);
  606. }
  607. }
  608. function hideNode(node) {
  609. if (canHideNode(node)) {
  610. node.style.setProperty("display", "none", "important");
  611. Array.from(node.childNodes).forEach(removeNode);
  612. }
  613. }
  614. function canHideNode(node) {
  615. if (node.nodeType == 1) {
  616. const tagName = node.tagName && node.tagName.toLowerCase();
  617. return (tagName != "svg" && tagName != "style" && tagName != "link");
  618. }
  619. }
  620. }
  621. insertVideoPosters() {
  622. if (this.options.posters) {
  623. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  624. let videoElement;
  625. if (element.tagName == "VIDEO") {
  626. videoElement = element;
  627. } else {
  628. videoElement = element.parentElement;
  629. }
  630. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  631. if (attributeValue) {
  632. const posterURL = this.options.posters[Number(attributeValue)];
  633. if (!videoElement.poster && posterURL) {
  634. videoElement.setAttribute("poster", posterURL);
  635. }
  636. }
  637. });
  638. }
  639. }
  640. removeFrames() {
  641. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  642. this.stats.set("discarded", "frames", frameElements.length);
  643. this.stats.set("processed", "frames", frameElements.length);
  644. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  645. }
  646. removeScripts() {
  647. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  648. this.doc.querySelectorAll("[href]").forEach(element => {
  649. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  650. element.setAttribute("href", "");
  651. }
  652. });
  653. this.doc.querySelectorAll("[src]").forEach(element => {
  654. if (element.src && element.src.match(/^\s*javascript:/)) {
  655. element.removeAttribute("src");
  656. }
  657. });
  658. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  659. this.stats.set("discarded", "scripts", scriptElements.length);
  660. this.stats.set("processed", "scripts", scriptElements.length);
  661. scriptElements.forEach(element => element.remove());
  662. }
  663. removeVideoSources() {
  664. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  665. this.stats.set("discarded", "video sources", videoSourceElements.length);
  666. this.stats.set("processed", "video sources", videoSourceElements.length);
  667. videoSourceElements.forEach(element => {
  668. if (element.tagName == "SOURCE") {
  669. element.remove();
  670. } else {
  671. videoSourceElements.forEach(element => element.removeAttribute("src"));
  672. }
  673. });
  674. }
  675. removeAudioSources() {
  676. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  677. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  678. this.stats.set("processed", "audio sources", audioSourceElements.length);
  679. audioSourceElements.forEach(element => {
  680. if (element.tagName == "SOURCE") {
  681. element.remove();
  682. } else {
  683. audioSourceElements.forEach(element => element.removeAttribute("src"));
  684. }
  685. });
  686. }
  687. removeDiscardedResources() {
  688. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  689. this.doc.querySelectorAll("button button").forEach(element => element.remove());
  690. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  691. 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\"])");
  692. this.stats.set("discarded", "objects", objectElements.length);
  693. this.stats.set("processed", "objects", objectElements.length);
  694. objectElements.forEach(element => element.remove());
  695. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  696. replacedAttributeValue.forEach(element => {
  697. let regExp;
  698. if (this.options.removeScripts) {
  699. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  700. } else {
  701. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  702. }
  703. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  704. if (relValue.length) {
  705. element.setAttribute("rel", relValue);
  706. } else {
  707. element.remove();
  708. }
  709. });
  710. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  711. if (this.options.removeHiddenElements) {
  712. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  713. }
  714. if (!this.options.saveFavicon) {
  715. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  716. }
  717. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  718. }
  719. resetCharsetMeta() {
  720. let charset;
  721. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  722. const charsetDeclaration = element.content.split(";")[1];
  723. if (charsetDeclaration && !charset) {
  724. charset = charsetDeclaration.split("=")[1];
  725. if (charset) {
  726. this.charset = charset.trim().toLowerCase();
  727. }
  728. }
  729. element.remove();
  730. });
  731. const metaElement = this.doc.createElement("meta");
  732. metaElement.setAttribute("charset", UTF8_CHARSET);
  733. if (this.doc.head.firstChild) {
  734. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  735. } else {
  736. this.doc.head.appendChild(metaElement);
  737. }
  738. }
  739. setInputValues() {
  740. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  741. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  742. input.setAttribute("value", value || "");
  743. });
  744. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  745. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  746. if (value == "true") {
  747. input.setAttribute("checked", "");
  748. }
  749. });
  750. this.doc.querySelectorAll("textarea").forEach(textarea => {
  751. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  752. textarea.textContent = value || "";
  753. });
  754. this.doc.querySelectorAll("select").forEach(select => {
  755. select.querySelectorAll("option").forEach(option => {
  756. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  757. if (selected) {
  758. option.setAttribute("selected", "");
  759. }
  760. });
  761. });
  762. }
  763. saveFavicon() {
  764. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  765. if (!faviconElement) {
  766. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  767. }
  768. if (!faviconElement) {
  769. faviconElement = this.doc.createElement("link");
  770. faviconElement.setAttribute("type", "image/x-icon");
  771. faviconElement.setAttribute("rel", "shortcut icon");
  772. faviconElement.setAttribute("href", "/favicon.ico");
  773. }
  774. this.doc.head.appendChild(faviconElement);
  775. }
  776. replaceCanvasElements() {
  777. if (this.options.canvases) {
  778. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  779. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  780. if (attributeValue) {
  781. const canvasData = this.options.canvases[Number(attributeValue)];
  782. if (canvasData) {
  783. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  784. this.stats.add("processed", "canvas", 1);
  785. }
  786. }
  787. });
  788. }
  789. }
  790. insertFonts() {
  791. if (this.options.fonts && this.options.fonts.length) {
  792. let stylesheetContent = "";
  793. this.options.fonts.forEach(fontData => {
  794. if (fontData["font-family"] && fontData.src) {
  795. stylesheetContent += "@font-face{";
  796. let stylesContent = "";
  797. Object.keys(fontData).forEach(fontStyle => {
  798. if (stylesContent) {
  799. stylesContent += ";";
  800. }
  801. stylesContent += fontStyle + ":" + fontData[fontStyle];
  802. });
  803. stylesheetContent += stylesContent + "}";
  804. }
  805. });
  806. if (stylesheetContent) {
  807. const styleElement = this.doc.createElement("style");
  808. styleElement.textContent = stylesheetContent;
  809. const existingStyleElement = this.doc.querySelector("style");
  810. if (existingStyleElement) {
  811. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  812. } else {
  813. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  814. }
  815. }
  816. }
  817. }
  818. removeHiddenElements() {
  819. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  820. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  821. this.stats.set("discarded", "hidden elements", removedElements.length);
  822. this.stats.set("processed", "hidden elements", removedElements.length);
  823. if (hiddenElements.length) {
  824. const styleElement = this.doc.createElement("style");
  825. styleElement.textContent = ".sf-hidden{display:none!important;}";
  826. this.doc.head.appendChild(styleElement);
  827. hiddenElements.forEach(element => {
  828. if (element.style.getPropertyValue("display") != "none") {
  829. if (element.style.getPropertyPriority("display") == "important") {
  830. element.style.setProperty("display", "none", "important");
  831. } else {
  832. element.classList.add("sf-hidden");
  833. }
  834. }
  835. });
  836. }
  837. removedElements.forEach(element => element.remove());
  838. }
  839. resolveHrefs() {
  840. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  841. const href = element.getAttribute("href").trim();
  842. if (element.tagName == "LINK" && element.rel.includes("stylesheet")) {
  843. if (this.options.saveOriginalURLs && !isDataURL(href)) {
  844. element.setAttribute("data-sf-original-href", href);
  845. }
  846. }
  847. if (!testIgnoredPath(href)) {
  848. let resolvedURL;
  849. try {
  850. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  851. } catch (error) {
  852. // ignored
  853. }
  854. if (resolvedURL) {
  855. const url = normalizeURL(this.options.url);
  856. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  857. resolvedURL = resolvedURL.substring(url.length);
  858. }
  859. try {
  860. element.setAttribute("href", resolvedURL);
  861. } catch (error) {
  862. // ignored
  863. }
  864. }
  865. }
  866. });
  867. }
  868. resolveStyleAttributeURLs() {
  869. this.doc.querySelectorAll("[style]").forEach(element => {
  870. let styleContent = element.getAttribute("style");
  871. if (this.options.compressCSS) {
  872. styleContent = util.compressCSS(styleContent);
  873. }
  874. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement, this.options.saveOriginalURLs);
  875. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  876. this.styles.set(element, declarationList);
  877. });
  878. }
  879. async resolveStylesheetURLs() {
  880. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]")).map(async element => {
  881. const options = {
  882. maxResourceSize: this.options.maxResourceSize,
  883. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  884. url: this.options.url,
  885. charset: this.charset,
  886. compressCSS: this.options.compressCSS,
  887. updatedResources: this.options.updatedResources,
  888. rootDocument: this.options.rootDocument,
  889. frameId: this.options.windowId,
  890. resourceReferrer: this.options.resourceReferrer,
  891. blockMixedContent: this.options.blockMixedContent,
  892. saveOriginalURLs: this.options.saveOriginalURLs,
  893. acceptHeaders: this.options.acceptHeaders
  894. };
  895. let mediaText;
  896. if (element.media) {
  897. mediaText = element.media.toLowerCase();
  898. }
  899. const stylesheetInfo = { mediaText };
  900. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  901. stylesheetInfo.scoped = true;
  902. }
  903. if (element.tagName == "LINK" && element.charset) {
  904. options.charset = element.charset;
  905. }
  906. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  907. }));
  908. if (this.options.rootDocument) {
  909. const newResources = Object.keys(this.options.updatedResources).filter(url => this.options.updatedResources[url].type == "stylesheet" && !this.options.updatedResources[url].retrieved).map(url => this.options.updatedResources[url]);
  910. await Promise.all(newResources.map(async resource => {
  911. resource.retrieved = true;
  912. const stylesheetInfo = {};
  913. const element = this.doc.createElement("style");
  914. this.doc.body.appendChild(element);
  915. element.textContent = resource.content;
  916. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, this.options, this.workStyleElement);
  917. }));
  918. }
  919. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  920. stylesheets.set(element, stylesheetInfo);
  921. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  922. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  923. options = Object.assign({}, options, { charset: getCharset(stylesheetContent) });
  924. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  925. }
  926. let stylesheet;
  927. try {
  928. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  929. } catch (error) {
  930. // ignored
  931. }
  932. if (stylesheet && stylesheet.children) {
  933. if (options.compressCSS) {
  934. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  935. }
  936. stylesheetInfo.stylesheet = stylesheet;
  937. } else {
  938. stylesheets.delete(element);
  939. }
  940. }
  941. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  942. let content;
  943. if (element.tagName == "LINK") {
  944. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  945. } else {
  946. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  947. }
  948. return content || "";
  949. }
  950. }
  951. async resolveFrameURLs() {
  952. if (!this.options.saveRawPage) {
  953. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  954. await Promise.all(frameElements.map(async frameElement => {
  955. if (frameElement.tagName == "OBJECT") {
  956. frameElement.setAttribute("data", "data:text/html,");
  957. } else {
  958. const src = frameElement.getAttribute("src");
  959. if (this.options.saveOriginalURLs && !isDataURL(src)) {
  960. frameElement.setAttribute("data-sf-original-src", src);
  961. }
  962. frameElement.removeAttribute("src");
  963. frameElement.removeAttribute("srcdoc");
  964. }
  965. Array.from(frameElement.childNodes).forEach(node => node.remove());
  966. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  967. if (this.options.frames && frameWindowId) {
  968. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  969. if (frameData) {
  970. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  971. }
  972. }
  973. }));
  974. }
  975. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  976. options.insertSingleFileComment = false;
  977. options.insertCanonicalLink = false;
  978. options.insertMetaNoIndex = false;
  979. options.saveFavicon = false;
  980. options.url = frameData.baseURI;
  981. options.windowId = frameWindowId;
  982. if (frameData.content) {
  983. options.content = frameData.content;
  984. options.canvases = frameData.canvases;
  985. options.fonts = frameData.fonts;
  986. options.stylesheets = frameData.stylesheets;
  987. options.images = frameData.images;
  988. options.posters = frameData.posters;
  989. options.usedFonts = frameData.usedFonts;
  990. options.shadowRoots = frameData.shadowRoots;
  991. options.imports = frameData.imports;
  992. frameData.runner = new Runner(options);
  993. frameData.frameElement = frameElement;
  994. await frameData.runner.loadPage();
  995. await frameData.runner.initialize();
  996. frameData.maxResources = batchRequest.getMaxResources();
  997. }
  998. }
  999. }
  1000. insertShadowRootContents() {
  1001. const doc = this.doc;
  1002. const options = this.options;
  1003. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  1004. processElement(this.doc);
  1005. if (this.options.removeScripts) {
  1006. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  1007. }
  1008. const scriptElement = doc.createElement("script");
  1009. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  1010. scriptElement.textContent = `(()=>{document.currentScript.remove();processNode(document);function processNode(node){node.querySelectorAll("template[${SHADOW_MODE_ATTRIBUTE_NAME}]").forEach(element=>{let shadowRoot = element.parentElement.shadowRoot;if (!shadowRoot) {try {shadowRoot=element.parentElement.attachShadow({mode:element.getAttribute("${SHADOW_MODE_ATTRIBUTE_NAME}"),delegatesFocus:Boolean(element.getAttribute("${SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME}"))});shadowRoot.innerHTML=element.innerHTML;element.remove()} catch (error) {} if (shadowRoot) {processNode(shadowRoot)}}})}})()`;
  1011. doc.body.appendChild(scriptElement);
  1012. }
  1013. function processElement(element) {
  1014. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  1015. shadowRootElements.forEach(element => {
  1016. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  1017. if (attributeValue) {
  1018. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  1019. if (shadowRootData) {
  1020. const templateElement = doc.createElement("template");
  1021. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  1022. if (shadowRootData.delegatesFocus) {
  1023. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME, "");
  1024. }
  1025. if (shadowRootData.adoptedStyleSheets) {
  1026. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  1027. const styleElement = doc.createElement("style");
  1028. styleElement.textContent = stylesheetContent;
  1029. templateElement.appendChild(styleElement);
  1030. });
  1031. }
  1032. const shadowDoc = util.parseDocContent(shadowRootData.content);
  1033. if (shadowDoc.head) {
  1034. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  1035. if (metaCharset) {
  1036. metaCharset.remove();
  1037. }
  1038. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1039. }
  1040. if (shadowDoc.body) {
  1041. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1042. }
  1043. processElement(templateElement);
  1044. if (element.firstChild) {
  1045. element.insertBefore(templateElement, element.firstChild);
  1046. } else {
  1047. element.appendChild(templateElement);
  1048. }
  1049. }
  1050. }
  1051. });
  1052. }
  1053. }
  1054. async resolveHtmlImportURLs() {
  1055. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1056. await Promise.all(linkElements.map(async linkElement => {
  1057. const resourceURL = linkElement.href;
  1058. if (this.options.saveOriginalURLs && !isDataURL(resourceURL)) {
  1059. linkElement.setAttribute("data-sf-original-href", resourceURL);
  1060. }
  1061. linkElement.removeAttribute("href");
  1062. const options = Object.create(this.options);
  1063. options.insertSingleFileComment = false;
  1064. options.insertCanonicalLink = false;
  1065. options.insertMetaNoIndex = false;
  1066. options.saveFavicon = false;
  1067. options.removeUnusedStyles = false;
  1068. options.removeAlternativeMedias = false;
  1069. options.removeUnusedFonts = false;
  1070. options.removeHiddenElements = false;
  1071. options.url = resourceURL;
  1072. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1073. if (attributeValue) {
  1074. const importData = options.imports[Number(attributeValue)];
  1075. if (importData) {
  1076. options.content = importData.content;
  1077. importData.runner = new Runner(options);
  1078. await importData.runner.loadPage();
  1079. await importData.runner.initialize();
  1080. if (!options.removeImports) {
  1081. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1082. }
  1083. importData.runner.getStyleSheets().forEach(stylesheet => {
  1084. const importedStyleElement = this.doc.createElement("style");
  1085. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1086. this.stylesheets.set(importedStyleElement, stylesheet);
  1087. });
  1088. }
  1089. }
  1090. if (options.removeImports) {
  1091. linkElement.remove();
  1092. this.stats.add("discarded", "HTML imports", 1);
  1093. }
  1094. }));
  1095. }
  1096. removeUnusedStyles() {
  1097. if (!this.mediaAllInfo) {
  1098. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1099. }
  1100. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1101. this.stats.set("processed", "CSS rules", stats.processed);
  1102. this.stats.set("discarded", "CSS rules", stats.discarded);
  1103. }
  1104. removeUnusedFonts() {
  1105. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1106. }
  1107. removeAlternativeMedias() {
  1108. const stats = util.minifyMedias(this.stylesheets);
  1109. this.stats.set("processed", "medias", stats.processed);
  1110. this.stats.set("discarded", "medias", stats.discarded);
  1111. }
  1112. async processStylesheets() {
  1113. this.options.fontURLs = new Map();
  1114. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1115. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1116. ));
  1117. }
  1118. async processStyleAttributes() {
  1119. return Promise.all([...this.styles].map(([, declarationList]) =>
  1120. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1121. ));
  1122. }
  1123. async processPageResources() {
  1124. const processAttributeArgs = [
  1125. ["link[href][rel*=\"icon\"]", "href", false, true],
  1126. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1127. ["img[src], input[src][type=image]", "src", true],
  1128. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1129. ["video[poster]", "poster"],
  1130. ["*[background]", "background"],
  1131. ["image", "xlink:href"],
  1132. ["image", "href"]
  1133. ];
  1134. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1135. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1136. );
  1137. resourcePromises = resourcePromises.concat([
  1138. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.doc, this.baseURI, this.options, this.batchRequest),
  1139. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1140. ]);
  1141. if (!this.options.removeAudioSrc) {
  1142. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1143. }
  1144. if (!this.options.removeVideoSrc) {
  1145. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1146. }
  1147. await Promise.all(resourcePromises);
  1148. if (this.options.saveFavicon) {
  1149. ProcessorHelper.processShortcutIcons(this.doc);
  1150. }
  1151. }
  1152. async processScripts() {
  1153. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1154. let resourceURL;
  1155. let scriptSrc;
  1156. if (element.tagName == "SCRIPT") {
  1157. scriptSrc = element.getAttribute("src");
  1158. if (this.options.saveOriginalURLs && !isDataURL(scriptSrc)) {
  1159. element.setAttribute("data-sf-original-src", scriptSrc);
  1160. }
  1161. } else {
  1162. scriptSrc = element.getAttribute("href");
  1163. if (this.options.saveOriginalURLs && !isDataURL(scriptSrc)) {
  1164. element.setAttribute("data-sf-original-href", scriptSrc);
  1165. }
  1166. }
  1167. element.removeAttribute("integrity");
  1168. element.textContent = "";
  1169. try {
  1170. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1171. } catch (error) {
  1172. // ignored
  1173. }
  1174. if (testValidURL(resourceURL)) {
  1175. if (element.tagName == "SCRIPT") {
  1176. element.removeAttribute("src");
  1177. } else {
  1178. element.removeAttribute("href");
  1179. }
  1180. this.stats.add("processed", "scripts", 1);
  1181. const content = await util.getContent(resourceURL, {
  1182. asBinary: true,
  1183. charset: this.charset != UTF8_CHARSET && this.charset,
  1184. maxResourceSize: this.options.maxResourceSize,
  1185. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1186. frameId: this.options.windowId,
  1187. resourceReferrer: this.options.resourceReferrer,
  1188. baseURI: this.options.baseURI,
  1189. blockMixedContent: this.options.blockMixedContent,
  1190. expectedType: "script",
  1191. acceptHeaders: this.options.acceptHeaders
  1192. });
  1193. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1194. if (element.tagName == "SCRIPT") {
  1195. element.setAttribute("src", content.data);
  1196. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1197. element.setAttribute("async", "");
  1198. }
  1199. } else {
  1200. const scriptElement = this.doc.createElement("script");
  1201. scriptElement.setAttribute("src", content.data);
  1202. scriptElement.setAttribute("async", "");
  1203. element.parentElement.replaceChild(scriptElement, element);
  1204. }
  1205. }
  1206. }));
  1207. }
  1208. removeAlternativeImages() {
  1209. util.removeAlternativeImages(this.doc);
  1210. }
  1211. async removeAlternativeFonts() {
  1212. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1213. }
  1214. async processFrames() {
  1215. if (this.options.frames) {
  1216. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1217. await Promise.all(frameElements.map(async frameElement => {
  1218. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1219. if (frameWindowId) {
  1220. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1221. if (frameData) {
  1222. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1223. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1224. this.stats.add("processed", "frames", 1);
  1225. await frameData.runner.run();
  1226. const pageData = await frameData.runner.getPageData();
  1227. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1228. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1229. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(CANVAS_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1230. sandbox += " allow-scripts allow-same-origin";
  1231. }
  1232. frameElement.setAttribute("sandbox", sandbox);
  1233. if (frameElement.tagName == "OBJECT") {
  1234. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1235. } else {
  1236. if (frameElement.tagName == "FRAME") {
  1237. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1238. } else {
  1239. frameElement.setAttribute("srcdoc", pageData.content);
  1240. frameElement.removeAttribute("src");
  1241. }
  1242. }
  1243. this.stats.addAll(pageData);
  1244. } else {
  1245. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1246. this.stats.add("discarded", "frames", 1);
  1247. }
  1248. }
  1249. }
  1250. }));
  1251. }
  1252. }
  1253. async processHtmlImports() {
  1254. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1255. await Promise.all(linkElements.map(async linkElement => {
  1256. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1257. if (attributeValue) {
  1258. const importData = this.options.imports[Number(attributeValue)];
  1259. if (importData.runner) {
  1260. this.stats.add("processed", "HTML imports", 1);
  1261. await importData.runner.run();
  1262. const pageData = await importData.runner.getPageData();
  1263. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1264. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1265. this.stats.addAll(pageData);
  1266. } else {
  1267. this.stats.add("discarded", "HTML imports", 1);
  1268. }
  1269. }
  1270. }));
  1271. }
  1272. replaceStylesheets() {
  1273. this.doc.querySelectorAll("style").forEach(styleElement => {
  1274. const stylesheetInfo = this.stylesheets.get(styleElement);
  1275. if (stylesheetInfo) {
  1276. this.stylesheets.delete(styleElement);
  1277. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1278. if (this.options.saveOriginalURLs) {
  1279. stylesheetContent = replaceOriginalURLs(stylesheetContent);
  1280. }
  1281. styleElement.textContent = stylesheetContent;
  1282. if (stylesheetInfo.mediaText) {
  1283. styleElement.media = stylesheetInfo.mediaText;
  1284. }
  1285. } else {
  1286. styleElement.remove();
  1287. }
  1288. });
  1289. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1290. const stylesheetInfo = this.stylesheets.get(linkElement);
  1291. if (stylesheetInfo) {
  1292. this.stylesheets.delete(linkElement);
  1293. const styleElement = this.doc.createElement("style");
  1294. if (stylesheetInfo.mediaText) {
  1295. styleElement.media = stylesheetInfo.mediaText;
  1296. }
  1297. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1298. if (this.options.saveOriginalURLs) {
  1299. stylesheetContent = replaceOriginalURLs(stylesheetContent);
  1300. styleElement.setAttribute("data-sf-original-href", linkElement.getAttribute("data-sf-original-href"));
  1301. }
  1302. styleElement.textContent = stylesheetContent;
  1303. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1304. } else {
  1305. linkElement.remove();
  1306. }
  1307. });
  1308. }
  1309. replaceStyleAttributes() {
  1310. this.doc.querySelectorAll("[style]").forEach(element => {
  1311. const declarations = this.styles.get(element);
  1312. if (declarations) {
  1313. this.styles.delete(element);
  1314. let styleContent = cssTree.generate(declarations);
  1315. if (this.options.saveOriginalURLs) {
  1316. styleContent = replaceOriginalURLs(styleContent);
  1317. }
  1318. element.setAttribute("style", styleContent);
  1319. } else {
  1320. element.setAttribute("style", "");
  1321. }
  1322. });
  1323. }
  1324. insertVariables() {
  1325. if (this.cssVariables.size) {
  1326. const styleElement = this.doc.createElement("style");
  1327. const firstStyleElement = this.doc.head.querySelector("style");
  1328. if (firstStyleElement) {
  1329. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1330. } else {
  1331. this.doc.head.appendChild(styleElement);
  1332. }
  1333. let stylesheetContent = "";
  1334. this.cssVariables.forEach(({ content, url }, indexResource) => {
  1335. this.cssVariables.delete(indexResource);
  1336. if (stylesheetContent) {
  1337. stylesheetContent += ";";
  1338. }
  1339. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}: `;
  1340. if (this.options.saveOriginalURLs) {
  1341. stylesheetContent += `/* original URL: ${url} */ `;
  1342. }
  1343. stylesheetContent += `url("${content}")`;
  1344. });
  1345. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1346. }
  1347. }
  1348. compressHTML() {
  1349. let size;
  1350. if (this.options.displayStats) {
  1351. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1352. }
  1353. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1354. if (this.options.displayStats) {
  1355. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1356. }
  1357. }
  1358. cleanupPage() {
  1359. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1360. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1361. if (metaCharset) {
  1362. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1363. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1364. this.doc.head.querySelector("meta[charset]").remove();
  1365. }
  1366. }
  1367. }
  1368. resetZoomLevel() {
  1369. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1370. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1371. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1372. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1373. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1374. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1375. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1376. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1377. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1378. this.doc.documentElement.style.removeProperty("-sf-transform");
  1379. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1380. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1381. }
  1382. async insertMAFFMetaData() {
  1383. const maffMetaData = await this.maffMetaDataPromise;
  1384. if (maffMetaData && maffMetaData.content) {
  1385. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1386. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1387. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1388. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1389. if (originalURLElement) {
  1390. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1391. }
  1392. if (archiveTimeElement) {
  1393. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1394. if (value) {
  1395. const date = new Date(value);
  1396. if (!isNaN(date.getTime())) {
  1397. this.options.saveDate = new Date(value);
  1398. }
  1399. }
  1400. }
  1401. }
  1402. }
  1403. async setDocInfo() {
  1404. const titleElement = this.doc.querySelector("title");
  1405. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1406. const authorElement = this.doc.querySelector("meta[name=author]");
  1407. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1408. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1409. const headingElement = this.doc.querySelector("h1");
  1410. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1411. this.options.info = {
  1412. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1413. lang: this.doc.documentElement.lang,
  1414. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1415. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1416. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1417. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1418. };
  1419. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1420. }
  1421. }
  1422. // ---------------
  1423. // ProcessorHelper
  1424. // ---------------
  1425. const DATA_URI_PREFIX = "data:";
  1426. const ABOUT_BLANK_URI = "about:blank";
  1427. const EMPTY_DATA_URI = "data:null;base64,";
  1428. const REGEXP_URL_HASH = /(#.+?)$/;
  1429. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1430. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1431. class ProcessorHelper {
  1432. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1433. const url = util.parseURL(options.saveUrl);
  1434. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1435. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1436. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1437. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1438. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1439. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1440. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1441. await evalDate(options.saveDate);
  1442. await evalDate(options.visitDate, "visit-");
  1443. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1444. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1445. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1446. const urlHref = decode(url.href);
  1447. template = await evalTemplateVariable(template, "url-href", () => urlHref || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1448. template = await evalTemplateVariable(template, "url-href-digest-sha-1", urlHref ? async () => util.digest("SHA-1", urlHref) : "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1449. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1450. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1451. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1452. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1453. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1454. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1455. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1456. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1457. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1458. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1459. for (const [name, value] of params) {
  1460. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1461. }
  1462. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1463. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1464. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1465. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1466. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1467. if (content) {
  1468. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1469. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1470. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1471. }
  1472. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1473. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1474. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1475. template = await evalTemplateVariable(template, "profile-name", () => options.profileName, dontReplaceSlash, options.filenameReplacementCharacter);
  1476. return template.trim();
  1477. function decode(value) {
  1478. try {
  1479. return decodeURI(value);
  1480. } catch (error) {
  1481. return value;
  1482. }
  1483. }
  1484. async function evalDate(date, prefix = "") {
  1485. if (date) {
  1486. template = await evalTemplateVariable(template, prefix + "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1487. template = await evalTemplateVariable(template, prefix + "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1488. template = await evalTemplateVariable(template, prefix + "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1489. template = await evalTemplateVariable(template, prefix + "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1490. template = await evalTemplateVariable(template, prefix + "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1491. template = await evalTemplateVariable(template, prefix + "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1492. template = await evalTemplateVariable(template, prefix + "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1493. template = await evalTemplateVariable(template, prefix + "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1494. template = await evalTemplateVariable(template, prefix + "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1495. template = await evalTemplateVariable(template, prefix + "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1496. template = await evalTemplateVariable(template, prefix + "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1497. template = await evalTemplateVariable(template, prefix + "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1498. template = await evalTemplateVariable(template, prefix + "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1499. template = await evalTemplateVariable(template, prefix + "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1500. template = await evalTemplateVariable(template, prefix + "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1501. template = await evalTemplateVariable(template, prefix + "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1502. template = await evalTemplateVariable(template, prefix + "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1503. template = await evalTemplateVariable(template, prefix + "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1504. template = await evalTemplateVariable(template, prefix + "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1505. template = await evalTemplateVariable(template, prefix + "time-ms", () => String(date.getTime()), dontReplaceSlash, options.filenameReplacementCharacter);
  1506. }
  1507. }
  1508. }
  1509. static setBackgroundImage(element, url, style) {
  1510. element.style.setProperty("background-blend-mode", "normal", "important");
  1511. element.style.setProperty("background-clip", "content-box", "important");
  1512. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1513. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1514. element.style.setProperty("background-image", url, "important");
  1515. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1516. element.style.setProperty("background-origin", "content-box", "important");
  1517. element.style.setProperty("background-repeat", "no-repeat", "important");
  1518. }
  1519. static processShortcutIcons(doc) {
  1520. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1521. if (!shortcutIcon) {
  1522. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1523. if (shortcutIcon) {
  1524. shortcutIcon.rel = "icon";
  1525. }
  1526. }
  1527. if (shortcutIcon) {
  1528. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1529. if (linkElement != shortcutIcon) {
  1530. linkElement.remove();
  1531. }
  1532. });
  1533. }
  1534. }
  1535. static removeSingleLineCssComments(stylesheet) {
  1536. const removedRules = [];
  1537. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1538. const ruleData = cssRule.data;
  1539. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1540. removedRules.push(cssRule);
  1541. }
  1542. }
  1543. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1544. }
  1545. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1546. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet, options.saveOriginalURLs);
  1547. const imports = getImportFunctions(stylesheetContent);
  1548. await Promise.all(imports.map(async cssImport => {
  1549. const match = matchImport(cssImport);
  1550. if (match) {
  1551. const regExpCssImport = getRegExp(cssImport);
  1552. let resourceURL = normalizeURL(match.resourceURL);
  1553. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1554. try {
  1555. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1556. } catch (error) {
  1557. // ignored
  1558. }
  1559. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1560. const content = await getStylesheetContent(resourceURL);
  1561. resourceURL = content.resourceURL;
  1562. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1563. if (content.data && content.data.match(/^<!doctype /i)) {
  1564. content.data = "";
  1565. }
  1566. let importedStylesheetContent = removeCssComments(content.data);
  1567. if (options.compressCSS) {
  1568. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1569. }
  1570. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1571. if (stylesheetContent.includes(cssImport)) {
  1572. const ancestorStyleSheets = new Set(importedStyleSheets);
  1573. ancestorStyleSheets.add(resourceURL);
  1574. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1575. workStylesheet.textContent = importedStylesheetContent;
  1576. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1577. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1578. } else {
  1579. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1580. }
  1581. }
  1582. } else {
  1583. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1584. }
  1585. } else {
  1586. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1587. }
  1588. }
  1589. }));
  1590. return stylesheetContent;
  1591. async function getStylesheetContent(resourceURL) {
  1592. const content = await util.getContent(resourceURL, {
  1593. maxResourceSize: options.maxResourceSize,
  1594. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1595. validateTextContentType: true,
  1596. frameId: options.frameId,
  1597. charset: options.charset,
  1598. resourceReferrer: options.resourceReferrer,
  1599. baseURI: options.baseURI,
  1600. blockMixedContent: options.blockMixedContent,
  1601. expectedType: "stylesheet",
  1602. acceptHeaders: options.acceptHeaders
  1603. });
  1604. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1605. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1606. return util.getContent(resourceURL, {
  1607. maxResourceSize: options.maxResourceSize,
  1608. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1609. validateTextContentType: true,
  1610. frameId: options.frameId,
  1611. charset: options.charset,
  1612. resourceReferrer: options.resourceReferrer,
  1613. baseURI: options.baseURI,
  1614. blockMixedContent: options.blockMixedContent,
  1615. expectedType: "stylesheet",
  1616. acceptHeaders: options.acceptHeaders
  1617. });
  1618. } else {
  1619. return content;
  1620. }
  1621. }
  1622. }
  1623. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet, saveOriginalURLs) {
  1624. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1625. if (saveOriginalURLs) {
  1626. stylesheetContent = addOriginalURLs(stylesheetContent);
  1627. }
  1628. urlFunctions.map(urlFunction => {
  1629. const originalResourceURL = matchURL(urlFunction);
  1630. let resourceURL = normalizeURL(originalResourceURL);
  1631. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1632. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1633. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1634. }
  1635. if (!testIgnoredPath(resourceURL)) {
  1636. if (!resourceURL || testValidPath(resourceURL)) {
  1637. let resolvedURL;
  1638. if (!originalResourceURL.startsWith("#")) {
  1639. try {
  1640. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1641. } catch (error) {
  1642. // ignored
  1643. }
  1644. }
  1645. if (testValidURL(resolvedURL) && originalResourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1646. try {
  1647. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1648. } catch (error) {
  1649. // ignored
  1650. }
  1651. }
  1652. } else {
  1653. let newUrlFunction;
  1654. if (originalResourceURL) {
  1655. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1656. } else {
  1657. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1658. }
  1659. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1660. }
  1661. }
  1662. });
  1663. return stylesheetContent;
  1664. }
  1665. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1666. resourceURL = normalizeURL(resourceURL);
  1667. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1668. const content = await util.getContent(resourceURL, {
  1669. maxResourceSize: options.maxResourceSize,
  1670. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1671. charset: options.charset,
  1672. frameId: options.frameId,
  1673. resourceReferrer: options.resourceReferrer,
  1674. validateTextContentType: true,
  1675. baseURI: baseURI,
  1676. blockMixedContent: options.blockMixedContent,
  1677. expectedType: "stylesheet",
  1678. acceptHeaders: options.acceptHeaders
  1679. });
  1680. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1681. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1682. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1683. }
  1684. resourceURL = content.resourceURL;
  1685. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1686. if (content.data && content.data.match(/^<!doctype /i)) {
  1687. content.data = "";
  1688. }
  1689. let stylesheetContent = removeCssComments(content.data);
  1690. if (options.compressCSS) {
  1691. stylesheetContent = util.compressCSS(stylesheetContent);
  1692. }
  1693. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1694. return stylesheetContent;
  1695. }
  1696. }
  1697. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1698. const promises = [];
  1699. const removedRules = [];
  1700. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1701. const ruleData = cssRule.data;
  1702. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1703. removedRules.push(cssRule);
  1704. } else if (ruleData.block && ruleData.block.children) {
  1705. if (ruleData.type == "Rule") {
  1706. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1707. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1708. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1709. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1710. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1711. }
  1712. }
  1713. }
  1714. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1715. await Promise.all(promises);
  1716. async function processFontFaceRule(ruleData, fontURLs) {
  1717. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1718. if (declaration.type == "Declaration" && declaration.value.children) {
  1719. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1720. await Promise.all(urlFunctions.map(async urlFunction => {
  1721. const originalResourceURL = matchURL(urlFunction);
  1722. const resourceURL = normalizeURL(originalResourceURL);
  1723. if (!testIgnoredPath(resourceURL)) {
  1724. if (testValidURL(resourceURL)) {
  1725. let { content } = await batchRequest.addURL(resourceURL,
  1726. { asBinary: true, expectedType: "font", baseURI, blockMixedContent: options.blockMixedContent });
  1727. let resourceURLs = fontURLs.get(declaration);
  1728. if (!resourceURLs) {
  1729. resourceURLs = [];
  1730. fontURLs.set(declaration, resourceURLs);
  1731. }
  1732. resourceURLs.push(resourceURL);
  1733. replaceURLs(declaration, originalResourceURL, content);
  1734. }
  1735. }
  1736. }));
  1737. }
  1738. }));
  1739. function replaceURLs(declaration, oldURL, newURL) {
  1740. declaration.value.children.forEach(token => {
  1741. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1742. token.value.value = newURL;
  1743. }
  1744. });
  1745. }
  1746. }
  1747. }
  1748. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1749. await Promise.all(declarations.map(async declaration => {
  1750. if (declaration.value && !declaration.value.children && declaration.value.type == "Raw") {
  1751. try {
  1752. declaration.value = cssTree.parse(declaration.value.value, { context: "value" });
  1753. } catch (error) {
  1754. // ignored
  1755. }
  1756. }
  1757. if (declaration.type == "Declaration" && declaration.value.children) {
  1758. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1759. await Promise.all(urlFunctions.map(async urlFunction => {
  1760. const originalResourceURL = matchURL(urlFunction);
  1761. const resourceURL = normalizeURL(originalResourceURL);
  1762. if (!testIgnoredPath(resourceURL)) {
  1763. if (testValidURL(resourceURL)) {
  1764. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL,
  1765. { asBinary: true, expectedType: "image", groupDuplicates: options.groupDuplicateImages });
  1766. let variableDefined;
  1767. const tokens = [];
  1768. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1769. if (!originalResourceURL.startsWith("#")) {
  1770. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1771. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1772. tokens.push({ parent, token, value });
  1773. variableDefined = true;
  1774. } else {
  1775. token.data.value.value = content;
  1776. }
  1777. }
  1778. });
  1779. if (variableDefined) {
  1780. cssVariables.set(indexResource, { content, url: originalResourceURL });
  1781. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1782. }
  1783. }
  1784. }
  1785. }));
  1786. }
  1787. }));
  1788. function findURLToken(url, children, callback, depth = 0) {
  1789. for (let token = children.head; token; token = token.next) {
  1790. if (token.data.children) {
  1791. findURLToken(url, token.data.children, callback, depth + 1);
  1792. }
  1793. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1794. callback(token, children, depth == 0);
  1795. }
  1796. }
  1797. }
  1798. }
  1799. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1800. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1801. let resourceURL = resourceElement.getAttribute(attributeName);
  1802. if (resourceURL != null) {
  1803. resourceURL = normalizeURL(resourceURL);
  1804. let originURL = resourceElement.dataset.singleFileOriginURL;
  1805. if (options.saveOriginalURLs && !isDataURL(resourceURL)) {
  1806. resourceElement.setAttribute("data-sf-original-" + attributeName, resourceURL);
  1807. }
  1808. delete resourceElement.dataset.singleFileOriginURL;
  1809. if (!testIgnoredPath(resourceURL)) {
  1810. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1811. if (testValidPath(resourceURL)) {
  1812. try {
  1813. resourceURL = util.resolveURL(resourceURL, baseURI);
  1814. } catch (error) {
  1815. // ignored
  1816. }
  1817. if (testValidURL(resourceURL)) {
  1818. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL,
  1819. { asBinary: true, expectedType: "image", groupDuplicates: options.groupDuplicateImages && resourceElement.tagName == "IMG" && attributeName == "src" });
  1820. if (originURL) {
  1821. if (content == EMPTY_DATA_URI) {
  1822. try {
  1823. originURL = util.resolveURL(originURL, baseURI);
  1824. } catch (error) {
  1825. // ignored
  1826. }
  1827. try {
  1828. resourceURL = originURL;
  1829. content = (await util.getContent(resourceURL, {
  1830. asBinary: true,
  1831. expectedType: "image",
  1832. maxResourceSize: options.maxResourceSize,
  1833. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1834. frameId: options.windowId,
  1835. resourceReferrer: options.resourceReferrer,
  1836. acceptHeaders: options.acceptHeaders
  1837. })).data;
  1838. } catch (error) {
  1839. // ignored
  1840. }
  1841. }
  1842. }
  1843. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1844. resourceElement.remove();
  1845. } else {
  1846. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1847. if (!forbiddenPrefixFound) {
  1848. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1849. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1850. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1851. cssVariables.set(indexResource, { content, url: originURL });
  1852. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1853. styles.set(resourceElement, declarationList);
  1854. } else {
  1855. resourceElement.setAttribute(attributeName, content);
  1856. }
  1857. } else {
  1858. resourceElement.setAttribute(attributeName, content);
  1859. }
  1860. }
  1861. }
  1862. }
  1863. }
  1864. }
  1865. }
  1866. }));
  1867. }
  1868. static async processXLinks(resourceElements, doc, baseURI, options, batchRequest) {
  1869. let attributeName = "xlink:href";
  1870. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1871. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1872. if (originalResourceURL == null) {
  1873. attributeName = "href";
  1874. originalResourceURL = resourceElement.getAttribute(attributeName);
  1875. }
  1876. if (options.saveOriginalURLs && !isDataURL(originalResourceURL)) {
  1877. resourceElement.setAttribute("data-sf-original-href", originalResourceURL);
  1878. }
  1879. let resourceURL = normalizeURL(originalResourceURL);
  1880. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1881. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1882. try {
  1883. resourceURL = util.resolveURL(resourceURL, baseURI);
  1884. } catch (error) {
  1885. // ignored
  1886. }
  1887. if (testValidURL(resourceURL)) {
  1888. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1889. if (originalResourceURL.startsWith(baseURI + "#")) {
  1890. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1891. } else {
  1892. const response = await batchRequest.addURL(resourceURL, { expectedType: "image" });
  1893. const svgDoc = util.parseSVGContent(response.content);
  1894. if (hashMatch && hashMatch[0]) {
  1895. let symbolElement;
  1896. try {
  1897. symbolElement = svgDoc.querySelector(hashMatch[0]);
  1898. } catch (error) {
  1899. // ignored
  1900. }
  1901. if (symbolElement) {
  1902. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1903. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1904. }
  1905. } else {
  1906. const content = await batchRequest.addURL(resourceURL, { expectedType: "image" });
  1907. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1908. }
  1909. }
  1910. }
  1911. } else if (resourceURL == options.url) {
  1912. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1913. }
  1914. }));
  1915. }
  1916. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1917. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1918. const originSrcset = resourceElement.getAttribute(attributeName);
  1919. const srcset = util.parseSrcset(originSrcset);
  1920. resourceElement.setAttribute("data-sf-original-srcset", originSrcset);
  1921. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1922. let resourceURL = normalizeURL(srcsetValue.url);
  1923. if (!testIgnoredPath(resourceURL)) {
  1924. if (testValidPath(resourceURL)) {
  1925. try {
  1926. resourceURL = util.resolveURL(resourceURL, baseURI);
  1927. } catch (error) {
  1928. // ignored
  1929. }
  1930. if (testValidURL(resourceURL)) {
  1931. const { content } = await batchRequest.addURL(resourceURL, { asBinary: true, expectedType: "image" });
  1932. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1933. if (forbiddenPrefixFound) {
  1934. return "";
  1935. }
  1936. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1937. } else {
  1938. return "";
  1939. }
  1940. } else {
  1941. return "";
  1942. }
  1943. } else {
  1944. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1945. }
  1946. }));
  1947. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1948. }));
  1949. }
  1950. static replaceImageSource(imgElement, variableName, options) {
  1951. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1952. if (attributeValue) {
  1953. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1954. if (imageData && imageData.replaceable) {
  1955. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1956. const backgroundStyle = {};
  1957. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1958. if (backgroundSize) {
  1959. backgroundStyle["background-size"] = imageData.objectFit;
  1960. }
  1961. if (imageData.objectPosition) {
  1962. backgroundStyle["background-position"] = imageData.objectPosition;
  1963. }
  1964. if (imageData.backgroundColor) {
  1965. backgroundStyle["background-color"] = imageData.backgroundColor;
  1966. }
  1967. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1968. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1969. return true;
  1970. }
  1971. }
  1972. }
  1973. }
  1974. // ----
  1975. // Util
  1976. // ----
  1977. const BLOB_URI_PREFIX = "blob:";
  1978. const HTTP_URI_PREFIX = /^https?:\/\//;
  1979. const FILE_URI_PREFIX = /^file:\/\//;
  1980. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1981. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1982. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1983. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1984. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1985. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1986. 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;
  1987. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1988. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1989. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1990. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1991. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1992. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1993. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1994. function getUpdatedResourceContent(resourceURL, content, options) {
  1995. if (options.rootDocument && options.updatedResources[resourceURL]) {
  1996. options.updatedResources[resourceURL].retrieved = true;
  1997. return options.updatedResources[resourceURL].content;
  1998. } else {
  1999. return content.data || "";
  2000. }
  2001. }
  2002. function normalizeURL(url) {
  2003. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  2004. return url;
  2005. } else {
  2006. return url.split("#")[0];
  2007. }
  2008. }
  2009. function getCSSValue(value) {
  2010. let result = "";
  2011. try {
  2012. result = cssTree.generate(value);
  2013. } catch (error) {
  2014. // ignored
  2015. }
  2016. return result;
  2017. }
  2018. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  2019. const stylesheetCharset = getCharset(stylesheetContent);
  2020. if (stylesheetCharset) {
  2021. return stylesheetCharset == charset.toLowerCase();
  2022. } else {
  2023. return true;
  2024. }
  2025. }
  2026. function getCharset(stylesheetContent) {
  2027. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  2028. if (match && match[1]) {
  2029. return match[1].toLowerCase().trim();
  2030. }
  2031. }
  2032. function getOnEventAttributeNames(doc) {
  2033. const element = doc.createElement("div");
  2034. const attributeNames = [];
  2035. for (const propertyName in element) {
  2036. if (propertyName.startsWith("on")) {
  2037. attributeNames.push(propertyName);
  2038. }
  2039. }
  2040. attributeNames.push("onunload");
  2041. return attributeNames;
  2042. }
  2043. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  2044. let maxLength;
  2045. if (template) {
  2046. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  2047. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  2048. if (template.match(replaceRegExp)) {
  2049. const matchedLength = template.match(replaceRegExp)[0];
  2050. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  2051. if (isNaN(maxLength) || maxLength <= 0) {
  2052. maxLength = null;
  2053. }
  2054. } else {
  2055. replaceRegExp = new RegExp(regExpVariable, "g");
  2056. }
  2057. if (template.match(replaceRegExp)) {
  2058. let value = await valueGetter();
  2059. if (!dontReplaceSlash) {
  2060. value = value.replace(/\/+/g, replacementCharacter);
  2061. }
  2062. if (maxLength) {
  2063. value = await util.truncateText(value, maxLength);
  2064. }
  2065. return template.replace(replaceRegExp, value);
  2066. }
  2067. }
  2068. return template;
  2069. }
  2070. function getLastSegment(url, replacementCharacter) {
  2071. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2072. if (!lastSegment) {
  2073. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  2074. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2075. }
  2076. if (!lastSegment) {
  2077. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  2078. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2079. }
  2080. if (!lastSegment) {
  2081. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  2082. }
  2083. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  2084. if (lastSegmentMatch && lastSegmentMatch[1]) {
  2085. lastSegment = lastSegmentMatch[1];
  2086. }
  2087. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  2088. return lastSegment;
  2089. }
  2090. function getRegExp(string) {
  2091. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  2092. }
  2093. function getUrlFunctions(stylesheetContent, unique) {
  2094. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  2095. if (unique) {
  2096. return [...new Set(result)];
  2097. } else {
  2098. return result;
  2099. }
  2100. }
  2101. function getImportFunctions(stylesheetContent) {
  2102. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  2103. }
  2104. function findShortcutIcon(shortcutIcons) {
  2105. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  2106. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  2107. return shortcutIcons[0];
  2108. }
  2109. function matchURL(stylesheetContent) {
  2110. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  2111. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  2112. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  2113. return match && match[1];
  2114. }
  2115. function addOriginalURLs(stylesheetContent) {
  2116. return stylesheetContent.replace(REGEXP_URL_FN, function (match, _0, url, _1, url2, _2, url3) {
  2117. url = url || url2 || url3;
  2118. if (isDataURL(url)) {
  2119. return match;
  2120. } else {
  2121. return "-sf-url-original(" + JSON.stringify(url) + ") " + match;
  2122. }
  2123. });
  2124. }
  2125. function isDataURL(url) {
  2126. return url && (url.startsWith(DATA_URI_PREFIX) || url.startsWith(BLOB_URI_PREFIX));
  2127. }
  2128. function replaceOriginalURLs(stylesheetContent) {
  2129. return stylesheetContent.replace(/-sf-url-original\("(.*?)"\)/g, "/* original URL: $1 */");
  2130. }
  2131. function testIgnoredPath(resourceURL) {
  2132. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  2133. }
  2134. function testValidPath(resourceURL) {
  2135. return resourceURL && !resourceURL.match(EMPTY_URL);
  2136. }
  2137. function testValidURL(resourceURL) {
  2138. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  2139. }
  2140. function matchImport(stylesheetContent) {
  2141. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  2142. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  2143. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2144. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2145. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2146. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2147. if (match) {
  2148. const [, resourceURL, media] = match;
  2149. return { resourceURL, media };
  2150. }
  2151. }
  2152. function removeCssComments(stylesheetContent) {
  2153. try {
  2154. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2155. } catch (error) {
  2156. let start, end;
  2157. do {
  2158. start = stylesheetContent.indexOf("/*");
  2159. end = stylesheetContent.indexOf("*/", start + 2);
  2160. if (start != -1 && end != -1) {
  2161. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2162. }
  2163. } while (start != -1 && end != -1);
  2164. return stylesheetContent;
  2165. }
  2166. }
  2167. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2168. if (mediaQuery) {
  2169. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2170. } else {
  2171. return stylesheetContent;
  2172. }
  2173. }
  2174. function log(...args) {
  2175. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2176. }
  2177. // -----
  2178. // Stats
  2179. // -----
  2180. const STATS_DEFAULT_VALUES = {
  2181. discarded: {
  2182. "HTML bytes": 0,
  2183. "hidden elements": 0,
  2184. "HTML imports": 0,
  2185. scripts: 0,
  2186. objects: 0,
  2187. "audio sources": 0,
  2188. "video sources": 0,
  2189. frames: 0,
  2190. "CSS rules": 0,
  2191. canvas: 0,
  2192. stylesheets: 0,
  2193. resources: 0,
  2194. medias: 0
  2195. },
  2196. processed: {
  2197. "HTML bytes": 0,
  2198. "hidden elements": 0,
  2199. "HTML imports": 0,
  2200. scripts: 0,
  2201. objects: 0,
  2202. "audio sources": 0,
  2203. "video sources": 0,
  2204. frames: 0,
  2205. "CSS rules": 0,
  2206. canvas: 0,
  2207. stylesheets: 0,
  2208. resources: 0,
  2209. medias: 0
  2210. }
  2211. };
  2212. class Stats {
  2213. constructor(options) {
  2214. this.options = options;
  2215. if (options.displayStats) {
  2216. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2217. }
  2218. }
  2219. set(type, subType, value) {
  2220. if (this.options.displayStats) {
  2221. this.data[type][subType] = value;
  2222. }
  2223. }
  2224. add(type, subType, value) {
  2225. if (this.options.displayStats) {
  2226. this.data[type][subType] += value;
  2227. }
  2228. }
  2229. addAll(pageData) {
  2230. if (this.options.displayStats) {
  2231. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2232. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2233. }
  2234. }
  2235. }
  2236. export {
  2237. getClass
  2238. };