single-file-core.js 97 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373
  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: "moveStylesInHead", action: "moveStylesInHead" },
  96. { option: "removeScripts", action: "removeScripts" },
  97. { option: "selected", action: "removeUnselectedElements" },
  98. { option: "removeVideoSrc", action: "insertVideoPosters" },
  99. { option: "removeFrames", action: "removeFrames" },
  100. { option: "removeVideoSrc", action: "removeVideoSources" },
  101. { option: "removeAudioSrc", action: "removeAudioSources" },
  102. { action: "removeDiscardedResources" },
  103. { option: "removeHiddenElements", action: "removeHiddenElements" },
  104. { action: "resolveHrefs" },
  105. { action: "resolveStyleAttributeURLs" }
  106. ],
  107. parallel: [
  108. { action: "resolveStylesheetURLs" },
  109. { option: "!removeFrames", action: "resolveFrameURLs" },
  110. { action: "resolveHtmlImportURLs" }
  111. ]
  112. }, {
  113. sequential: [
  114. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  115. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" },
  116. { option: "removeUnusedFonts", action: "removeUnusedFonts" }
  117. ],
  118. parallel: [
  119. { action: "processStylesheets" },
  120. { action: "processStyleAttributes" },
  121. { action: "processPageResources" },
  122. { option: "!removeScripts", action: "processScripts" }
  123. ]
  124. }, {
  125. sequential: [
  126. { option: "removeAlternativeImages", action: "removeAlternativeImages" }
  127. ],
  128. parallel: [
  129. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  130. { option: "!removeFrames", action: "processFrames" },
  131. { option: "!removeImports", action: "processHtmlImports" },
  132. ]
  133. }, {
  134. sequential: [
  135. { action: "replaceStylesheets" },
  136. { action: "replaceStyleAttributes" },
  137. { action: "insertVariables" },
  138. { option: "compressHTML", action: "compressHTML" },
  139. { action: "cleanupPage" }
  140. ],
  141. parallel: [
  142. { option: "enableMaff", action: "insertMAFFMetaData" },
  143. { action: "setDocInfo" }
  144. ]
  145. }];
  146. class Runner {
  147. constructor(options, root) {
  148. const rootDocDefined = root && options.doc;
  149. this.root = root;
  150. this.options = options;
  151. this.options.url = this.options.url || (rootDocDefined && this.options.doc.location.href);
  152. const matchResourceReferrer = this.options.url.match(/^.*\//);
  153. this.options.resourceReferrer = this.options.passReferrerOnError && matchResourceReferrer && matchResourceReferrer[0];
  154. this.options.baseURI = rootDocDefined && this.options.doc.baseURI;
  155. this.options.rootDocument = root;
  156. this.options.updatedResources = this.options.updatedResources || {};
  157. this.options.fontTests = new Map();
  158. this.batchRequest = new BatchRequest();
  159. this.processor = new Processor(options, this.batchRequest);
  160. if (rootDocDefined) {
  161. const docData = util.preProcessDoc(this.options.doc, this.options.win, this.options);
  162. this.options.canvases = docData.canvases;
  163. this.options.fonts = docData.fonts;
  164. this.options.stylesheets = docData.stylesheets;
  165. this.options.images = docData.images;
  166. this.options.posters = docData.posters;
  167. this.options.usedFonts = docData.usedFonts;
  168. this.options.shadowRoots = docData.shadowRoots;
  169. this.options.imports = docData.imports;
  170. this.options.referrer = docData.referrer;
  171. this.markedElements = docData.markedElements;
  172. }
  173. if (this.options.saveRawPage) {
  174. this.options.removeFrames = true;
  175. }
  176. this.options.content = this.options.content || (rootDocDefined ? util.serialize(this.options.doc) : null);
  177. this.onprogress = options.onprogress || (() => { });
  178. }
  179. async loadPage() {
  180. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url, frame: !this.root }));
  181. await this.processor.loadPage(this.options.content);
  182. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url, frame: !this.root }));
  183. }
  184. async initialize() {
  185. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  186. await this.executeStage(RESOLVE_URLS_STAGE);
  187. this.pendingPromises = this.executeStage(REPLACE_DATA_STAGE);
  188. if (this.root && this.options.doc) {
  189. util.postProcessDoc(this.options.doc, this.markedElements);
  190. }
  191. }
  192. cancel() {
  193. this.cancelled = true;
  194. this.batchRequest.cancel();
  195. if (this.root) {
  196. if (this.options.frames) {
  197. this.options.frames.forEach(cancelRunner);
  198. }
  199. if (this.options.imports) {
  200. this.options.imports.forEach(cancelRunner);
  201. }
  202. }
  203. function cancelRunner(resourceData) {
  204. if (resourceData.runner) {
  205. resourceData.runner.cancel();
  206. }
  207. }
  208. }
  209. async run() {
  210. if (this.root) {
  211. this.processor.initialize(this.batchRequest);
  212. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, max: this.processor.maxResources }));
  213. }
  214. await this.batchRequest.run(detail => {
  215. detail.pageURL = this.options.url;
  216. this.onprogress(new ProgressEvent(RESOURCE_LOADED, detail));
  217. }, this.options);
  218. await this.pendingPromises;
  219. this.options.doc = null;
  220. this.options.win = null;
  221. await this.executeStage(REPLACE_DOCS_STAGE);
  222. await this.executeStage(POST_PROCESS_STAGE);
  223. this.processor.finalize();
  224. }
  225. getDocument() {
  226. return this.processor.doc;
  227. }
  228. getStyleSheets() {
  229. return this.processor.stylesheets;
  230. }
  231. getPageData() {
  232. if (this.root) {
  233. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  234. }
  235. return this.processor.getPageData();
  236. }
  237. async executeStage(step) {
  238. if (DEBUG) {
  239. log("**** STARTED STAGE", step, "****");
  240. }
  241. const frame = !this.root;
  242. this.onprogress(new ProgressEvent(STAGE_STARTED, { pageURL: this.options.url, step, frame }));
  243. STAGES[step].sequential.forEach(task => {
  244. let startTime;
  245. if (DEBUG) {
  246. startTime = Date.now();
  247. log(" -- STARTED task =", task.action);
  248. }
  249. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  250. if (!this.cancelled) {
  251. this.executeTask(task);
  252. }
  253. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  254. if (DEBUG) {
  255. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  256. }
  257. });
  258. let parallelTasksPromise;
  259. if (STAGES[step].parallel) {
  260. parallelTasksPromise = await Promise.all(STAGES[step].parallel.map(async task => {
  261. let startTime;
  262. if (DEBUG) {
  263. startTime = Date.now();
  264. log(" // STARTED task =", task.action);
  265. }
  266. this.onprogress(new ProgressEvent(STAGE_TASK_STARTED, { pageURL: this.options.url, step, task: task.action, frame }));
  267. if (!this.cancelled) {
  268. await this.executeTask(task);
  269. }
  270. this.onprogress(new ProgressEvent(STAGE_TASK_ENDED, { pageURL: this.options.url, step, task: task.action, frame }));
  271. if (DEBUG) {
  272. log(" // ENDED task =", task.action, "delay =", Date.now() - startTime);
  273. }
  274. }));
  275. } else {
  276. parallelTasksPromise = Promise.resolve();
  277. }
  278. this.onprogress(new ProgressEvent(STAGE_ENDED, { pageURL: this.options.url, step, frame }));
  279. if (DEBUG) {
  280. log("**** ENDED STAGE", step, "****");
  281. }
  282. return parallelTasksPromise;
  283. }
  284. executeTask(task) {
  285. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  286. return this.processor[task.action]();
  287. }
  288. }
  289. }
  290. // ------------
  291. // BatchRequest
  292. // ------------
  293. class BatchRequest {
  294. constructor() {
  295. this.requests = new Map();
  296. this.duplicates = new Map();
  297. }
  298. addURL(resourceURL, { asBinary, expectedType, groupDuplicates, baseURI, blockMixedContent } = {}) {
  299. return new Promise((resolve, reject) => {
  300. const requestKey = JSON.stringify([resourceURL, asBinary, expectedType, baseURI, blockMixedContent]);
  301. let resourceRequests = this.requests.get(requestKey);
  302. if (!resourceRequests) {
  303. resourceRequests = [];
  304. this.requests.set(requestKey, resourceRequests);
  305. }
  306. const callbacks = { resolve, reject };
  307. resourceRequests.push(callbacks);
  308. if (groupDuplicates) {
  309. let duplicateRequests = this.duplicates.get(requestKey);
  310. if (!duplicateRequests) {
  311. duplicateRequests = [];
  312. this.duplicates.set(requestKey, duplicateRequests);
  313. }
  314. duplicateRequests.push(callbacks);
  315. }
  316. });
  317. }
  318. getMaxResources() {
  319. return this.requests.size;
  320. }
  321. run(onloadListener, options) {
  322. const resourceURLs = [...this.requests.keys()];
  323. let indexResource = 0;
  324. return Promise.all(resourceURLs.map(async requestKey => {
  325. const [resourceURL, asBinary, expectedType, baseURI, blockMixedContent] = JSON.parse(requestKey);
  326. const resourceRequests = this.requests.get(requestKey);
  327. try {
  328. const currentIndexResource = indexResource;
  329. indexResource = indexResource + 1;
  330. const content = await util.getContent(resourceURL, {
  331. asBinary,
  332. expectedType,
  333. maxResourceSize: options.maxResourceSize,
  334. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  335. frameId: options.windowId,
  336. resourceReferrer: options.resourceReferrer,
  337. baseURI,
  338. blockMixedContent,
  339. acceptHeaders: options.acceptHeaders
  340. });
  341. onloadListener({ url: resourceURL });
  342. if (!this.cancelled) {
  343. resourceRequests.forEach(callbacks => {
  344. const duplicateCallbacks = this.duplicates.get(requestKey);
  345. const duplicate = duplicateCallbacks && duplicateCallbacks.length > 1 && duplicateCallbacks.includes(callbacks);
  346. callbacks.resolve({ content: content.data, indexResource: currentIndexResource, duplicate });
  347. });
  348. }
  349. } catch (error) {
  350. indexResource = indexResource + 1;
  351. onloadListener({ url: resourceURL });
  352. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  353. }
  354. this.requests.delete(requestKey);
  355. }));
  356. }
  357. cancel() {
  358. this.cancelled = true;
  359. const resourceURLs = [...this.requests.keys()];
  360. resourceURLs.forEach(requestKey => {
  361. const resourceRequests = this.requests.get(requestKey);
  362. resourceRequests.forEach(callbacks => callbacks.reject());
  363. this.requests.delete(requestKey);
  364. });
  365. }
  366. }
  367. // ---------
  368. // Processor
  369. // ---------
  370. const PREFIXES_FORBIDDEN_DATA_URI = ["data:text/"];
  371. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  372. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  373. const SCRIPT_TAG_FOUND = /<script/gi;
  374. const NOSCRIPT_TAG_FOUND = /<noscript/gi;
  375. const CANVAS_TAG_FOUND = /<canvas/gi;
  376. const SHADOW_MODE_ATTRIBUTE_NAME = "shadowmode";
  377. const SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME = "delegatesfocus";
  378. const SCRIPT_TEMPLATE_SHADOW_ROOT = "data-template-shadow-root";
  379. const UTF8_CHARSET = "utf-8";
  380. class Processor {
  381. constructor(options, batchRequest) {
  382. this.options = options;
  383. this.stats = new Stats(options);
  384. this.baseURI = normalizeURL(options.baseURI || options.url);
  385. this.batchRequest = batchRequest;
  386. this.stylesheets = new Map();
  387. this.styles = new Map();
  388. this.cssVariables = new Map();
  389. this.fontTests = options.fontTests;
  390. }
  391. initialize() {
  392. this.options.saveDate = new Date();
  393. this.options.saveUrl = this.options.url;
  394. if (this.options.enableMaff) {
  395. this.maffMetaDataPromise = this.batchRequest.addURL(util.resolveURL("index.rdf", this.options.baseURI || this.options.url), { expectedType: "document" });
  396. }
  397. this.maxResources = this.batchRequest.getMaxResources();
  398. if (!this.options.saveRawPage && !this.options.removeFrames && this.options.frames) {
  399. this.options.frames.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  400. }
  401. if (!this.options.removeImports && this.options.imports) {
  402. this.options.imports.forEach(importData => this.maxResources += importData.maxResources || 0);
  403. }
  404. this.stats.set("processed", "resources", this.maxResources);
  405. }
  406. async loadPage(pageContent, charset) {
  407. let content;
  408. if (!pageContent || this.options.saveRawPage) {
  409. content = await util.getContent(this.baseURI, {
  410. maxResourceSize: this.options.maxResourceSize,
  411. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  412. charset,
  413. frameId: this.options.windowId,
  414. resourceReferrer: this.options.resourceReferrer,
  415. expectedType: "document",
  416. acceptHeaders: this.options.acceptHeaders
  417. });
  418. pageContent = content.data;
  419. }
  420. this.doc = util.parseDocContent(pageContent, this.baseURI);
  421. if (this.options.saveRawPage) {
  422. let charset;
  423. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  424. const charsetDeclaration = element.content.split(";")[1];
  425. if (charsetDeclaration && !charset) {
  426. charset = charsetDeclaration.split("=")[1].trim().toLowerCase();
  427. }
  428. });
  429. if (charset && content.charset && charset.toLowerCase() != content.charset.toLowerCase()) {
  430. return this.loadPage(pageContent, charset);
  431. }
  432. }
  433. this.workStyleElement = this.doc.createElement("style");
  434. this.doc.body.appendChild(this.workStyleElement);
  435. this.onEventAttributeNames = getOnEventAttributeNames(this.doc);
  436. }
  437. finalize() {
  438. if (this.workStyleElement.parentNode) {
  439. this.workStyleElement.remove();
  440. }
  441. }
  442. async getPageData() {
  443. util.postProcessDoc(this.doc);
  444. const url = util.parseURL(this.baseURI);
  445. if (this.options.insertSingleFileComment) {
  446. const firstComment = this.doc.documentElement.firstChild;
  447. let infobarURL = this.options.saveUrl, infobarSaveDate = this.options.saveDate;
  448. if (firstComment.nodeType == 8 && (firstComment.textContent.includes(util.COMMENT_HEADER_LEGACY) || firstComment.textContent.includes(util.COMMENT_HEADER))) {
  449. const info = this.doc.documentElement.firstChild.textContent.split("\n");
  450. const [, , url, saveDate] = info;
  451. infobarURL = url.split("url: ")[1];
  452. infobarSaveDate = saveDate.split("saved date: ")[1];
  453. firstComment.remove();
  454. }
  455. const infobarContent = (this.options.infobarContent || "").replace(/\\n/g, "\n").replace(/\\t/g, "\t");
  456. const commentNode = this.doc.createComment("\n " + (this.options.useLegacyCommentHeader ? util.COMMENT_HEADER_LEGACY : util.COMMENT_HEADER) +
  457. " \n url: " + infobarURL +
  458. " \n saved date: " + infobarSaveDate +
  459. (infobarContent ? " \n info: " + infobarContent : "") + "\n");
  460. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  461. }
  462. if (this.options.insertCanonicalLink && this.options.saveUrl.match(HTTP_URI_PREFIX)) {
  463. let canonicalLink = this.doc.querySelector("link[rel=canonical]");
  464. if (!canonicalLink) {
  465. canonicalLink = this.doc.createElement("link");
  466. canonicalLink.setAttribute("rel", "canonical");
  467. this.doc.head.appendChild(canonicalLink);
  468. }
  469. if (canonicalLink && !canonicalLink.href) {
  470. canonicalLink.href = this.options.saveUrl;
  471. }
  472. }
  473. if (this.options.insertMetaCSP) {
  474. const metaTag = this.doc.createElement("meta");
  475. metaTag.httpEquiv = "content-security-policy";
  476. 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:;";
  477. this.doc.head.appendChild(metaTag);
  478. }
  479. if (this.options.insertMetaNoIndex) {
  480. let metaElement = this.doc.querySelector("meta[name=robots][content*=noindex]");
  481. if (!metaElement) {
  482. metaElement = this.doc.createElement("meta");
  483. metaElement.setAttribute("name", "robots");
  484. metaElement.setAttribute("content", "noindex");
  485. this.doc.head.appendChild(metaElement);
  486. }
  487. }
  488. let size;
  489. if (this.options.displayStats) {
  490. size = util.getContentSize(this.doc.documentElement.outerHTML);
  491. }
  492. const content = util.serialize(this.doc, this.options.compressHTML);
  493. if (this.options.displayStats) {
  494. const contentSize = util.getContentSize(content);
  495. this.stats.set("processed", "HTML bytes", contentSize);
  496. this.stats.add("discarded", "HTML bytes", size - contentSize);
  497. }
  498. let filename = await ProcessorHelper.evalTemplate(this.options.filenameTemplate, this.options, content) || "";
  499. const replacementCharacter = this.options.filenameReplacementCharacter;
  500. filename = util.getValidFilename(filename, this.options.filenameReplacedCharacters, replacementCharacter);
  501. if (!this.options.backgroundSave) {
  502. filename = filename.replace(/\//g, replacementCharacter);
  503. }
  504. if (!this.options.saveToGDrive && !this.options.saveToGitHub && !this.options.saveWithCompanion && util.getContentSize(filename) > this.options.filenameMaxLength) {
  505. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  506. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  507. filename = await util.truncateText(filename, this.options.filenameMaxLength - extension.length);
  508. filename = filename + "…" + extension;
  509. }
  510. if (!filename) {
  511. filename = "Unnamed page";
  512. }
  513. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  514. const pageData = {
  515. stats: this.stats.data,
  516. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  517. filename,
  518. content
  519. };
  520. if (this.options.addProof) {
  521. pageData.hash = await util.digest("SHA-256", content);
  522. }
  523. if (this.options.retrieveLinks) {
  524. pageData.links = Array.from(new Set(Array.from(this.doc.links).map(linkElement => linkElement.href)));
  525. }
  526. return pageData;
  527. }
  528. preProcessPage() {
  529. if (this.options.win) {
  530. 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));
  531. }
  532. if (this.options.images && !this.options.saveRawPage) {
  533. this.doc.querySelectorAll("img[" + util.IMAGE_ATTRIBUTE_NAME + "]").forEach(imgElement => {
  534. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  535. if (attributeValue) {
  536. const imageData = this.options.images[Number(attributeValue)];
  537. if (imageData) {
  538. if (this.options.removeHiddenElements && (
  539. (imageData.size && !imageData.size.pxWidth && !imageData.size.pxHeight) ||
  540. (imgElement.getAttribute(util.HIDDEN_CONTENT_ATTRIBUTE_NAME) == "")
  541. )) {
  542. imgElement.setAttribute("src", EMPTY_IMAGE);
  543. } else {
  544. if (imageData.currentSrc) {
  545. imgElement.dataset.singleFileOriginURL = imgElement.getAttribute("src");
  546. imgElement.setAttribute("src", imageData.currentSrc);
  547. }
  548. if (this.options.loadDeferredImages) {
  549. if ((!imgElement.getAttribute("src") || imgElement.getAttribute("src") == EMPTY_IMAGE) && imgElement.getAttribute("data-src")) {
  550. imageData.src = imgElement.dataset.src;
  551. imgElement.setAttribute("src", imgElement.dataset.src);
  552. imgElement.removeAttribute("data-src");
  553. }
  554. }
  555. }
  556. }
  557. }
  558. });
  559. if (this.options.loadDeferredImages) {
  560. this.doc.querySelectorAll("img[data-srcset]").forEach(imgElement => {
  561. if (!imgElement.getAttribute("srcset") && imgElement.getAttribute("data-srcset")) {
  562. imgElement.setAttribute("srcset", imgElement.dataset.srcset);
  563. imgElement.removeAttribute("data-srcset");
  564. }
  565. });
  566. }
  567. }
  568. }
  569. replaceStyleContents() {
  570. if (this.options.stylesheets) {
  571. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  572. const attributeValue = styleElement.getAttribute(util.STYLESHEET_ATTRIBUTE_NAME);
  573. if (attributeValue) {
  574. const stylesheetContent = this.options.stylesheets[Number(styleIndex)];
  575. if (stylesheetContent) {
  576. styleElement.textContent = stylesheetContent;
  577. }
  578. }
  579. });
  580. }
  581. }
  582. removeUnselectedElements() {
  583. removeUnmarkedElements(this.doc.body);
  584. this.doc.body.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  585. function removeUnmarkedElements(element) {
  586. let selectedElementFound = false;
  587. Array.from(element.childNodes).forEach(node => {
  588. if (node.nodeType == 1) {
  589. const isSelectedElement = node.getAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME) == "";
  590. selectedElementFound = selectedElementFound || isSelectedElement;
  591. if (isSelectedElement) {
  592. node.removeAttribute(util.SELECTED_CONTENT_ATTRIBUTE_NAME);
  593. removeUnmarkedElements(node);
  594. } else if (selectedElementFound) {
  595. removeNode(node);
  596. } else {
  597. hideNode(node);
  598. }
  599. }
  600. });
  601. }
  602. function removeNode(node) {
  603. if ((node.nodeType != 1 || !node.querySelector("svg,style,link")) && canHideNode(node)) {
  604. node.remove();
  605. } else {
  606. hideNode(node);
  607. }
  608. }
  609. function hideNode(node) {
  610. if (canHideNode(node)) {
  611. node.style.setProperty("display", "none", "important");
  612. Array.from(node.childNodes).forEach(removeNode);
  613. }
  614. }
  615. function canHideNode(node) {
  616. if (node.nodeType == 1) {
  617. const tagName = node.tagName && node.tagName.toLowerCase();
  618. return (tagName != "svg" && tagName != "style" && tagName != "link");
  619. }
  620. }
  621. }
  622. insertVideoPosters() {
  623. if (this.options.posters) {
  624. this.doc.querySelectorAll("video[src], video > source[src]").forEach(element => {
  625. let videoElement;
  626. if (element.tagName == "VIDEO") {
  627. videoElement = element;
  628. } else {
  629. videoElement = element.parentElement;
  630. }
  631. const attributeValue = element.getAttribute(util.POSTER_ATTRIBUTE_NAME);
  632. if (attributeValue) {
  633. const posterURL = this.options.posters[Number(attributeValue)];
  634. if (!videoElement.poster && posterURL) {
  635. videoElement.setAttribute("poster", posterURL);
  636. }
  637. }
  638. });
  639. }
  640. }
  641. removeFrames() {
  642. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  643. this.stats.set("discarded", "frames", frameElements.length);
  644. this.stats.set("processed", "frames", frameElements.length);
  645. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  646. }
  647. removeScripts() {
  648. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  649. this.doc.querySelectorAll("[href]").forEach(element => {
  650. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  651. element.setAttribute("href", "");
  652. }
  653. });
  654. this.doc.querySelectorAll("[src]").forEach(element => {
  655. if (element.src && element.src.match(/^\s*javascript:/)) {
  656. element.removeAttribute("src");
  657. }
  658. });
  659. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"]):not([" + SCRIPT_TEMPLATE_SHADOW_ROOT + "])");
  660. this.stats.set("discarded", "scripts", scriptElements.length);
  661. this.stats.set("processed", "scripts", scriptElements.length);
  662. scriptElements.forEach(element => element.remove());
  663. }
  664. removeVideoSources() {
  665. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source");
  666. this.stats.set("discarded", "video sources", videoSourceElements.length);
  667. this.stats.set("processed", "video sources", videoSourceElements.length);
  668. videoSourceElements.forEach(element => {
  669. if (element.tagName == "SOURCE") {
  670. element.remove();
  671. } else {
  672. videoSourceElements.forEach(element => element.removeAttribute("src"));
  673. }
  674. });
  675. }
  676. removeAudioSources() {
  677. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  678. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  679. this.stats.set("processed", "audio sources", audioSourceElements.length);
  680. audioSourceElements.forEach(element => {
  681. if (element.tagName == "SOURCE") {
  682. element.remove();
  683. } else {
  684. audioSourceElements.forEach(element => element.removeAttribute("src"));
  685. }
  686. });
  687. }
  688. removeDiscardedResources() {
  689. this.doc.querySelectorAll("." + util.SINGLE_FILE_UI_ELEMENT_CLASS).forEach(element => element.remove());
  690. this.doc.querySelectorAll("button button").forEach(element => element.remove());
  691. this.doc.querySelectorAll("meta[http-equiv=refresh], meta[disabled-http-equiv], meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  692. 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\"])");
  693. this.stats.set("discarded", "objects", objectElements.length);
  694. this.stats.set("processed", "objects", objectElements.length);
  695. objectElements.forEach(element => element.remove());
  696. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=manifest], link[rel~=prefetch]");
  697. replacedAttributeValue.forEach(element => {
  698. let regExp;
  699. if (this.options.removeScripts) {
  700. regExp = /(preconnect|prerender|dns-prefetch|preload|prefetch|manifest)/g;
  701. } else {
  702. regExp = /(preconnect|prerender|dns-prefetch|prefetch|manifest)/g;
  703. }
  704. const relValue = element.getAttribute("rel").replace(regExp, "").trim();
  705. if (relValue.length) {
  706. element.setAttribute("rel", relValue);
  707. } else {
  708. element.remove();
  709. }
  710. });
  711. this.doc.querySelectorAll("link[rel*=stylesheet][rel*=alternate][title],link[rel*=stylesheet]:not([href]),link[rel*=stylesheet][href=\"\"]").forEach(element => element.remove());
  712. if (this.options.removeHiddenElements) {
  713. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  714. }
  715. if (!this.options.saveFavicon) {
  716. this.doc.querySelectorAll("link[rel*=\"icon\"]").forEach(element => element.remove());
  717. }
  718. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  719. }
  720. resetCharsetMeta() {
  721. let charset;
  722. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => {
  723. const charsetDeclaration = element.content.split(";")[1];
  724. if (charsetDeclaration && !charset) {
  725. charset = charsetDeclaration.split("=")[1];
  726. if (charset) {
  727. this.charset = charset.trim().toLowerCase();
  728. }
  729. }
  730. element.remove();
  731. });
  732. const metaElement = this.doc.createElement("meta");
  733. metaElement.setAttribute("charset", UTF8_CHARSET);
  734. if (this.doc.head.firstChild) {
  735. this.doc.head.insertBefore(metaElement, this.doc.head.firstChild);
  736. } else {
  737. this.doc.head.appendChild(metaElement);
  738. }
  739. }
  740. setInputValues() {
  741. this.doc.querySelectorAll("input:not([type=radio]):not([type=checkbox])").forEach(input => {
  742. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  743. input.setAttribute("value", value || "");
  744. });
  745. this.doc.querySelectorAll("input[type=radio], input[type=checkbox]").forEach(input => {
  746. const value = input.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  747. if (value == "true") {
  748. input.setAttribute("checked", "");
  749. }
  750. });
  751. this.doc.querySelectorAll("textarea").forEach(textarea => {
  752. const value = textarea.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME);
  753. textarea.textContent = value || "";
  754. });
  755. this.doc.querySelectorAll("select").forEach(select => {
  756. select.querySelectorAll("option").forEach(option => {
  757. const selected = option.getAttribute(util.INPUT_VALUE_ATTRIBUTE_NAME) != null;
  758. if (selected) {
  759. option.setAttribute("selected", "");
  760. }
  761. });
  762. });
  763. }
  764. moveStylesInHead() {
  765. this.doc.querySelectorAll("style").forEach(stylesheet => {
  766. if (stylesheet.getAttribute(util.STYLE_ATTRIBUTE_NAME) == "") {
  767. this.doc.head.appendChild(stylesheet);
  768. }
  769. });
  770. }
  771. saveFavicon() {
  772. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  773. if (!faviconElement) {
  774. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  775. }
  776. if (!faviconElement) {
  777. faviconElement = this.doc.createElement("link");
  778. faviconElement.setAttribute("type", "image/x-icon");
  779. faviconElement.setAttribute("rel", "shortcut icon");
  780. faviconElement.setAttribute("href", "/favicon.ico");
  781. }
  782. this.doc.head.appendChild(faviconElement);
  783. }
  784. replaceCanvasElements() {
  785. if (this.options.canvases) {
  786. this.doc.querySelectorAll("canvas").forEach(canvasElement => {
  787. const attributeValue = canvasElement.getAttribute(util.CANVAS_ATTRIBUTE_NAME);
  788. if (attributeValue) {
  789. const canvasData = this.options.canvases[Number(attributeValue)];
  790. if (canvasData) {
  791. ProcessorHelper.setBackgroundImage(canvasElement, "url(" + canvasData.dataURI + ")");
  792. this.stats.add("processed", "canvas", 1);
  793. }
  794. }
  795. });
  796. }
  797. }
  798. insertFonts() {
  799. if (this.options.fonts && this.options.fonts.length) {
  800. let stylesheetContent = "";
  801. this.options.fonts.forEach(fontData => {
  802. if (fontData["font-family"] && fontData.src) {
  803. stylesheetContent += "@font-face{";
  804. let stylesContent = "";
  805. Object.keys(fontData).forEach(fontStyle => {
  806. if (stylesContent) {
  807. stylesContent += ";";
  808. }
  809. stylesContent += fontStyle + ":" + fontData[fontStyle];
  810. });
  811. stylesheetContent += stylesContent + "}";
  812. }
  813. });
  814. if (stylesheetContent) {
  815. const styleElement = this.doc.createElement("style");
  816. styleElement.textContent = stylesheetContent;
  817. const existingStyleElement = this.doc.querySelector("style");
  818. if (existingStyleElement) {
  819. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  820. } else {
  821. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  822. }
  823. }
  824. }
  825. }
  826. removeHiddenElements() {
  827. const hiddenElements = this.doc.querySelectorAll("[" + util.HIDDEN_CONTENT_ATTRIBUTE_NAME + "]");
  828. const removedElements = this.doc.querySelectorAll("[" + util.REMOVED_CONTENT_ATTRIBUTE_NAME + "]");
  829. this.stats.set("discarded", "hidden elements", removedElements.length);
  830. this.stats.set("processed", "hidden elements", removedElements.length);
  831. if (hiddenElements.length) {
  832. const styleElement = this.doc.createElement("style");
  833. styleElement.textContent = ".sf-hidden{display:none!important;}";
  834. this.doc.head.appendChild(styleElement);
  835. hiddenElements.forEach(element => {
  836. if (element.style.getPropertyValue("display") != "none") {
  837. if (element.style.getPropertyPriority("display") == "important") {
  838. element.style.setProperty("display", "none", "important");
  839. } else {
  840. element.classList.add("sf-hidden");
  841. }
  842. }
  843. });
  844. }
  845. removedElements.forEach(element => element.remove());
  846. }
  847. resolveHrefs() {
  848. this.doc.querySelectorAll("a[href], area[href], link[href]").forEach(element => {
  849. const href = element.getAttribute("href").trim();
  850. if (element.tagName == "LINK" && element.rel.includes("stylesheet")) {
  851. if (this.options.saveOriginalURLs && !isDataURL(href)) {
  852. element.setAttribute("data-sf-original-href", href);
  853. }
  854. }
  855. if (!testIgnoredPath(href)) {
  856. let resolvedURL;
  857. try {
  858. resolvedURL = util.resolveURL(href, this.options.baseURI || this.options.url);
  859. } catch (error) {
  860. // ignored
  861. }
  862. if (resolvedURL) {
  863. const url = normalizeURL(this.options.url);
  864. if (resolvedURL.startsWith(url + "#") && !resolvedURL.startsWith(url + "#!") && !this.options.resolveFragmentIdentifierURLs) {
  865. resolvedURL = resolvedURL.substring(url.length);
  866. }
  867. try {
  868. element.setAttribute("href", resolvedURL);
  869. } catch (error) {
  870. // ignored
  871. }
  872. }
  873. }
  874. });
  875. }
  876. resolveStyleAttributeURLs() {
  877. this.doc.querySelectorAll("[style]").forEach(element => {
  878. let styleContent = element.getAttribute("style");
  879. if (this.options.compressCSS) {
  880. styleContent = util.compressCSS(styleContent);
  881. }
  882. styleContent = ProcessorHelper.resolveStylesheetURLs(styleContent, this.baseURI, this.workStyleElement, this.options.saveOriginalURLs);
  883. const declarationList = cssTree.parse(styleContent, { context: "declarationList" });
  884. this.styles.set(element, declarationList);
  885. });
  886. }
  887. async resolveStylesheetURLs() {
  888. await Promise.all(Array.from(this.doc.querySelectorAll("style, link[rel*=stylesheet]")).map(async element => {
  889. const options = {
  890. maxResourceSize: this.options.maxResourceSize,
  891. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  892. url: this.options.url,
  893. charset: this.charset,
  894. compressCSS: this.options.compressCSS,
  895. updatedResources: this.options.updatedResources,
  896. rootDocument: this.options.rootDocument,
  897. frameId: this.options.windowId,
  898. resourceReferrer: this.options.resourceReferrer,
  899. blockMixedContent: this.options.blockMixedContent,
  900. saveOriginalURLs: this.options.saveOriginalURLs,
  901. acceptHeaders: this.options.acceptHeaders
  902. };
  903. let mediaText;
  904. if (element.media) {
  905. mediaText = element.media.toLowerCase();
  906. }
  907. const stylesheetInfo = { mediaText };
  908. if (element.closest("[" + SHADOW_MODE_ATTRIBUTE_NAME + "]")) {
  909. stylesheetInfo.scoped = true;
  910. }
  911. if (element.tagName == "LINK" && element.charset) {
  912. options.charset = element.charset;
  913. }
  914. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, options, this.workStyleElement);
  915. }));
  916. if (this.options.rootDocument) {
  917. 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]);
  918. await Promise.all(newResources.map(async resource => {
  919. resource.retrieved = true;
  920. const stylesheetInfo = {};
  921. const element = this.doc.createElement("style");
  922. this.doc.body.appendChild(element);
  923. element.textContent = resource.content;
  924. await processElement(element, stylesheetInfo, this.stylesheets, this.baseURI, this.options, this.workStyleElement);
  925. }));
  926. }
  927. async function processElement(element, stylesheetInfo, stylesheets, baseURI, options, workStyleElement) {
  928. stylesheets.set(element, stylesheetInfo);
  929. let stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  930. if (!matchCharsetEquals(stylesheetContent, options.charset)) {
  931. options = Object.assign({}, options, { charset: getCharset(stylesheetContent) });
  932. stylesheetContent = await getStylesheetContent(element, baseURI, options, workStyleElement);
  933. }
  934. let stylesheet;
  935. try {
  936. stylesheet = cssTree.parse(removeCssComments(stylesheetContent));
  937. } catch (error) {
  938. // ignored
  939. }
  940. if (stylesheet && stylesheet.children) {
  941. if (options.compressCSS) {
  942. ProcessorHelper.removeSingleLineCssComments(stylesheet);
  943. }
  944. stylesheetInfo.stylesheet = stylesheet;
  945. } else {
  946. stylesheets.delete(element);
  947. }
  948. }
  949. async function getStylesheetContent(element, baseURI, options, workStyleElement) {
  950. let content;
  951. if (element.tagName == "LINK") {
  952. content = await ProcessorHelper.resolveLinkStylesheetURLs(element.href, baseURI, options, workStyleElement);
  953. } else {
  954. content = await ProcessorHelper.resolveImportURLs(element.textContent, baseURI, options, workStyleElement);
  955. }
  956. return content || "";
  957. }
  958. }
  959. async resolveFrameURLs() {
  960. if (!this.options.saveRawPage) {
  961. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  962. await Promise.all(frameElements.map(async frameElement => {
  963. if (frameElement.tagName == "OBJECT") {
  964. frameElement.setAttribute("data", "data:text/html,");
  965. } else {
  966. const src = frameElement.getAttribute("src");
  967. if (this.options.saveOriginalURLs && !isDataURL(src)) {
  968. frameElement.setAttribute("data-sf-original-src", src);
  969. }
  970. frameElement.removeAttribute("src");
  971. frameElement.removeAttribute("srcdoc");
  972. }
  973. Array.from(frameElement.childNodes).forEach(node => node.remove());
  974. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  975. if (this.options.frames && frameWindowId) {
  976. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  977. if (frameData) {
  978. await initializeProcessor(frameData, frameElement, frameWindowId, this.batchRequest, Object.create(this.options));
  979. }
  980. }
  981. }));
  982. }
  983. async function initializeProcessor(frameData, frameElement, frameWindowId, batchRequest, options) {
  984. options.insertSingleFileComment = false;
  985. options.insertCanonicalLink = false;
  986. options.insertMetaNoIndex = false;
  987. options.saveFavicon = false;
  988. options.url = frameData.baseURI;
  989. options.windowId = frameWindowId;
  990. if (frameData.content) {
  991. options.content = frameData.content;
  992. options.canvases = frameData.canvases;
  993. options.fonts = frameData.fonts;
  994. options.stylesheets = frameData.stylesheets;
  995. options.images = frameData.images;
  996. options.posters = frameData.posters;
  997. options.usedFonts = frameData.usedFonts;
  998. options.shadowRoots = frameData.shadowRoots;
  999. options.imports = frameData.imports;
  1000. frameData.runner = new Runner(options);
  1001. frameData.frameElement = frameElement;
  1002. await frameData.runner.loadPage();
  1003. await frameData.runner.initialize();
  1004. frameData.maxResources = batchRequest.getMaxResources();
  1005. }
  1006. }
  1007. }
  1008. insertShadowRootContents() {
  1009. const doc = this.doc;
  1010. const options = this.options;
  1011. if (this.options.shadowRoots && this.options.shadowRoots.length) {
  1012. processElement(this.doc);
  1013. if (this.options.removeScripts) {
  1014. this.doc.querySelectorAll("script[" + SCRIPT_TEMPLATE_SHADOW_ROOT + "]").forEach(element => element.remove());
  1015. }
  1016. const scriptElement = doc.createElement("script");
  1017. scriptElement.setAttribute(SCRIPT_TEMPLATE_SHADOW_ROOT, "");
  1018. 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)}}})}})()`;
  1019. doc.body.appendChild(scriptElement);
  1020. }
  1021. function processElement(element) {
  1022. const shadowRootElements = Array.from((element.querySelectorAll("[" + util.SHADOW_ROOT_ATTRIBUTE_NAME + "]")));
  1023. shadowRootElements.forEach(element => {
  1024. const attributeValue = element.getAttribute(util.SHADOW_ROOT_ATTRIBUTE_NAME);
  1025. if (attributeValue) {
  1026. const shadowRootData = options.shadowRoots[Number(attributeValue)];
  1027. if (shadowRootData) {
  1028. const templateElement = doc.createElement("template");
  1029. templateElement.setAttribute(SHADOW_MODE_ATTRIBUTE_NAME, shadowRootData.mode);
  1030. if (shadowRootData.delegatesFocus) {
  1031. templateElement.setAttribute(SHADOW_DELEGATE_FOCUS_ATTRIBUTE_NAME, "");
  1032. }
  1033. if (shadowRootData.adoptedStyleSheets) {
  1034. shadowRootData.adoptedStyleSheets.forEach(stylesheetContent => {
  1035. const styleElement = doc.createElement("style");
  1036. styleElement.textContent = stylesheetContent;
  1037. templateElement.appendChild(styleElement);
  1038. });
  1039. }
  1040. const shadowDoc = util.parseDocContent(shadowRootData.content);
  1041. if (shadowDoc.head) {
  1042. const metaCharset = shadowDoc.head.querySelector("meta[charset]");
  1043. if (metaCharset) {
  1044. metaCharset.remove();
  1045. }
  1046. shadowDoc.head.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1047. }
  1048. if (shadowDoc.body) {
  1049. shadowDoc.body.childNodes.forEach(node => templateElement.appendChild(shadowDoc.importNode(node, true)));
  1050. }
  1051. processElement(templateElement);
  1052. if (element.firstChild) {
  1053. element.insertBefore(templateElement, element.firstChild);
  1054. } else {
  1055. element.appendChild(templateElement);
  1056. }
  1057. }
  1058. }
  1059. });
  1060. }
  1061. }
  1062. async resolveHtmlImportURLs() {
  1063. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  1064. await Promise.all(linkElements.map(async linkElement => {
  1065. const resourceURL = linkElement.href;
  1066. if (this.options.saveOriginalURLs && !isDataURL(resourceURL)) {
  1067. linkElement.setAttribute("data-sf-original-href", resourceURL);
  1068. }
  1069. linkElement.removeAttribute("href");
  1070. const options = Object.create(this.options);
  1071. options.insertSingleFileComment = false;
  1072. options.insertCanonicalLink = false;
  1073. options.insertMetaNoIndex = false;
  1074. options.saveFavicon = false;
  1075. options.removeUnusedStyles = false;
  1076. options.removeAlternativeMedias = false;
  1077. options.removeUnusedFonts = false;
  1078. options.removeHiddenElements = false;
  1079. options.url = resourceURL;
  1080. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1081. if (attributeValue) {
  1082. const importData = options.imports[Number(attributeValue)];
  1083. if (importData) {
  1084. options.content = importData.content;
  1085. importData.runner = new Runner(options);
  1086. await importData.runner.loadPage();
  1087. await importData.runner.initialize();
  1088. if (!options.removeImports) {
  1089. importData.maxResources = importData.runner.batchRequest.getMaxResources();
  1090. }
  1091. importData.runner.getStyleSheets().forEach(stylesheet => {
  1092. const importedStyleElement = this.doc.createElement("style");
  1093. linkElement.insertAdjacentElement("afterEnd", importedStyleElement);
  1094. this.stylesheets.set(importedStyleElement, stylesheet);
  1095. });
  1096. }
  1097. }
  1098. if (options.removeImports) {
  1099. linkElement.remove();
  1100. this.stats.add("discarded", "HTML imports", 1);
  1101. }
  1102. }));
  1103. }
  1104. removeUnusedStyles() {
  1105. if (!this.mediaAllInfo) {
  1106. this.mediaAllInfo = util.getMediaAllInfo(this.doc, this.stylesheets, this.styles);
  1107. }
  1108. const stats = util.minifyCSSRules(this.stylesheets, this.styles, this.mediaAllInfo);
  1109. this.stats.set("processed", "CSS rules", stats.processed);
  1110. this.stats.set("discarded", "CSS rules", stats.discarded);
  1111. }
  1112. removeUnusedFonts() {
  1113. util.removeUnusedFonts(this.doc, this.stylesheets, this.styles, this.options);
  1114. }
  1115. removeAlternativeMedias() {
  1116. const stats = util.minifyMedias(this.stylesheets);
  1117. this.stats.set("processed", "medias", stats.processed);
  1118. this.stats.set("discarded", "medias", stats.discarded);
  1119. }
  1120. async processStylesheets() {
  1121. this.options.fontURLs = new Map();
  1122. await Promise.all([...this.stylesheets].map(([, stylesheetInfo]) =>
  1123. ProcessorHelper.processStylesheet(stylesheetInfo.stylesheet.children, this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1124. ));
  1125. }
  1126. async processStyleAttributes() {
  1127. return Promise.all([...this.styles].map(([, declarationList]) =>
  1128. ProcessorHelper.processStyle(declarationList.children.toArray(), this.baseURI, this.options, this.cssVariables, this.batchRequest)
  1129. ));
  1130. }
  1131. async processPageResources() {
  1132. const processAttributeArgs = [
  1133. ["link[href][rel*=\"icon\"]", "href", false, true],
  1134. ["object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]", "data"],
  1135. ["img[src], input[src][type=image]", "src", true],
  1136. ["embed[src*=\".svg\"], embed[src*=\".pdf\"]", "src"],
  1137. ["video[poster]", "poster"],
  1138. ["*[background]", "background"],
  1139. ["image", "xlink:href"],
  1140. ["image", "href"]
  1141. ];
  1142. let resourcePromises = processAttributeArgs.map(([selector, attributeName, processDuplicates, removeElementIfMissing]) =>
  1143. ProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), attributeName, this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest, processDuplicates, removeElementIfMissing)
  1144. );
  1145. resourcePromises = resourcePromises.concat([
  1146. ProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.doc, this.baseURI, this.options, this.batchRequest),
  1147. ProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", this.baseURI, this.batchRequest)
  1148. ]);
  1149. if (!this.options.removeAudioSrc) {
  1150. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1151. }
  1152. if (!this.options.removeVideoSrc) {
  1153. resourcePromises.push(ProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", this.baseURI, this.options, this.cssVariables, this.styles, this.batchRequest));
  1154. }
  1155. await Promise.all(resourcePromises);
  1156. if (this.options.saveFavicon) {
  1157. ProcessorHelper.processShortcutIcons(this.doc);
  1158. }
  1159. }
  1160. async processScripts() {
  1161. await Promise.all(Array.from(this.doc.querySelectorAll("script[src], link[rel=preload][as=script][href]")).map(async element => {
  1162. let resourceURL;
  1163. let scriptSrc;
  1164. if (element.tagName == "SCRIPT") {
  1165. scriptSrc = element.getAttribute("src");
  1166. if (this.options.saveOriginalURLs && !isDataURL(scriptSrc)) {
  1167. element.setAttribute("data-sf-original-src", scriptSrc);
  1168. }
  1169. } else {
  1170. scriptSrc = element.getAttribute("href");
  1171. if (this.options.saveOriginalURLs && !isDataURL(scriptSrc)) {
  1172. element.setAttribute("data-sf-original-href", scriptSrc);
  1173. }
  1174. }
  1175. element.removeAttribute("integrity");
  1176. element.textContent = "";
  1177. try {
  1178. resourceURL = util.resolveURL(scriptSrc, this.baseURI);
  1179. } catch (error) {
  1180. // ignored
  1181. }
  1182. if (testValidURL(resourceURL)) {
  1183. if (element.tagName == "SCRIPT") {
  1184. element.removeAttribute("src");
  1185. } else {
  1186. element.removeAttribute("href");
  1187. }
  1188. this.stats.add("processed", "scripts", 1);
  1189. const content = await util.getContent(resourceURL, {
  1190. asBinary: true,
  1191. charset: this.charset != UTF8_CHARSET && this.charset,
  1192. maxResourceSize: this.options.maxResourceSize,
  1193. maxResourceSizeEnabled: this.options.maxResourceSizeEnabled,
  1194. frameId: this.options.windowId,
  1195. resourceReferrer: this.options.resourceReferrer,
  1196. baseURI: this.options.baseURI,
  1197. blockMixedContent: this.options.blockMixedContent,
  1198. expectedType: "script",
  1199. acceptHeaders: this.options.acceptHeaders
  1200. });
  1201. content.data = getUpdatedResourceContent(resourceURL, content, this.options);
  1202. if (element.tagName == "SCRIPT") {
  1203. element.setAttribute("src", content.data);
  1204. if (element.getAttribute("async") == "" || element.getAttribute("async") == "async" || element.getAttribute(util.ASYNC_SCRIPT_ATTRIBUTE_NAME) == "") {
  1205. element.setAttribute("async", "");
  1206. }
  1207. } else {
  1208. const scriptElement = this.doc.createElement("script");
  1209. scriptElement.setAttribute("src", content.data);
  1210. scriptElement.setAttribute("async", "");
  1211. element.parentElement.replaceChild(scriptElement, element);
  1212. }
  1213. }
  1214. }));
  1215. }
  1216. removeAlternativeImages() {
  1217. util.removeAlternativeImages(this.doc);
  1218. }
  1219. async removeAlternativeFonts() {
  1220. await util.removeAlternativeFonts(this.doc, this.stylesheets, this.options.fontURLs, this.options.fontTests);
  1221. }
  1222. async processFrames() {
  1223. if (this.options.frames) {
  1224. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  1225. await Promise.all(frameElements.map(async frameElement => {
  1226. const frameWindowId = frameElement.getAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1227. if (frameWindowId) {
  1228. const frameData = this.options.frames.find(frame => frame.windowId == frameWindowId);
  1229. if (frameData) {
  1230. this.options.frames = this.options.frames.filter(frame => frame.windowId != frameWindowId);
  1231. if (frameData.runner && frameElement.getAttribute(util.HIDDEN_FRAME_ATTRIBUTE_NAME) != "") {
  1232. this.stats.add("processed", "frames", 1);
  1233. await frameData.runner.run();
  1234. const pageData = await frameData.runner.getPageData();
  1235. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1236. let sandbox = "allow-popups allow-top-navigation allow-top-navigation-by-user-activation";
  1237. if (pageData.content.match(NOSCRIPT_TAG_FOUND) || pageData.content.match(CANVAS_TAG_FOUND) || pageData.content.match(SCRIPT_TAG_FOUND)) {
  1238. sandbox += " allow-scripts allow-same-origin";
  1239. }
  1240. frameElement.setAttribute("sandbox", sandbox);
  1241. if (frameElement.tagName == "OBJECT") {
  1242. frameElement.setAttribute("data", "data:text/html," + pageData.content);
  1243. } else {
  1244. if (frameElement.tagName == "FRAME") {
  1245. frameElement.setAttribute("src", "data:text/html," + pageData.content.replace(/%/g, "%25").replace(/#/g, "%23"));
  1246. } else {
  1247. frameElement.setAttribute("srcdoc", pageData.content);
  1248. frameElement.removeAttribute("src");
  1249. }
  1250. }
  1251. this.stats.addAll(pageData);
  1252. } else {
  1253. frameElement.removeAttribute(util.WIN_ID_ATTRIBUTE_NAME);
  1254. this.stats.add("discarded", "frames", 1);
  1255. }
  1256. }
  1257. }
  1258. }));
  1259. }
  1260. }
  1261. async processHtmlImports() {
  1262. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import]"));
  1263. await Promise.all(linkElements.map(async linkElement => {
  1264. const attributeValue = linkElement.getAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1265. if (attributeValue) {
  1266. const importData = this.options.imports[Number(attributeValue)];
  1267. if (importData.runner) {
  1268. this.stats.add("processed", "HTML imports", 1);
  1269. await importData.runner.run();
  1270. const pageData = await importData.runner.getPageData();
  1271. linkElement.removeAttribute(util.HTML_IMPORT_ATTRIBUTE_NAME);
  1272. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  1273. this.stats.addAll(pageData);
  1274. } else {
  1275. this.stats.add("discarded", "HTML imports", 1);
  1276. }
  1277. }
  1278. }));
  1279. }
  1280. replaceStylesheets() {
  1281. this.doc.querySelectorAll("style").forEach(styleElement => {
  1282. const stylesheetInfo = this.stylesheets.get(styleElement);
  1283. if (stylesheetInfo) {
  1284. this.stylesheets.delete(styleElement);
  1285. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1286. if (this.options.saveOriginalURLs) {
  1287. stylesheetContent = replaceOriginalURLs(stylesheetContent);
  1288. }
  1289. styleElement.textContent = stylesheetContent;
  1290. if (stylesheetInfo.mediaText) {
  1291. styleElement.media = stylesheetInfo.mediaText;
  1292. }
  1293. } else {
  1294. styleElement.remove();
  1295. }
  1296. });
  1297. this.doc.querySelectorAll("link[rel*=stylesheet]").forEach(linkElement => {
  1298. const stylesheetInfo = this.stylesheets.get(linkElement);
  1299. if (stylesheetInfo) {
  1300. this.stylesheets.delete(linkElement);
  1301. const styleElement = this.doc.createElement("style");
  1302. if (stylesheetInfo.mediaText) {
  1303. styleElement.media = stylesheetInfo.mediaText;
  1304. }
  1305. let stylesheetContent = cssTree.generate(stylesheetInfo.stylesheet);
  1306. if (this.options.saveOriginalURLs) {
  1307. stylesheetContent = replaceOriginalURLs(stylesheetContent);
  1308. styleElement.setAttribute("data-sf-original-href", linkElement.getAttribute("data-sf-original-href"));
  1309. }
  1310. styleElement.textContent = stylesheetContent;
  1311. linkElement.parentElement.replaceChild(styleElement, linkElement);
  1312. } else {
  1313. linkElement.remove();
  1314. }
  1315. });
  1316. }
  1317. replaceStyleAttributes() {
  1318. this.doc.querySelectorAll("[style]").forEach(element => {
  1319. const declarations = this.styles.get(element);
  1320. if (declarations) {
  1321. this.styles.delete(element);
  1322. let styleContent = cssTree.generate(declarations);
  1323. if (this.options.saveOriginalURLs) {
  1324. styleContent = replaceOriginalURLs(styleContent);
  1325. }
  1326. element.setAttribute("style", styleContent);
  1327. } else {
  1328. element.setAttribute("style", "");
  1329. }
  1330. });
  1331. }
  1332. insertVariables() {
  1333. if (this.cssVariables.size) {
  1334. const styleElement = this.doc.createElement("style");
  1335. const firstStyleElement = this.doc.head.querySelector("style");
  1336. if (firstStyleElement) {
  1337. this.doc.head.insertBefore(styleElement, firstStyleElement);
  1338. } else {
  1339. this.doc.head.appendChild(styleElement);
  1340. }
  1341. let stylesheetContent = "";
  1342. this.cssVariables.forEach(({ content, url }, indexResource) => {
  1343. this.cssVariables.delete(indexResource);
  1344. if (stylesheetContent) {
  1345. stylesheetContent += ";";
  1346. }
  1347. stylesheetContent += `${SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource}: `;
  1348. if (this.options.saveOriginalURLs) {
  1349. stylesheetContent += `/* original URL: ${url} */ `;
  1350. }
  1351. stylesheetContent += `url("${content}")`;
  1352. });
  1353. styleElement.textContent = ":root{" + stylesheetContent + "}";
  1354. }
  1355. }
  1356. compressHTML() {
  1357. let size;
  1358. if (this.options.displayStats) {
  1359. size = util.getContentSize(this.doc.documentElement.outerHTML);
  1360. }
  1361. util.minifyHTML(this.doc, { PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME: util.PRESERVED_SPACE_ELEMENT_ATTRIBUTE_NAME });
  1362. if (this.options.displayStats) {
  1363. this.stats.add("discarded", "HTML bytes", size - util.getContentSize(this.doc.documentElement.outerHTML));
  1364. }
  1365. }
  1366. cleanupPage() {
  1367. this.doc.querySelectorAll("base").forEach(element => element.remove());
  1368. const metaCharset = this.doc.head.querySelector("meta[charset]");
  1369. if (metaCharset) {
  1370. this.doc.head.insertBefore(metaCharset, this.doc.head.firstChild);
  1371. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.body.childNodes.length == 0) {
  1372. this.doc.head.querySelector("meta[charset]").remove();
  1373. }
  1374. }
  1375. }
  1376. resetZoomLevel() {
  1377. const transform = this.doc.documentElement.style.getPropertyValue("-sf-transform");
  1378. const transformPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform");
  1379. const transformOrigin = this.doc.documentElement.style.getPropertyValue("-sf-transform-origin");
  1380. const transformOriginPriority = this.doc.documentElement.style.getPropertyPriority("-sf-transform-origin");
  1381. const minHeight = this.doc.documentElement.style.getPropertyValue("-sf-min-height");
  1382. const minHeightPriority = this.doc.documentElement.style.getPropertyPriority("-sf-min-height");
  1383. this.doc.documentElement.style.setProperty("transform", transform, transformPriority);
  1384. this.doc.documentElement.style.setProperty("transform-origin", transformOrigin, transformOriginPriority);
  1385. this.doc.documentElement.style.setProperty("min-height", minHeight, minHeightPriority);
  1386. this.doc.documentElement.style.removeProperty("-sf-transform");
  1387. this.doc.documentElement.style.removeProperty("-sf-transform-origin");
  1388. this.doc.documentElement.style.removeProperty("-sf-min-height");
  1389. }
  1390. async insertMAFFMetaData() {
  1391. const maffMetaData = await this.maffMetaDataPromise;
  1392. if (maffMetaData && maffMetaData.content) {
  1393. const NAMESPACE_RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#";
  1394. const maffDoc = util.parseXMLContent(maffMetaData.content);
  1395. const originalURLElement = maffDoc.querySelector("RDF > Description > originalurl");
  1396. const archiveTimeElement = maffDoc.querySelector("RDF > Description > archivetime");
  1397. if (originalURLElement) {
  1398. this.options.saveUrl = originalURLElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1399. }
  1400. if (archiveTimeElement) {
  1401. const value = archiveTimeElement.getAttributeNS(NAMESPACE_RDF, "resource");
  1402. if (value) {
  1403. const date = new Date(value);
  1404. if (!isNaN(date.getTime())) {
  1405. this.options.saveDate = new Date(value);
  1406. }
  1407. }
  1408. }
  1409. }
  1410. }
  1411. async setDocInfo() {
  1412. const titleElement = this.doc.querySelector("title");
  1413. const descriptionElement = this.doc.querySelector("meta[name=description]");
  1414. const authorElement = this.doc.querySelector("meta[name=author]");
  1415. const creatorElement = this.doc.querySelector("meta[name=creator]");
  1416. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  1417. const headingElement = this.doc.querySelector("h1");
  1418. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  1419. this.options.info = {
  1420. description: descriptionElement && descriptionElement.content ? descriptionElement.content.trim() : "",
  1421. lang: this.doc.documentElement.lang,
  1422. author: authorElement && authorElement.content ? authorElement.content.trim() : "",
  1423. creator: creatorElement && creatorElement.content ? creatorElement.content.trim() : "",
  1424. publisher: publisherElement && publisherElement.content ? publisherElement.content.trim() : "",
  1425. heading: headingElement && headingElement.textContent ? headingElement.textContent.trim() : ""
  1426. };
  1427. this.options.infobarContent = await ProcessorHelper.evalTemplate(this.options.infobarTemplate, this.options, null, true);
  1428. }
  1429. }
  1430. // ---------------
  1431. // ProcessorHelper
  1432. // ---------------
  1433. const DATA_URI_PREFIX = "data:";
  1434. const ABOUT_BLANK_URI = "about:blank";
  1435. const EMPTY_DATA_URI = "data:null;base64,";
  1436. const REGEXP_URL_HASH = /(#.+?)$/;
  1437. const SINGLE_FILE_VARIABLE_NAME_PREFIX = "--sf-img-";
  1438. const SINGLE_FILE_VARIABLE_MAX_SIZE = 512 * 1024;
  1439. class ProcessorHelper {
  1440. static async evalTemplate(template = "", options, content, dontReplaceSlash) {
  1441. const url = util.parseURL(options.saveUrl);
  1442. template = await evalTemplateVariable(template, "page-title", () => options.title || "No title", dontReplaceSlash, options.filenameReplacementCharacter);
  1443. template = await evalTemplateVariable(template, "page-heading", () => options.info.heading || "No heading", dontReplaceSlash, options.filenameReplacementCharacter);
  1444. template = await evalTemplateVariable(template, "page-language", () => options.info.lang || "No language", dontReplaceSlash, options.filenameReplacementCharacter);
  1445. template = await evalTemplateVariable(template, "page-description", () => options.info.description || "No description", dontReplaceSlash, options.filenameReplacementCharacter);
  1446. template = await evalTemplateVariable(template, "page-author", () => options.info.author || "No author", dontReplaceSlash, options.filenameReplacementCharacter);
  1447. template = await evalTemplateVariable(template, "page-creator", () => options.info.creator || "No creator", dontReplaceSlash, options.filenameReplacementCharacter);
  1448. template = await evalTemplateVariable(template, "page-publisher", () => options.info.publisher || "No publisher", dontReplaceSlash, options.filenameReplacementCharacter);
  1449. await evalDate(options.saveDate);
  1450. await evalDate(options.visitDate, "visit-");
  1451. template = await evalTemplateVariable(template, "url-hash", () => url.hash.substring(1) || "No hash", dontReplaceSlash, options.filenameReplacementCharacter);
  1452. template = await evalTemplateVariable(template, "url-host", () => url.host.replace(/\/$/, "") || "No host", dontReplaceSlash, options.filenameReplacementCharacter);
  1453. template = await evalTemplateVariable(template, "url-hostname", () => url.hostname.replace(/\/$/, "") || "No hostname", dontReplaceSlash, options.filenameReplacementCharacter);
  1454. const urlHref = decode(url.href);
  1455. template = await evalTemplateVariable(template, "url-href", () => urlHref || "No href", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1456. template = await evalTemplateVariable(template, "url-href-digest-sha-1", urlHref ? async () => util.digest("SHA-1", urlHref) : "No href", dontReplaceSlash, options.filenameReplacementCharacter);
  1457. template = await evalTemplateVariable(template, "url-href-flat", () => decode(url.href) || "No href", false, options.filenameReplacementCharacter);
  1458. template = await evalTemplateVariable(template, "url-referrer", () => decode(options.referrer) || "No referrer", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1459. template = await evalTemplateVariable(template, "url-referrer-flat", () => decode(options.referrer) || "No referrer", false, options.filenameReplacementCharacter);
  1460. template = await evalTemplateVariable(template, "url-password", () => url.password || "No password", dontReplaceSlash, options.filenameReplacementCharacter);
  1461. template = await evalTemplateVariable(template, "url-pathname", () => decode(url.pathname).replace(/^\//, "").replace(/\/$/, "") || "No pathname", dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1462. template = await evalTemplateVariable(template, "url-pathname-flat", () => decode(url.pathname) || "No pathname", false, options.filenameReplacementCharacter);
  1463. template = await evalTemplateVariable(template, "url-port", () => url.port || "No port", dontReplaceSlash, options.filenameReplacementCharacter);
  1464. template = await evalTemplateVariable(template, "url-protocol", () => url.protocol || "No protocol", dontReplaceSlash, options.filenameReplacementCharacter);
  1465. template = await evalTemplateVariable(template, "url-search", () => url.search.substring(1) || "No search", dontReplaceSlash, options.filenameReplacementCharacter);
  1466. const params = url.search.substring(1).split("&").map(parameter => parameter.split("="));
  1467. for (const [name, value] of params) {
  1468. template = await evalTemplateVariable(template, "url-search-" + name, () => value || "", dontReplaceSlash, options.filenameReplacementCharacter);
  1469. }
  1470. template = template.replace(/{\s*url-search-[^}\s]*\s*}/gi, "");
  1471. template = await evalTemplateVariable(template, "url-username", () => url.username || "No username", dontReplaceSlash, options.filenameReplacementCharacter);
  1472. template = await evalTemplateVariable(template, "tab-id", () => String(options.tabId || "No tab id"), dontReplaceSlash, options.filenameReplacementCharacter);
  1473. template = await evalTemplateVariable(template, "tab-index", () => String(options.tabIndex || "No tab index"), dontReplaceSlash, options.filenameReplacementCharacter);
  1474. template = await evalTemplateVariable(template, "url-last-segment", () => decode(getLastSegment(url, options.filenameReplacementCharacter)) || "No last segment", dontReplaceSlash, options.filenameReplacementCharacter);
  1475. if (content) {
  1476. template = await evalTemplateVariable(template, "digest-sha-256", async () => util.digest("SHA-256", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1477. template = await evalTemplateVariable(template, "digest-sha-384", async () => util.digest("SHA-384", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1478. template = await evalTemplateVariable(template, "digest-sha-512", async () => util.digest("SHA-512", content), dontReplaceSlash, options.filenameReplacementCharacter);
  1479. }
  1480. const bookmarkFolder = (options.bookmarkFolders && options.bookmarkFolders.join("/")) || "";
  1481. template = await evalTemplateVariable(template, "bookmark-pathname", () => bookmarkFolder, dontReplaceSlash === undefined ? true : dontReplaceSlash, options.filenameReplacementCharacter);
  1482. template = await evalTemplateVariable(template, "bookmark-pathname-flat", () => bookmarkFolder, false, options.filenameReplacementCharacter);
  1483. template = await evalTemplateVariable(template, "profile-name", () => options.profileName, dontReplaceSlash, options.filenameReplacementCharacter);
  1484. return template.trim();
  1485. function decode(value) {
  1486. try {
  1487. return decodeURI(value);
  1488. } catch (error) {
  1489. return value;
  1490. }
  1491. }
  1492. async function evalDate(date, prefix = "") {
  1493. if (date) {
  1494. template = await evalTemplateVariable(template, prefix + "datetime-iso", () => date.toISOString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1495. template = await evalTemplateVariable(template, prefix + "date-iso", () => date.toISOString().split("T")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1496. template = await evalTemplateVariable(template, prefix + "time-iso", () => date.toISOString().split("T")[1].split("Z")[0], dontReplaceSlash, options.filenameReplacementCharacter);
  1497. template = await evalTemplateVariable(template, prefix + "date-locale", () => date.toLocaleDateString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1498. template = await evalTemplateVariable(template, prefix + "time-locale", () => date.toLocaleTimeString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1499. template = await evalTemplateVariable(template, prefix + "day-locale", () => String(date.getDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1500. template = await evalTemplateVariable(template, prefix + "month-locale", () => String(date.getMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1501. template = await evalTemplateVariable(template, prefix + "year-locale", () => String(date.getFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1502. template = await evalTemplateVariable(template, prefix + "datetime-locale", () => date.toLocaleString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1503. template = await evalTemplateVariable(template, prefix + "datetime-utc", () => date.toUTCString(), dontReplaceSlash, options.filenameReplacementCharacter);
  1504. template = await evalTemplateVariable(template, prefix + "day-utc", () => String(date.getUTCDate()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1505. template = await evalTemplateVariable(template, prefix + "month-utc", () => String(date.getUTCMonth() + 1).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1506. template = await evalTemplateVariable(template, prefix + "year-utc", () => String(date.getUTCFullYear()), dontReplaceSlash, options.filenameReplacementCharacter);
  1507. template = await evalTemplateVariable(template, prefix + "hours-locale", () => String(date.getHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1508. template = await evalTemplateVariable(template, prefix + "minutes-locale", () => String(date.getMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1509. template = await evalTemplateVariable(template, prefix + "seconds-locale", () => String(date.getSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1510. template = await evalTemplateVariable(template, prefix + "hours-utc", () => String(date.getUTCHours()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1511. template = await evalTemplateVariable(template, prefix + "minutes-utc", () => String(date.getUTCMinutes()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1512. template = await evalTemplateVariable(template, prefix + "seconds-utc", () => String(date.getUTCSeconds()).padStart(2, "0"), dontReplaceSlash, options.filenameReplacementCharacter);
  1513. template = await evalTemplateVariable(template, prefix + "time-ms", () => String(date.getTime()), dontReplaceSlash, options.filenameReplacementCharacter);
  1514. }
  1515. }
  1516. }
  1517. static setBackgroundImage(element, url, style) {
  1518. element.style.setProperty("background-blend-mode", "normal", "important");
  1519. element.style.setProperty("background-clip", "content-box", "important");
  1520. element.style.setProperty("background-position", style && style["background-position"] ? style["background-position"] : "center", "important");
  1521. element.style.setProperty("background-color", style && style["background-color"] ? style["background-color"] : "transparent", "important");
  1522. element.style.setProperty("background-image", url, "important");
  1523. element.style.setProperty("background-size", style && style["background-size"] ? style["background-size"] : "100% 100%", "important");
  1524. element.style.setProperty("background-origin", "content-box", "important");
  1525. element.style.setProperty("background-repeat", "no-repeat", "important");
  1526. }
  1527. static processShortcutIcons(doc) {
  1528. let shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel=\"icon\"], link[href][rel=\"shortcut icon\"]")));
  1529. if (!shortcutIcon) {
  1530. shortcutIcon = findShortcutIcon(Array.from(doc.querySelectorAll("link[href][rel*=\"icon\"]")));
  1531. if (shortcutIcon) {
  1532. shortcutIcon.rel = "icon";
  1533. }
  1534. }
  1535. if (shortcutIcon) {
  1536. doc.querySelectorAll("link[href][rel*=\"icon\"]").forEach(linkElement => {
  1537. if (linkElement != shortcutIcon) {
  1538. linkElement.remove();
  1539. }
  1540. });
  1541. }
  1542. }
  1543. static removeSingleLineCssComments(stylesheet) {
  1544. const removedRules = [];
  1545. for (let cssRule = stylesheet.children.head; cssRule; cssRule = cssRule.next) {
  1546. const ruleData = cssRule.data;
  1547. if (ruleData.type == "Raw" && ruleData.value && ruleData.value.trim().startsWith("//")) {
  1548. removedRules.push(cssRule);
  1549. }
  1550. }
  1551. removedRules.forEach(cssRule => stylesheet.children.remove(cssRule));
  1552. }
  1553. static async resolveImportURLs(stylesheetContent, baseURI, options, workStylesheet, importedStyleSheets = new Set()) {
  1554. stylesheetContent = ProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet, options.saveOriginalURLs);
  1555. const imports = getImportFunctions(stylesheetContent);
  1556. await Promise.all(imports.map(async cssImport => {
  1557. const match = matchImport(cssImport);
  1558. if (match) {
  1559. const regExpCssImport = getRegExp(cssImport);
  1560. let resourceURL = normalizeURL(match.resourceURL);
  1561. if (!testIgnoredPath(resourceURL) && testValidPath(resourceURL)) {
  1562. try {
  1563. resourceURL = util.resolveURL(match.resourceURL, baseURI);
  1564. } catch (error) {
  1565. // ignored
  1566. }
  1567. if (testValidURL(resourceURL) && !importedStyleSheets.has(resourceURL)) {
  1568. const content = await getStylesheetContent(resourceURL);
  1569. resourceURL = content.resourceURL;
  1570. content.data = getUpdatedResourceContent(resourceURL, content, options);
  1571. if (content.data && content.data.match(/^<!doctype /i)) {
  1572. content.data = "";
  1573. }
  1574. let importedStylesheetContent = removeCssComments(content.data);
  1575. if (options.compressCSS) {
  1576. importedStylesheetContent = util.compressCSS(importedStylesheetContent);
  1577. }
  1578. importedStylesheetContent = wrapMediaQuery(importedStylesheetContent, match.media);
  1579. if (stylesheetContent.includes(cssImport)) {
  1580. const ancestorStyleSheets = new Set(importedStyleSheets);
  1581. ancestorStyleSheets.add(resourceURL);
  1582. importedStylesheetContent = await ProcessorHelper.resolveImportURLs(importedStylesheetContent, resourceURL, options, workStylesheet, ancestorStyleSheets);
  1583. workStylesheet.textContent = importedStylesheetContent;
  1584. if ((workStylesheet.sheet && workStylesheet.sheet.cssRules.length) || (!workStylesheet.sheet && importedStylesheetContent)) {
  1585. stylesheetContent = stylesheetContent.replace(regExpCssImport, importedStylesheetContent);
  1586. } else {
  1587. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1588. }
  1589. }
  1590. } else {
  1591. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1592. }
  1593. } else {
  1594. stylesheetContent = stylesheetContent.replace(regExpCssImport, "");
  1595. }
  1596. }
  1597. }));
  1598. return stylesheetContent;
  1599. async function getStylesheetContent(resourceURL) {
  1600. const content = await util.getContent(resourceURL, {
  1601. maxResourceSize: options.maxResourceSize,
  1602. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1603. validateTextContentType: true,
  1604. frameId: options.frameId,
  1605. charset: options.charset,
  1606. resourceReferrer: options.resourceReferrer,
  1607. baseURI: options.baseURI,
  1608. blockMixedContent: options.blockMixedContent,
  1609. expectedType: "stylesheet",
  1610. acceptHeaders: options.acceptHeaders
  1611. });
  1612. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1613. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1614. return util.getContent(resourceURL, {
  1615. maxResourceSize: options.maxResourceSize,
  1616. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1617. validateTextContentType: true,
  1618. frameId: options.frameId,
  1619. charset: options.charset,
  1620. resourceReferrer: options.resourceReferrer,
  1621. baseURI: options.baseURI,
  1622. blockMixedContent: options.blockMixedContent,
  1623. expectedType: "stylesheet",
  1624. acceptHeaders: options.acceptHeaders
  1625. });
  1626. } else {
  1627. return content;
  1628. }
  1629. }
  1630. }
  1631. static resolveStylesheetURLs(stylesheetContent, baseURI, workStylesheet, saveOriginalURLs) {
  1632. const urlFunctions = getUrlFunctions(stylesheetContent, true);
  1633. if (saveOriginalURLs) {
  1634. stylesheetContent = addOriginalURLs(stylesheetContent);
  1635. }
  1636. urlFunctions.map(urlFunction => {
  1637. const originalResourceURL = matchURL(urlFunction);
  1638. let resourceURL = normalizeURL(originalResourceURL);
  1639. workStylesheet.textContent = "tmp { content:\"" + resourceURL + "\"}";
  1640. if (workStylesheet.sheet && workStylesheet.sheet.cssRules) {
  1641. resourceURL = util.removeQuotes(workStylesheet.sheet.cssRules[0].style.getPropertyValue("content"));
  1642. }
  1643. if (!testIgnoredPath(resourceURL)) {
  1644. if (!resourceURL || testValidPath(resourceURL)) {
  1645. let resolvedURL;
  1646. if (!originalResourceURL.startsWith("#")) {
  1647. try {
  1648. resolvedURL = util.resolveURL(resourceURL, baseURI);
  1649. } catch (error) {
  1650. // ignored
  1651. }
  1652. }
  1653. if (testValidURL(resolvedURL) && originalResourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  1654. try {
  1655. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), originalResourceURL ? urlFunction.replace(originalResourceURL, resolvedURL) : "url(" + resolvedURL + ")");
  1656. } catch (error) {
  1657. // ignored
  1658. }
  1659. }
  1660. } else {
  1661. let newUrlFunction;
  1662. if (originalResourceURL) {
  1663. newUrlFunction = urlFunction.replace(originalResourceURL, EMPTY_DATA_URI);
  1664. } else {
  1665. newUrlFunction = "url(" + EMPTY_DATA_URI + ")";
  1666. }
  1667. stylesheetContent = stylesheetContent.replace(getRegExp(urlFunction), newUrlFunction);
  1668. }
  1669. }
  1670. });
  1671. return stylesheetContent;
  1672. }
  1673. static async resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet) {
  1674. resourceURL = normalizeURL(resourceURL);
  1675. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  1676. const content = await util.getContent(resourceURL, {
  1677. maxResourceSize: options.maxResourceSize,
  1678. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1679. charset: options.charset,
  1680. frameId: options.frameId,
  1681. resourceReferrer: options.resourceReferrer,
  1682. validateTextContentType: true,
  1683. baseURI: baseURI,
  1684. blockMixedContent: options.blockMixedContent,
  1685. expectedType: "stylesheet",
  1686. acceptHeaders: options.acceptHeaders
  1687. });
  1688. if (!matchCharsetEquals(content.data, content.charset || options.charset)) {
  1689. options = Object.assign({}, options, { charset: getCharset(content.data) });
  1690. return ProcessorHelper.resolveLinkStylesheetURLs(resourceURL, baseURI, options, workStylesheet);
  1691. }
  1692. resourceURL = content.resourceURL;
  1693. content.data = getUpdatedResourceContent(content.resourceURL, content, options);
  1694. if (content.data && content.data.match(/^<!doctype /i)) {
  1695. content.data = "";
  1696. }
  1697. let stylesheetContent = removeCssComments(content.data);
  1698. if (options.compressCSS) {
  1699. stylesheetContent = util.compressCSS(stylesheetContent);
  1700. }
  1701. stylesheetContent = await ProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options, workStylesheet);
  1702. return stylesheetContent;
  1703. }
  1704. }
  1705. static async processStylesheet(cssRules, baseURI, options, cssVariables, batchRequest) {
  1706. const promises = [];
  1707. const removedRules = [];
  1708. for (let cssRule = cssRules.head; cssRule; cssRule = cssRule.next) {
  1709. const ruleData = cssRule.data;
  1710. if (ruleData.type == "Atrule" && ruleData.name == "charset") {
  1711. removedRules.push(cssRule);
  1712. } else if (ruleData.block && ruleData.block.children) {
  1713. if (ruleData.type == "Rule") {
  1714. promises.push(this.processStyle(ruleData.block.children.toArray(), baseURI, options, cssVariables, batchRequest));
  1715. } else if (ruleData.type == "Atrule" && (ruleData.name == "media" || ruleData.name == "supports")) {
  1716. promises.push(this.processStylesheet(ruleData.block.children, baseURI, options, cssVariables, batchRequest));
  1717. } else if (ruleData.type == "Atrule" && ruleData.name == "font-face") {
  1718. promises.push(processFontFaceRule(ruleData, options.fontURLs));
  1719. }
  1720. }
  1721. }
  1722. removedRules.forEach(cssRule => cssRules.remove(cssRule));
  1723. await Promise.all(promises);
  1724. async function processFontFaceRule(ruleData, fontURLs) {
  1725. await Promise.all(ruleData.block.children.toArray().map(async declaration => {
  1726. if (declaration.type == "Declaration" && declaration.value.children) {
  1727. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value), true);
  1728. await Promise.all(urlFunctions.map(async urlFunction => {
  1729. const originalResourceURL = matchURL(urlFunction);
  1730. const resourceURL = normalizeURL(originalResourceURL);
  1731. if (!testIgnoredPath(resourceURL)) {
  1732. if (testValidURL(resourceURL)) {
  1733. let { content } = await batchRequest.addURL(resourceURL,
  1734. { asBinary: true, expectedType: "font", baseURI, blockMixedContent: options.blockMixedContent });
  1735. let resourceURLs = fontURLs.get(declaration);
  1736. if (!resourceURLs) {
  1737. resourceURLs = [];
  1738. fontURLs.set(declaration, resourceURLs);
  1739. }
  1740. resourceURLs.push(resourceURL);
  1741. replaceURLs(declaration, originalResourceURL, content);
  1742. }
  1743. }
  1744. }));
  1745. }
  1746. }));
  1747. function replaceURLs(declaration, oldURL, newURL) {
  1748. declaration.value.children.forEach(token => {
  1749. if (token.type == "Url" && util.removeQuotes(getCSSValue(token.value)) == oldURL) {
  1750. token.value.value = newURL;
  1751. }
  1752. });
  1753. }
  1754. }
  1755. }
  1756. static async processStyle(declarations, baseURI, options, cssVariables, batchRequest) {
  1757. await Promise.all(declarations.map(async declaration => {
  1758. if (declaration.value && !declaration.value.children && declaration.value.type == "Raw") {
  1759. try {
  1760. declaration.value = cssTree.parse(declaration.value.value, { context: "value" });
  1761. } catch (error) {
  1762. // ignored
  1763. }
  1764. }
  1765. if (declaration.type == "Declaration" && declaration.value.children) {
  1766. const urlFunctions = getUrlFunctions(getCSSValue(declaration.value));
  1767. await Promise.all(urlFunctions.map(async urlFunction => {
  1768. const originalResourceURL = matchURL(urlFunction);
  1769. const resourceURL = normalizeURL(originalResourceURL);
  1770. if (!testIgnoredPath(resourceURL)) {
  1771. if (testValidURL(resourceURL)) {
  1772. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL,
  1773. { asBinary: true, expectedType: "image", groupDuplicates: options.groupDuplicateImages });
  1774. let variableDefined;
  1775. const tokens = [];
  1776. findURLToken(originalResourceURL, declaration.value.children, (token, parent, rootFunction) => {
  1777. if (!originalResourceURL.startsWith("#")) {
  1778. if (duplicate && options.groupDuplicateImages && rootFunction && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1779. const value = cssTree.parse("var(" + SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource + ")", { context: "value" }).children.head;
  1780. tokens.push({ parent, token, value });
  1781. variableDefined = true;
  1782. } else {
  1783. token.data.value.value = content;
  1784. }
  1785. }
  1786. });
  1787. if (variableDefined) {
  1788. cssVariables.set(indexResource, { content, url: originalResourceURL });
  1789. tokens.forEach(({ parent, token, value }) => parent.replace(token, value));
  1790. }
  1791. }
  1792. }
  1793. }));
  1794. }
  1795. }));
  1796. function findURLToken(url, children, callback, depth = 0) {
  1797. for (let token = children.head; token; token = token.next) {
  1798. if (token.data.children) {
  1799. findURLToken(url, token.data.children, callback, depth + 1);
  1800. }
  1801. if (token.data.type == "Url" && util.removeQuotes(getCSSValue(token.data.value)) == url) {
  1802. callback(token, children, depth == 0);
  1803. }
  1804. }
  1805. }
  1806. }
  1807. static async processAttribute(resourceElements, attributeName, baseURI, options, cssVariables, styles, batchRequest, processDuplicates, removeElementIfMissing) {
  1808. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1809. let resourceURL = resourceElement.getAttribute(attributeName);
  1810. if (resourceURL != null) {
  1811. resourceURL = normalizeURL(resourceURL);
  1812. let originURL = resourceElement.dataset.singleFileOriginURL;
  1813. if (options.saveOriginalURLs && !isDataURL(resourceURL)) {
  1814. resourceElement.setAttribute("data-sf-original-" + attributeName, resourceURL);
  1815. }
  1816. delete resourceElement.dataset.singleFileOriginURL;
  1817. if (!testIgnoredPath(resourceURL)) {
  1818. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1819. if (testValidPath(resourceURL)) {
  1820. try {
  1821. resourceURL = util.resolveURL(resourceURL, baseURI);
  1822. } catch (error) {
  1823. // ignored
  1824. }
  1825. if (testValidURL(resourceURL)) {
  1826. let { content, indexResource, duplicate } = await batchRequest.addURL(resourceURL,
  1827. { asBinary: true, expectedType: "image", groupDuplicates: options.groupDuplicateImages && resourceElement.tagName == "IMG" && attributeName == "src" });
  1828. if (originURL) {
  1829. if (content == EMPTY_DATA_URI) {
  1830. try {
  1831. originURL = util.resolveURL(originURL, baseURI);
  1832. } catch (error) {
  1833. // ignored
  1834. }
  1835. try {
  1836. resourceURL = originURL;
  1837. content = (await util.getContent(resourceURL, {
  1838. asBinary: true,
  1839. expectedType: "image",
  1840. maxResourceSize: options.maxResourceSize,
  1841. maxResourceSizeEnabled: options.maxResourceSizeEnabled,
  1842. frameId: options.windowId,
  1843. resourceReferrer: options.resourceReferrer,
  1844. acceptHeaders: options.acceptHeaders
  1845. })).data;
  1846. } catch (error) {
  1847. // ignored
  1848. }
  1849. }
  1850. }
  1851. if (removeElementIfMissing && content == EMPTY_DATA_URI) {
  1852. resourceElement.remove();
  1853. } else {
  1854. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1855. if (!forbiddenPrefixFound) {
  1856. const isSVG = content.startsWith(PREFIX_DATA_URI_IMAGE_SVG);
  1857. if (processDuplicates && duplicate && options.groupDuplicateImages && !isSVG && util.getContentSize(content) < SINGLE_FILE_VARIABLE_MAX_SIZE) {
  1858. if (ProcessorHelper.replaceImageSource(resourceElement, SINGLE_FILE_VARIABLE_NAME_PREFIX + indexResource, options)) {
  1859. cssVariables.set(indexResource, { content, url: originURL });
  1860. const declarationList = cssTree.parse(resourceElement.getAttribute("style"), { context: "declarationList" });
  1861. styles.set(resourceElement, declarationList);
  1862. } else {
  1863. resourceElement.setAttribute(attributeName, content);
  1864. }
  1865. } else {
  1866. resourceElement.setAttribute(attributeName, content);
  1867. }
  1868. }
  1869. }
  1870. }
  1871. }
  1872. }
  1873. }
  1874. }));
  1875. }
  1876. static async processXLinks(resourceElements, doc, baseURI, options, batchRequest) {
  1877. let attributeName = "xlink:href";
  1878. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1879. let originalResourceURL = resourceElement.getAttribute(attributeName);
  1880. if (originalResourceURL == null) {
  1881. attributeName = "href";
  1882. originalResourceURL = resourceElement.getAttribute(attributeName);
  1883. }
  1884. if (options.saveOriginalURLs && !isDataURL(originalResourceURL)) {
  1885. resourceElement.setAttribute("data-sf-original-href", originalResourceURL);
  1886. }
  1887. let resourceURL = normalizeURL(originalResourceURL);
  1888. if (testValidPath(resourceURL) && !testIgnoredPath(resourceURL)) {
  1889. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1890. try {
  1891. resourceURL = util.resolveURL(resourceURL, baseURI);
  1892. } catch (error) {
  1893. // ignored
  1894. }
  1895. if (testValidURL(resourceURL)) {
  1896. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  1897. if (originalResourceURL.startsWith(baseURI + "#")) {
  1898. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1899. } else {
  1900. const response = await batchRequest.addURL(resourceURL, { expectedType: "image" });
  1901. const svgDoc = util.parseSVGContent(response.content);
  1902. if (hashMatch && hashMatch[0]) {
  1903. let symbolElement;
  1904. try {
  1905. symbolElement = svgDoc.querySelector(hashMatch[0]);
  1906. } catch (error) {
  1907. // ignored
  1908. }
  1909. if (symbolElement) {
  1910. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1911. resourceElement.parentElement.insertBefore(symbolElement, resourceElement.parentElement.firstChild);
  1912. }
  1913. } else {
  1914. const content = await batchRequest.addURL(resourceURL, { expectedType: "image" });
  1915. resourceElement.setAttribute(attributeName, PREFIX_DATA_URI_IMAGE_SVG + "," + content);
  1916. }
  1917. }
  1918. }
  1919. } else if (resourceURL == options.url) {
  1920. resourceElement.setAttribute(attributeName, originalResourceURL.substring(resourceURL.length));
  1921. }
  1922. }));
  1923. }
  1924. static async processSrcset(resourceElements, attributeName, baseURI, batchRequest) {
  1925. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1926. const originSrcset = resourceElement.getAttribute(attributeName);
  1927. const srcset = util.parseSrcset(originSrcset);
  1928. resourceElement.setAttribute("data-sf-original-srcset", originSrcset);
  1929. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1930. let resourceURL = normalizeURL(srcsetValue.url);
  1931. if (!testIgnoredPath(resourceURL)) {
  1932. if (testValidPath(resourceURL)) {
  1933. try {
  1934. resourceURL = util.resolveURL(resourceURL, baseURI);
  1935. } catch (error) {
  1936. // ignored
  1937. }
  1938. if (testValidURL(resourceURL)) {
  1939. const { content } = await batchRequest.addURL(resourceURL, { asBinary: true, expectedType: "image" });
  1940. const forbiddenPrefixFound = PREFIXES_FORBIDDEN_DATA_URI.filter(prefixDataURI => content.startsWith(prefixDataURI)).length;
  1941. if (forbiddenPrefixFound) {
  1942. return "";
  1943. }
  1944. return content + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1945. } else {
  1946. return "";
  1947. }
  1948. } else {
  1949. return "";
  1950. }
  1951. } else {
  1952. return resourceURL + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1953. }
  1954. }));
  1955. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1956. }));
  1957. }
  1958. static replaceImageSource(imgElement, variableName, options) {
  1959. const attributeValue = imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1960. if (attributeValue) {
  1961. const imageData = options.images[Number(imgElement.getAttribute(util.IMAGE_ATTRIBUTE_NAME))];
  1962. if (imageData && imageData.replaceable) {
  1963. imgElement.setAttribute("src", `${PREFIX_DATA_URI_IMAGE_SVG},<svg xmlns="http://www.w3.org/2000/svg" width="${imageData.size.pxWidth}" height="${imageData.size.pxHeight}"><rect fill-opacity="0"/></svg>`);
  1964. const backgroundStyle = {};
  1965. const backgroundSize = (imageData.objectFit == "content" || imageData.objectFit == "cover") && imageData.objectFit;
  1966. if (backgroundSize) {
  1967. backgroundStyle["background-size"] = imageData.objectFit;
  1968. }
  1969. if (imageData.objectPosition) {
  1970. backgroundStyle["background-position"] = imageData.objectPosition;
  1971. }
  1972. if (imageData.backgroundColor) {
  1973. backgroundStyle["background-color"] = imageData.backgroundColor;
  1974. }
  1975. ProcessorHelper.setBackgroundImage(imgElement, "var(" + variableName + ")", backgroundStyle);
  1976. imgElement.removeAttribute(util.IMAGE_ATTRIBUTE_NAME);
  1977. return true;
  1978. }
  1979. }
  1980. }
  1981. }
  1982. // ----
  1983. // Util
  1984. // ----
  1985. const BLOB_URI_PREFIX = "blob:";
  1986. const HTTP_URI_PREFIX = /^https?:\/\//;
  1987. const FILE_URI_PREFIX = /^file:\/\//;
  1988. const EMPTY_URL = /^https?:\/\/+\s*$/;
  1989. const NOT_EMPTY_URL = /^(https?:\/\/|file:\/\/|blob:).+/;
  1990. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1991. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1992. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1993. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1994. 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;
  1995. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1996. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1997. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1998. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1999. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  2000. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  2001. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  2002. function getUpdatedResourceContent(resourceURL, content, options) {
  2003. if (options.rootDocument && options.updatedResources[resourceURL]) {
  2004. options.updatedResources[resourceURL].retrieved = true;
  2005. return options.updatedResources[resourceURL].content;
  2006. } else {
  2007. return content.data || "";
  2008. }
  2009. }
  2010. function normalizeURL(url) {
  2011. if (!url || url.startsWith(DATA_URI_PREFIX)) {
  2012. return url;
  2013. } else {
  2014. return url.split("#")[0];
  2015. }
  2016. }
  2017. function getCSSValue(value) {
  2018. let result = "";
  2019. try {
  2020. result = cssTree.generate(value);
  2021. } catch (error) {
  2022. // ignored
  2023. }
  2024. return result;
  2025. }
  2026. function matchCharsetEquals(stylesheetContent, charset = "utf-8") {
  2027. const stylesheetCharset = getCharset(stylesheetContent);
  2028. if (stylesheetCharset) {
  2029. return stylesheetCharset == charset.toLowerCase();
  2030. } else {
  2031. return true;
  2032. }
  2033. }
  2034. function getCharset(stylesheetContent) {
  2035. const match = stylesheetContent.match(/^@charset\s+"([^"]*)";/i);
  2036. if (match && match[1]) {
  2037. return match[1].toLowerCase().trim();
  2038. }
  2039. }
  2040. function getOnEventAttributeNames(doc) {
  2041. const element = doc.createElement("div");
  2042. const attributeNames = [];
  2043. for (const propertyName in element) {
  2044. if (propertyName.startsWith("on")) {
  2045. attributeNames.push(propertyName);
  2046. }
  2047. }
  2048. attributeNames.push("onunload");
  2049. return attributeNames;
  2050. }
  2051. async function evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash, replacementCharacter) {
  2052. let maxLength;
  2053. if (template) {
  2054. const regExpVariable = "{\\s*" + variableName.replace(/\W|_/g, "[$&]") + "\\s*}";
  2055. let replaceRegExp = new RegExp(regExpVariable + "\\[\\d+\\]", "g");
  2056. if (template.match(replaceRegExp)) {
  2057. const matchedLength = template.match(replaceRegExp)[0];
  2058. maxLength = Number(matchedLength.match(/\[(\d+)\]$/)[1]);
  2059. if (isNaN(maxLength) || maxLength <= 0) {
  2060. maxLength = null;
  2061. }
  2062. } else {
  2063. replaceRegExp = new RegExp(regExpVariable, "g");
  2064. }
  2065. if (template.match(replaceRegExp)) {
  2066. let value = await valueGetter();
  2067. if (!dontReplaceSlash) {
  2068. value = value.replace(/\/+/g, replacementCharacter);
  2069. }
  2070. if (maxLength) {
  2071. value = await util.truncateText(value, maxLength);
  2072. }
  2073. return template.replace(replaceRegExp, value);
  2074. }
  2075. }
  2076. return template;
  2077. }
  2078. function getLastSegment(url, replacementCharacter) {
  2079. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/), lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2080. if (!lastSegment) {
  2081. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  2082. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2083. }
  2084. if (!lastSegment) {
  2085. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  2086. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  2087. }
  2088. if (!lastSegment) {
  2089. lastSegment = url.hostname.replace(/\/+/g, replacementCharacter).replace(/\/$/, "");
  2090. }
  2091. lastSegmentMatch = lastSegment.match(/(.*)\.[^.]+$/);
  2092. if (lastSegmentMatch && lastSegmentMatch[1]) {
  2093. lastSegment = lastSegmentMatch[1];
  2094. }
  2095. lastSegment = lastSegment.replace(/\/$/, "").replace(/^\//, "");
  2096. return lastSegment;
  2097. }
  2098. function getRegExp(string) {
  2099. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  2100. }
  2101. function getUrlFunctions(stylesheetContent, unique) {
  2102. const result = stylesheetContent.match(REGEXP_URL_FN) || [];
  2103. if (unique) {
  2104. return [...new Set(result)];
  2105. } else {
  2106. return result;
  2107. }
  2108. }
  2109. function getImportFunctions(stylesheetContent) {
  2110. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  2111. }
  2112. function findShortcutIcon(shortcutIcons) {
  2113. shortcutIcons = shortcutIcons.filter(linkElement => linkElement.href != EMPTY_IMAGE);
  2114. shortcutIcons.sort((linkElement1, linkElement2) => (parseInt(linkElement2.sizes, 10) || 16) - (parseInt(linkElement1.sizes, 10) || 16));
  2115. return shortcutIcons[0];
  2116. }
  2117. function matchURL(stylesheetContent) {
  2118. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  2119. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  2120. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  2121. return match && match[1];
  2122. }
  2123. function addOriginalURLs(stylesheetContent) {
  2124. return stylesheetContent.replace(REGEXP_URL_FN, function (match, _0, url, _1, url2, _2, url3) {
  2125. url = url || url2 || url3;
  2126. if (isDataURL(url)) {
  2127. return match;
  2128. } else {
  2129. return "-sf-url-original(" + JSON.stringify(url) + ") " + match;
  2130. }
  2131. });
  2132. }
  2133. function isDataURL(url) {
  2134. return url && (url.startsWith(DATA_URI_PREFIX) || url.startsWith(BLOB_URI_PREFIX));
  2135. }
  2136. function replaceOriginalURLs(stylesheetContent) {
  2137. return stylesheetContent.replace(/-sf-url-original\("(.*?)"\)/g, "/* original URL: $1 */");
  2138. }
  2139. function testIgnoredPath(resourceURL) {
  2140. return resourceURL && (resourceURL.startsWith(DATA_URI_PREFIX) || resourceURL == ABOUT_BLANK_URI);
  2141. }
  2142. function testValidPath(resourceURL) {
  2143. return resourceURL && !resourceURL.match(EMPTY_URL);
  2144. }
  2145. function testValidURL(resourceURL) {
  2146. return testValidPath(resourceURL) && (resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(FILE_URI_PREFIX) || resourceURL.startsWith(BLOB_URI_PREFIX)) && resourceURL.match(NOT_EMPTY_URL);
  2147. }
  2148. function matchImport(stylesheetContent) {
  2149. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  2150. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  2151. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  2152. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  2153. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  2154. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  2155. if (match) {
  2156. const [, resourceURL, media] = match;
  2157. return { resourceURL, media };
  2158. }
  2159. }
  2160. function removeCssComments(stylesheetContent) {
  2161. try {
  2162. return stylesheetContent.replace(/\/\*(.|\s)*?\*\//g, "");
  2163. } catch (error) {
  2164. let start, end;
  2165. do {
  2166. start = stylesheetContent.indexOf("/*");
  2167. end = stylesheetContent.indexOf("*/", start + 2);
  2168. if (start != -1 && end != -1) {
  2169. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  2170. }
  2171. } while (start != -1 && end != -1);
  2172. return stylesheetContent;
  2173. }
  2174. }
  2175. function wrapMediaQuery(stylesheetContent, mediaQuery) {
  2176. if (mediaQuery) {
  2177. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  2178. } else {
  2179. return stylesheetContent;
  2180. }
  2181. }
  2182. function log(...args) {
  2183. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  2184. }
  2185. // -----
  2186. // Stats
  2187. // -----
  2188. const STATS_DEFAULT_VALUES = {
  2189. discarded: {
  2190. "HTML bytes": 0,
  2191. "hidden elements": 0,
  2192. "HTML imports": 0,
  2193. scripts: 0,
  2194. objects: 0,
  2195. "audio sources": 0,
  2196. "video sources": 0,
  2197. frames: 0,
  2198. "CSS rules": 0,
  2199. canvas: 0,
  2200. stylesheets: 0,
  2201. resources: 0,
  2202. medias: 0
  2203. },
  2204. processed: {
  2205. "HTML bytes": 0,
  2206. "hidden elements": 0,
  2207. "HTML imports": 0,
  2208. scripts: 0,
  2209. objects: 0,
  2210. "audio sources": 0,
  2211. "video sources": 0,
  2212. frames: 0,
  2213. "CSS rules": 0,
  2214. canvas: 0,
  2215. stylesheets: 0,
  2216. resources: 0,
  2217. medias: 0
  2218. }
  2219. };
  2220. class Stats {
  2221. constructor(options) {
  2222. this.options = options;
  2223. if (options.displayStats) {
  2224. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  2225. }
  2226. }
  2227. set(type, subType, value) {
  2228. if (this.options.displayStats) {
  2229. this.data[type][subType] = value;
  2230. }
  2231. }
  2232. add(type, subType, value) {
  2233. if (this.options.displayStats) {
  2234. this.data[type][subType] += value;
  2235. }
  2236. }
  2237. addAll(pageData) {
  2238. if (this.options.displayStats) {
  2239. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  2240. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  2241. }
  2242. }
  2243. }
  2244. export {
  2245. getClass
  2246. };