single-file-core.js 54 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294
  1. /*
  2. * Copyright 2018 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile.
  6. *
  7. * SingleFile is free software: you can redistribute it and/or modify
  8. * it under the terms of the GNU Lesser General Public License as published by
  9. * the Free Software Foundation, either version 3 of the License, or
  10. * (at your option) any later version.
  11. *
  12. * SingleFile 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
  15. * GNU Lesser General Public License for more details.
  16. *
  17. * You should have received a copy of the GNU Lesser General Public License
  18. * along with SingleFile. If not, see <http://www.gnu.org/licenses/>.
  19. */
  20. this.SingleFileCore = this.SingleFileCore || (() => {
  21. const SELECTED_CONTENT_ATTRIBUTE_NAME = "data-single-file-selected-content";
  22. const SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = "data-single-file-selected-content-root";
  23. const DEBUG = false;
  24. let Download, DOM, URL, sessionId = 0;
  25. function getClass(...args) {
  26. [Download, DOM, URL] = args;
  27. return class {
  28. constructor(options) {
  29. this.options = options;
  30. options.sessionId = sessionId;
  31. sessionId++;
  32. this.SELECTED_CONTENT_ATTRIBUTE_NAME = SELECTED_CONTENT_ATTRIBUTE_NAME;
  33. this.SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME = SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME;
  34. }
  35. async initialize() {
  36. this.processor = new PageProcessor(this.options);
  37. await this.processor.loadPage();
  38. await this.processor.initialize();
  39. }
  40. async preparePageData() {
  41. await this.processor.preparePageData();
  42. }
  43. async getPageData() {
  44. return this.processor.getPageData();
  45. }
  46. };
  47. }
  48. // -------------
  49. // ProgressEvent
  50. // -------------
  51. const PAGE_LOADING = "page-loading";
  52. const PAGE_LOADED = "page-loaded";
  53. const RESOURCES_INITIALIZING = "resource-initializing";
  54. const RESOURCES_INITIALIZED = "resources-initialized";
  55. const RESOURCE_LOADED = "resource-loaded";
  56. const PAGE_ENDED = "page-ended";
  57. class ProgressEvent {
  58. constructor(type, details) {
  59. return { type, details, PAGE_LOADING, PAGE_LOADED, RESOURCES_INITIALIZING, RESOURCES_INITIALIZED, RESOURCE_LOADED, PAGE_ENDED };
  60. }
  61. }
  62. // -------------
  63. // PageProcessor
  64. // -------------
  65. const STAGES = [{
  66. sequential: [
  67. { action: "removeUIElements" },
  68. { action: "replaceStyleContents" },
  69. { option: "removeVideoSrc", action: "insertVideoPosters" },
  70. { option: "removeSrcSet", action: "removeSrcSet" },
  71. { option: "removeFrames", action: "removeFrames" },
  72. { option: "removeImports", action: "removeImports" },
  73. { option: "removeScripts", action: "removeScripts" },
  74. { action: "removeDiscardedResources" },
  75. { action: "resetCharsetMeta" },
  76. { action: "setInputValues" },
  77. { option: "insertFaviconLink", action: "insertFaviconLink" },
  78. { action: "resolveHrefs" },
  79. { action: "replaceCanvasElements" },
  80. { action: "insertFonts" },
  81. { option: "removeHiddenElements", action: "removeHiddenElements" }
  82. ],
  83. parallel: [
  84. { action: "inlineStylesheets" },
  85. { action: "linkStylesheets" },
  86. { action: "attributeStyles" },
  87. { option: "!removeFrames", action: "frames" },
  88. { option: "!removeImports", action: "htmlImports" }
  89. ]
  90. }, {
  91. sequential: [
  92. { option: "removeUnusedStyles", action: "removeUnusedStyles" },
  93. { option: "removeAlternativeFonts", action: "removeAlternativeFonts" },
  94. { option: "removeAlternativeMedias", action: "removeAlternativeMedias" }
  95. ],
  96. parallel: [
  97. { action: "inlineStylesheets" },
  98. { action: "attributeStyles" },
  99. { action: "pageResources" },
  100. { option: "!removeScripts", action: "scripts" }
  101. ]
  102. }, {
  103. sequential: [
  104. { option: "lazyLoadImages", action: "lazyLoadImages" },
  105. { option: "removeAlternativeFonts", action: "postRemoveAlternativeFonts" },
  106. { option: "compressCSS", action: "compressCSS" }
  107. ],
  108. parallel: [
  109. { option: "!removeFrames", action: "frames" },
  110. { option: "!removeImports", action: "htmlImports" },
  111. ]
  112. }, {
  113. sequential: [
  114. { option: "compressHTML", action: "compressHTML" },
  115. { option: "insertSingleFileComment", action: "insertSingleFileComment" },
  116. { action: "removeDefaultHeadTags" }
  117. ]
  118. }];
  119. class PageProcessor {
  120. constructor(options) {
  121. this.options = options;
  122. this.options.url = this.options.url || this.options.doc.location.href;
  123. this.options.baseURI = this.options.doc && this.options.doc.baseURI;
  124. this.batchRequest = new BatchRequest();
  125. this.processor = new DOMProcessor(options, this.batchRequest);
  126. if (this.options.doc) {
  127. const docData = DOM.preProcessDoc(this.options.doc, this.options.win, this.options);
  128. this.options.canvasData = docData.canvasData;
  129. this.options.fontsData = docData.fontsData;
  130. this.options.stylesheetContents = docData.stylesheetContents;
  131. this.options.responsiveImageData = docData.responsiveImageData;
  132. this.options.imageData = docData.imageData;
  133. this.options.postersData = docData.postersData;
  134. }
  135. this.options.content = this.options.content || (this.options.doc ? DOM.serialize(this.options.doc, false) : null);
  136. this.onprogress = options.onprogress || (() => { });
  137. }
  138. async loadPage() {
  139. this.onprogress(new ProgressEvent(PAGE_LOADING, { pageURL: this.options.url }));
  140. await this.processor.loadPage(this.options.content);
  141. this.onprogress(new ProgressEvent(PAGE_LOADED, { pageURL: this.options.url }));
  142. }
  143. async initialize() {
  144. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZING, { pageURL: this.options.url }));
  145. await this.executeStage(0);
  146. this.pendingPromises = this.executeStage(1);
  147. if (this.options.doc) {
  148. DOM.postProcessDoc(this.options.doc, this.options);
  149. this.options.doc = null;
  150. this.options.win = null;
  151. }
  152. }
  153. async preparePageData() {
  154. if (!this.options.windowId) {
  155. this.processor.initialize(this.batchRequest);
  156. this.onprogress(new ProgressEvent(RESOURCES_INITIALIZED, { pageURL: this.options.url, index: 0, max: this.processor.maxResources }));
  157. }
  158. await this.batchRequest.run(details => {
  159. details.pageURL = this.options.url;
  160. this.onprogress(new ProgressEvent(RESOURCE_LOADED, details));
  161. }, this.options);
  162. await this.pendingPromises;
  163. await this.executeStage(2);
  164. await this.executeStage(3);
  165. }
  166. async getPageData() {
  167. if (!this.options.windowId) {
  168. this.onprogress(new ProgressEvent(PAGE_ENDED, { pageURL: this.options.url }));
  169. }
  170. return this.processor.getPageData();
  171. }
  172. async executeStage(step) {
  173. if (DEBUG) {
  174. log("**** STARTED STAGE", step, "****");
  175. }
  176. STAGES[step].sequential.forEach(task => {
  177. let startTime;
  178. if (DEBUG) {
  179. startTime = Date.now();
  180. log(" -- STARTED task =", task.action);
  181. }
  182. this.executeTask(task, !step);
  183. if (DEBUG) {
  184. log(" -- ENDED task =", task.action, "delay =", Date.now() - startTime);
  185. }
  186. });
  187. if (STAGES[step].parallel) {
  188. return await Promise.all(STAGES[step].parallel.map(task => {
  189. let startTime;
  190. if (DEBUG) {
  191. startTime = Date.now();
  192. log(" // STARTED task =", task.action);
  193. }
  194. const promise = this.executeTask(task, !step);
  195. if (DEBUG) {
  196. promise.then(() => log(" // ENDED task =", task.action, "delay =", Date.now() - startTime));
  197. }
  198. return promise;
  199. }));
  200. }
  201. if (DEBUG) {
  202. log("**** ENDED STAGE", step, "****");
  203. }
  204. }
  205. executeTask(task, initialization) {
  206. if (!task.option || ((task.option.startsWith("!") && !this.options[task.option]) || this.options[task.option])) {
  207. return this.processor[task.action](initialization);
  208. }
  209. }
  210. }
  211. // --------
  212. // BatchRequest
  213. // --------
  214. class BatchRequest {
  215. constructor() {
  216. this.requests = new Map();
  217. }
  218. async addURL(resourceURL, asDataURI = true) {
  219. return new Promise((resolve, reject) => {
  220. const requestKey = JSON.stringify([resourceURL, asDataURI]);
  221. const resourceRequests = this.requests.get(requestKey);
  222. if (resourceRequests) {
  223. resourceRequests.push({ resolve, reject });
  224. } else {
  225. this.requests.set(requestKey, [{ resolve, reject }]);
  226. }
  227. });
  228. }
  229. getMaxResources() {
  230. return Array.from(this.requests.keys()).length;
  231. }
  232. async run(onloadListener, options) {
  233. const resourceURLs = Array.from(this.requests.keys());
  234. let indexResource = 0;
  235. return Promise.all(resourceURLs.map(async requestKey => {
  236. const [resourceURL, asDataURI] = JSON.parse(requestKey);
  237. const resourceRequests = this.requests.get(requestKey);
  238. try {
  239. const dataURI = await Download.getContent(resourceURL, { asDataURI, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  240. indexResource = indexResource + 1;
  241. onloadListener({ index: indexResource, url: resourceURL });
  242. resourceRequests.forEach(resourceRequest => resourceRequest.resolve(dataURI));
  243. } catch (error) {
  244. indexResource = indexResource + 1;
  245. onloadListener({ index: indexResource, url: resourceURL });
  246. resourceRequests.forEach(resourceRequest => resourceRequest.reject(error));
  247. }
  248. this.requests.delete(requestKey);
  249. }));
  250. }
  251. }
  252. // ------------
  253. // DOMProcessor
  254. // ------------
  255. const ESCAPED_FRAGMENT = "_escaped_fragment_=";
  256. const EMPTY_DATA_URI = "data:base64,";
  257. const EMPTY_IMAGE = "data:image/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  258. class DOMProcessor {
  259. constructor(options, batchRequest) {
  260. this.options = options;
  261. this.stats = new Stats(options);
  262. this.baseURI = DomUtil.normalizeURL(options.baseURI || options.url);
  263. this.batchRequest = batchRequest;
  264. }
  265. initialize() {
  266. this.maxResources = this.batchRequest.getMaxResources();
  267. if (!this.options.removeFrames) {
  268. this.options.framesData.forEach(frameData => this.maxResources += frameData.maxResources || 0);
  269. }
  270. this.stats.set("processed", "resources", this.maxResources);
  271. }
  272. async loadPage(pageContent) {
  273. if (!pageContent || this.options.saveRawPage) {
  274. pageContent = await Download.getContent(this.baseURI, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  275. }
  276. this.doc = DOM.createDoc(pageContent, this.baseURI);
  277. this.onEventAttributeNames = DOM.getOnEventAttributeNames(this.doc);
  278. if (!pageContent && this.doc.querySelector("meta[name=fragment][content=\"!\"]") && !this.baseURI.endsWith("?" + ESCAPED_FRAGMENT) && !this.baseURI.endsWith("&" + ESCAPED_FRAGMENT)) {
  279. await DOMProcessor.loadEscapedFragmentPage();
  280. }
  281. }
  282. async loadEscapedFragmentPage() {
  283. if (this.baseURI.includes("?")) {
  284. this.baseURI += "&";
  285. } else {
  286. this.baseURI += "?";
  287. }
  288. this.baseURI += ESCAPED_FRAGMENT;
  289. await this.loadPage();
  290. }
  291. async getPageData() {
  292. DOM.postProcessDoc(this.doc, this.options);
  293. if (this.options.selected) {
  294. const rootElement = this.doc.querySelector("[" + SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME + "]");
  295. if (rootElement) {
  296. DomProcessorHelper.isolateElements(rootElement);
  297. rootElement.removeAttribute(SELECTED_CONTENT_ROOT_ATTRIBUTE_NAME);
  298. rootElement.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  299. }
  300. }
  301. const titleElement = this.doc.querySelector("title");
  302. this.options.title = titleElement ? titleElement.textContent.trim() : "";
  303. this.options.info = {};
  304. const descriptionElement = this.doc.querySelector("meta[name=description]");
  305. this.options.info.description = descriptionElement ? descriptionElement.content.trim() : "";
  306. this.options.info.lang = this.doc.documentElement.lang;
  307. const authorElement = this.doc.querySelector("meta[name=author]");
  308. this.options.info.author = authorElement ? authorElement.content.trim() : "";
  309. const creatorElement = this.doc.querySelector("meta[name=creator]");
  310. this.options.info.creator = creatorElement ? creatorElement.content.trim() : "";
  311. const publisherElement = this.doc.querySelector("meta[name=publisher]");
  312. this.options.info.publisher = creatorElement ? publisherElement.content.trim() : "";
  313. const url = new URL(this.baseURI);
  314. let size;
  315. if (this.options.displayStats) {
  316. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  317. }
  318. const content = DOM.serialize(this.doc, this.options.compressHTML);
  319. if (this.options.displayStats) {
  320. const contentSize = DOM.getContentSize(content);
  321. this.stats.set("processed", "HTML bytes", contentSize);
  322. this.stats.add("discarded", "HTML bytes", size - contentSize);
  323. }
  324. const filename = await DomProcessorHelper.getFilename(this.options, content);
  325. const matchTitle = this.baseURI.match(/([^/]*)\/?(\.html?.*)$/) || this.baseURI.match(/\/\/([^/]*)\/?$/);
  326. return {
  327. stats: this.stats.data,
  328. title: this.options.title || (this.baseURI && matchTitle ? matchTitle[1] : (url.hostname ? url.hostname : "")),
  329. filename,
  330. content
  331. };
  332. }
  333. setInputValues() {
  334. this.doc.querySelectorAll("input").forEach(input => {
  335. const value = input.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  336. if (value) {
  337. input.setAttribute("value", value);
  338. }
  339. });
  340. this.doc.querySelectorAll("textarea").forEach(textarea => {
  341. const value = textarea.getAttribute(DOM.inputValueAttributeName(this.options.sessionId));
  342. if (value) {
  343. textarea.textContent = value;
  344. }
  345. });
  346. this.doc.querySelectorAll("select").forEach(select => {
  347. select.querySelectorAll("option").forEach(option => {
  348. const selected = option.getAttribute(DOM.inputValueAttributeName(this.options.sessionId)) != null;
  349. if (selected) {
  350. option.setAttribute("selected", "");
  351. }
  352. });
  353. });
  354. }
  355. lazyLoadImages() {
  356. DOM.lazyLoad(this.doc);
  357. }
  358. removeDiscardedResources() {
  359. const objectElements = this.doc.querySelectorAll("applet, meta[http-equiv=refresh], object:not([type=\"image/svg+xml\"]):not([type=\"image/svg-xml\"]):not([type=\"text/html\"]), embed:not([src*=\".svg\"])");
  360. this.stats.set("discarded", "objects", objectElements.length);
  361. this.stats.set("processed", "objects", objectElements.length);
  362. objectElements.forEach(element => element.remove());
  363. const replacedAttributeValue = this.doc.querySelectorAll("link[rel~=preconnect], link[rel~=prerender], link[rel~=dns-prefetch], link[rel~=preload], link[rel~=prefetch]");
  364. replacedAttributeValue.forEach(element => {
  365. const relValue = element.getAttribute("rel").replace(/(preconnect|prerender|dns-prefetch|preload|prefetch)/g, "").trim();
  366. if (relValue.length) {
  367. element.setAttribute("rel", relValue);
  368. } else {
  369. element.remove();
  370. }
  371. });
  372. this.doc.querySelectorAll("meta[http-equiv=\"content-security-policy\"]").forEach(element => element.remove());
  373. if (this.options.compressHTML) {
  374. this.doc.querySelectorAll("input[type=hidden]").forEach(element => element.remove());
  375. }
  376. this.doc.querySelectorAll("a[ping]").forEach(element => element.removeAttribute("ping"));
  377. if (this.options.removeScripts) {
  378. this.onEventAttributeNames.forEach(attributeName => this.doc.querySelectorAll("[" + attributeName + "]").forEach(element => element.removeAttribute(attributeName)));
  379. this.doc.querySelectorAll("[href]").forEach(element => {
  380. if (element.href && element.href.match && element.href.match(/^\s*javascript:/)) {
  381. element.removeAttribute("href");
  382. }
  383. });
  384. this.doc.querySelectorAll("[src]").forEach(element => {
  385. if (element.src && element.src.match(/^\s*javascript:/)) {
  386. element.removeAttribute("src");
  387. }
  388. });
  389. }
  390. const audioSourceElements = this.doc.querySelectorAll("audio[src], audio > source[src]");
  391. this.stats.set("processed", "audio sources", audioSourceElements.length);
  392. if (this.options.removeAudioSrc) {
  393. this.stats.set("discarded", "audio sources", audioSourceElements.length);
  394. audioSourceElements.forEach(element => element.removeAttribute("src"));
  395. }
  396. const videoSourceElements = this.doc.querySelectorAll("video[src], video > source[src]");
  397. this.stats.set("processed", "video sources", videoSourceElements.length);
  398. if (this.options.removeVideoSrc) {
  399. this.stats.set("discarded", "video sources", videoSourceElements.length);
  400. videoSourceElements.forEach(element => element.removeAttribute("src"));
  401. }
  402. }
  403. removeDefaultHeadTags() {
  404. this.doc.querySelectorAll("base").forEach(element => element.remove());
  405. if (this.doc.head.querySelectorAll("*").length == 1 && this.doc.head.querySelector("meta[charset]") && this.doc.body.childNodes.length == 0) {
  406. this.doc.head.querySelector("meta[charset]").remove();
  407. }
  408. }
  409. removeUIElements() {
  410. this.doc.querySelectorAll("singlefile-infobar, singlefile-mask").forEach(element => element.remove());
  411. }
  412. removeScripts() {
  413. const scriptElements = this.doc.querySelectorAll("script:not([type=\"application/ld+json\"])");
  414. this.stats.set("discarded", "scripts", scriptElements.length);
  415. this.stats.set("processed", "scripts", scriptElements.length);
  416. scriptElements.forEach(element => element.remove());
  417. }
  418. removeFrames() {
  419. const frameElements = this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]");
  420. this.stats.set("discarded", "frames", frameElements.length);
  421. this.stats.set("processed", "frames", frameElements.length);
  422. this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]").forEach(element => element.remove());
  423. }
  424. removeImports() {
  425. const importElements = this.doc.querySelectorAll("link[rel=import]");
  426. this.stats.set("discarded", "HTML imports", importElements.length);
  427. this.stats.set("processed", "HTML imports", importElements.length);
  428. importElements.forEach(element => element.remove());
  429. }
  430. resetCharsetMeta() {
  431. this.doc.querySelectorAll("meta[charset], meta[http-equiv=\"content-type\"]").forEach(element => element.remove());
  432. const metaElement = this.doc.createElement("meta");
  433. metaElement.setAttribute("charset", "utf-8");
  434. this.doc.head.insertBefore(metaElement, this.doc.head.firstElementChild);
  435. }
  436. insertFaviconLink() {
  437. let faviconElement = this.doc.querySelector("link[href][rel=\"icon\"]");
  438. if (!faviconElement) {
  439. faviconElement = this.doc.querySelector("link[href][rel=\"shortcut icon\"]");
  440. }
  441. if (!faviconElement) {
  442. faviconElement = this.doc.createElement("link");
  443. faviconElement.setAttribute("type", "image/x-icon");
  444. faviconElement.setAttribute("rel", "shortcut icon");
  445. faviconElement.setAttribute("href", "/favicon.ico");
  446. }
  447. this.doc.head.appendChild(faviconElement);
  448. }
  449. resolveHrefs() {
  450. this.doc.querySelectorAll("[href]").forEach(element => {
  451. if (element.href) {
  452. const href = element.href.baseVal ? element.href.baseVal : element.href;
  453. const normalizedHref = DomUtil.normalizeURL(href);
  454. if (!normalizedHref || normalizedHref != this.baseURI) {
  455. element.setAttribute("href", href);
  456. }
  457. }
  458. });
  459. }
  460. removeUnusedStyles() {
  461. const stats = DOM.minifyCSS(this.doc);
  462. this.stats.set("processed", "CSS rules", stats.processed);
  463. this.stats.set("discarded", "CSS rules", stats.discarded);
  464. }
  465. removeAlternativeFonts() {
  466. DOM.minifyFonts(this.doc);
  467. }
  468. removeSrcSet() {
  469. this.doc.querySelectorAll("picture, img[srcset]").forEach(element => {
  470. const tagName = element.tagName.toLowerCase();
  471. const dataAttributeName = DOM.responsiveImagesAttributeName(this.options.sessionId);
  472. const responsiveImageData = this.options.responsiveImageData[Number(element.getAttribute(dataAttributeName))];
  473. element.removeAttribute(dataAttributeName);
  474. if (responsiveImageData) {
  475. if (tagName == "img") {
  476. if (responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  477. element.removeAttribute("srcset");
  478. element.removeAttribute("sizes");
  479. element.src = responsiveImageData.source.src;
  480. }
  481. }
  482. if (tagName == "picture") {
  483. const imageElement = element.querySelector("img");
  484. if (responsiveImageData.source && responsiveImageData.source.src && responsiveImageData.source.naturalWidth > 1 && responsiveImageData.source.naturalHeight > 1) {
  485. imageElement.removeAttribute("srcset");
  486. imageElement.removeAttribute("sizes");
  487. imageElement.src = responsiveImageData.source.src;
  488. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  489. } else {
  490. if (responsiveImageData.sources) {
  491. element.querySelectorAll("source").forEach(sourceElement => {
  492. if (!sourceElement.srcset && !sourceElement.dataset.srcset && !sourceElement.src) {
  493. sourceElement.remove();
  494. }
  495. });
  496. const sourceElements = element.querySelectorAll("source");
  497. if (sourceElements.length) {
  498. const lastSourceElement = sourceElements[sourceElements.length - 1];
  499. if (imageElement) {
  500. if (lastSourceElement.src) {
  501. imageElement.src = lastSourceElement.src;
  502. } else {
  503. imageElement.removeAttribute("src");
  504. }
  505. if (lastSourceElement.srcset || lastSourceElement.dataset.srcset) {
  506. imageElement.srcset = lastSourceElement.srcset || lastSourceElement.dataset.srcset;
  507. } else {
  508. imageElement.removeAttribute("srcset");
  509. }
  510. element.querySelectorAll("source").forEach(sourceElement => sourceElement.remove());
  511. }
  512. }
  513. }
  514. }
  515. }
  516. }
  517. });
  518. }
  519. postRemoveAlternativeFonts() {
  520. DOM.minifyFonts(this.doc, true);
  521. }
  522. removeHiddenElements() {
  523. const hiddenElements = this.doc.querySelectorAll("[" + DOM.removedContentAttributeName(this.options.sessionId) + "]");
  524. this.stats.set("discarded", "hidden elements", hiddenElements.length);
  525. this.stats.set("processed", "hidden elements", hiddenElements.length);
  526. hiddenElements.forEach(element => element.remove());
  527. }
  528. compressHTML() {
  529. let size;
  530. if (this.options.displayStats) {
  531. size = DOM.getContentSize(this.doc.documentElement.outerHTML);
  532. }
  533. DOM.minifyHTML(this.doc, { preservedSpaceAttributeName: DOM.preservedSpaceAttributeName(this.options.sessionId) });
  534. if (this.options.displayStats) {
  535. this.stats.add("discarded", "HTML bytes", size - DOM.getContentSize(this.doc.documentElement.outerHTML));
  536. }
  537. }
  538. removeAlternativeMedias() {
  539. const stats = DOM.minifyMedias(this.doc);
  540. this.stats.set("processed", "medias", stats.processed);
  541. this.stats.set("discarded", "medias", stats.discarded);
  542. }
  543. compressCSS() {
  544. this.doc.querySelectorAll("style").forEach(styleElement => {
  545. if (styleElement) {
  546. styleElement.textContent = DOM.compressCSS(styleElement.textContent);
  547. }
  548. });
  549. this.doc.querySelectorAll("[style]").forEach(element => {
  550. element.setAttribute("style", DOM.compressCSS(element.getAttribute("style")));
  551. });
  552. }
  553. insertSingleFileComment() {
  554. const commentNode = this.doc.createComment("\n Archive processed by SingleFile \n url: " + this.baseURI + " \n saved date: " + new Date() + " \n");
  555. this.doc.documentElement.insertBefore(commentNode, this.doc.documentElement.firstChild);
  556. }
  557. replaceCanvasElements() {
  558. if (this.options.canvasData) {
  559. this.doc.querySelectorAll("canvas").forEach((canvasElement, indexCanvasElement) => {
  560. const canvasData = this.options.canvasData[indexCanvasElement];
  561. if (canvasData) {
  562. const imgElement = this.doc.createElement("img");
  563. imgElement.setAttribute("src", canvasData.dataURI);
  564. Array.from(canvasElement.attributes).forEach(attribute => {
  565. if (attribute.value) {
  566. imgElement.setAttribute(attribute.name, attribute.value);
  567. }
  568. });
  569. if (!imgElement.width && canvasData.width) {
  570. imgElement.style.pixelWidth = canvasData.width;
  571. }
  572. if (!imgElement.height && canvasData.height) {
  573. imgElement.style.pixelHeight = canvasData.height;
  574. }
  575. canvasElement.parentElement.replaceChild(imgElement, canvasElement);
  576. this.stats.add("processed", "canvas", 1);
  577. }
  578. });
  579. }
  580. }
  581. insertFonts() {
  582. if (this.options.fontsData && this.options.fontsData.length) {
  583. let stylesheetContent = "";
  584. this.options.fontsData.forEach(fontStyles => {
  585. if (fontStyles["font-family"] && fontStyles.src) {
  586. stylesheetContent += "@font-face{";
  587. let stylesContent = "";
  588. Object.keys(fontStyles).forEach(fontStyle => {
  589. if (stylesContent) {
  590. stylesContent += ";";
  591. }
  592. stylesContent += fontStyle + ":" + fontStyles[fontStyle];
  593. });
  594. stylesheetContent += stylesContent + "}";
  595. }
  596. });
  597. if (stylesheetContent) {
  598. const styleElement = this.doc.createElement("style");
  599. styleElement.textContent = stylesheetContent;
  600. const existingStyleElement = this.doc.querySelector("style");
  601. if (existingStyleElement) {
  602. existingStyleElement.parentElement.insertBefore(styleElement, existingStyleElement);
  603. } else {
  604. this.doc.head.insertBefore(styleElement, this.doc.head.firstChild);
  605. }
  606. }
  607. }
  608. }
  609. replaceStyleContents() {
  610. if (this.options.stylesheetContents) {
  611. this.doc.querySelectorAll("style").forEach((styleElement, styleIndex) => {
  612. if (this.options.stylesheetContents[styleIndex]) {
  613. styleElement.textContent = this.options.stylesheetContents[styleIndex];
  614. }
  615. });
  616. }
  617. }
  618. insertVideoPosters() {
  619. if (this.options.postersData) {
  620. this.doc.querySelectorAll("video[src], video > source[src]").forEach((videoElement, videoIndex) => {
  621. if (!videoElement.poster && this.options.postersData[videoIndex]) {
  622. videoElement.setAttribute("poster", this.options.postersData[videoIndex]);
  623. }
  624. });
  625. }
  626. }
  627. async pageResources() {
  628. const resourcePromises = [
  629. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("link[href][rel*=\"icon\"]"), "href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  630. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("object[type=\"image/svg+xml\"], object[type=\"image/svg-xml\"]"), "data", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest),
  631. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("img[src], input[src][type=image]"), "src", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  632. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("embed[src*=\".svg\"]"), "src", PREFIX_DATA_URI_IMAGE_SVG, this.baseURI, this.batchRequest),
  633. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[poster]"), "poster", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  634. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("*[background]"), "background", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  635. DomProcessorHelper.processAttribute(this.doc.querySelectorAll("image"), "xlink:href", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest),
  636. DomProcessorHelper.processXLinks(this.doc.querySelectorAll("use"), this.baseURI, this.batchRequest),
  637. DomProcessorHelper.processSrcset(this.doc.querySelectorAll("img[srcset], source[srcset]"), "srcset", PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)
  638. ];
  639. if (!this.options.removeAudioSrc) {
  640. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("audio[src], audio > source[src]"), "src", PREFIX_DATA_URI_AUDIO, this.baseURI, this.batchRequest));
  641. }
  642. if (!this.options.removeVideoSrc) {
  643. resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll("video[src], video > source[src]"), "src", PREFIX_DATA_URI_VIDEO, this.baseURI, this.batchRequest));
  644. }
  645. if (this.options.lazyLoadImages) {
  646. const imageSelectors = DOM.lazyLoaderImageSelectors();
  647. Object.keys(imageSelectors.src).forEach(selector => resourcePromises.push(DomProcessorHelper.processAttribute(this.doc.querySelectorAll(selector), imageSelectors.src[selector], PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)));
  648. Object.keys(imageSelectors.srcset).forEach(selector => resourcePromises.push(DomProcessorHelper.processSrcset(this.doc.querySelectorAll(selector), imageSelectors.srcset[selector], PREFIX_DATA_URI_IMAGE, this.baseURI, this.batchRequest)));
  649. }
  650. await resourcePromises;
  651. }
  652. async inlineStylesheets(initialization) {
  653. await Promise.all(Array.from(this.doc.querySelectorAll("style")).map(async (styleElement, indexStyle) => {
  654. if (!initialization) {
  655. this.stats.add("processed", "stylesheets", 1);
  656. }
  657. let stylesheetContent = styleElement.textContent;
  658. if (initialization) {
  659. stylesheetContent = await DomProcessorHelper.resolveImportURLs(styleElement.textContent, this.baseURI, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  660. } else {
  661. stylesheetContent = await DomProcessorHelper.processStylesheet(styleElement.textContent, this.baseURI, this.options, false, indexStyle, this.batchRequest);
  662. }
  663. styleElement.textContent = stylesheetContent;
  664. }));
  665. }
  666. async scripts() {
  667. await Promise.all(Array.from(this.doc.querySelectorAll("script[src]")).map(async scriptElement => {
  668. if (scriptElement.src) {
  669. this.stats.add("processed", "scripts", 1);
  670. const scriptContent = await Download.getContent(scriptElement.src, { asDataURI: false, maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  671. scriptElement.textContent = scriptContent.replace(/<\/script>/gi, "<\\/script>");
  672. }
  673. scriptElement.removeAttribute("src");
  674. }));
  675. }
  676. async frames(initialization) {
  677. if (this.options.framesData) {
  678. const frameElements = Array.from(this.doc.querySelectorAll("iframe, frame, object[type=\"text/html\"][data]"));
  679. await Promise.all(frameElements.map(async frameElement => {
  680. DomProcessorHelper.setFrameEmptySrc(frameElement);
  681. frameElement.setAttribute("sandbox", "allow-scripts allow-same-origin");
  682. const frameWindowId = frameElement.getAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  683. if (frameWindowId) {
  684. const frameData = this.options.framesData.find(frame => frame.windowId == frameWindowId);
  685. if (frameData) {
  686. if (initialization) {
  687. const options = Object.create(this.options);
  688. options.insertSingleFileComment = false;
  689. options.insertFaviconLink = false;
  690. options.doc = null;
  691. options.win = null;
  692. options.url = frameData.baseURI;
  693. options.windowId = frameWindowId;
  694. if (frameData.content) {
  695. options.content = frameData.content;
  696. options.canvasData = frameData.canvasData;
  697. options.stylesheetContents = frameData.stylesheetContents;
  698. options.currentSrcImages = frameData.currentSrcImages;
  699. frameData.processor = new PageProcessor(options);
  700. frameData.frameElement = frameElement;
  701. await frameData.processor.loadPage();
  702. await frameData.processor.initialize();
  703. frameData.maxResources = this.batchRequest.getMaxResources();
  704. }
  705. } else {
  706. if (frameData.processor) {
  707. this.stats.add("processed", "frames", 1);
  708. await frameData.processor.preparePageData();
  709. const pageData = await frameData.processor.getPageData();
  710. frameElement.removeAttribute(DOM.windowIdAttributeName(this.options.sessionId));
  711. DomProcessorHelper.setFrameContent(frameElement, pageData.content);
  712. this.stats.addAll(pageData);
  713. } else {
  714. this.stats.add("discarded", "frames", 1);
  715. }
  716. }
  717. }
  718. }
  719. }));
  720. }
  721. }
  722. async htmlImports(initialization) {
  723. const linkElements = Array.from(this.doc.querySelectorAll("link[rel=import][href]"));
  724. if (!this.relImportProcessors) {
  725. this.relImportProcessors = new Map();
  726. }
  727. await Promise.all(linkElements.map(async linkElement => {
  728. if (initialization) {
  729. const resourceURL = linkElement.href;
  730. const options = Object.create(this.options);
  731. options.insertSingleFileComment = false;
  732. options.insertFaviconLink = false;
  733. options.doc = null;
  734. options.win = null;
  735. options.url = resourceURL;
  736. if (resourceURL) {
  737. if (resourceURL && resourceURL != this.baseURI && DomUtil.testValidPath(resourceURL)) {
  738. const processor = new PageProcessor(options);
  739. this.relImportProcessors.set(linkElement, processor);
  740. await processor.loadPage();
  741. return processor.initialize();
  742. }
  743. }
  744. } else {
  745. linkElement.setAttribute("href", EMPTY_DATA_URI);
  746. const processor = this.relImportProcessors.get(linkElement);
  747. if (processor) {
  748. this.stats.add("processed", "HTML imports", 1);
  749. this.relImportProcessors.delete(linkElement);
  750. const pageData = await processor.getPageData();
  751. linkElement.setAttribute("href", "data:text/html," + pageData.content);
  752. this.stats.addAll(pageData);
  753. } else {
  754. this.stats.add("discarded", "HTML imports", 1);
  755. }
  756. }
  757. }));
  758. }
  759. async attributeStyles(initialization) {
  760. await Promise.all(Array.from(this.doc.querySelectorAll("[style]")).map(async element => {
  761. let stylesheetContent = element.getAttribute("style");
  762. if (initialization) {
  763. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, this.baseURI);
  764. } else {
  765. stylesheetContent = await DomProcessorHelper.processStylesheet(element.getAttribute("style"), this.baseURI, this.options, true, 0, this.batchRequest);
  766. }
  767. element.setAttribute("style", stylesheetContent);
  768. }));
  769. }
  770. async linkStylesheets() {
  771. await Promise.all(Array.from(this.doc.querySelectorAll("link[rel*=stylesheet]")).map(async linkElement => {
  772. const stylesheetContent = await DomProcessorHelper.resolveLinkStylesheetURLs(linkElement.href, this.baseURI, linkElement.media, { maxResourceSize: this.options.maxResourceSize, maxResourceSizeEnabled: this.options.maxResourceSizeEnabled });
  773. const styleElement = this.doc.createElement("style");
  774. styleElement.textContent = stylesheetContent;
  775. linkElement.parentElement.replaceChild(styleElement, linkElement);
  776. }));
  777. }
  778. }
  779. // ---------
  780. // DomHelper
  781. // ---------
  782. const REGEXP_AMP = /&/g;
  783. const REGEXP_NBSP = /\u00a0/g;
  784. const REGEXP_START_TAG = /</g;
  785. const REGEXP_END_TAG = />/g;
  786. const REGEXP_URL_HASH = /(#.+?)$/;
  787. const PREFIX_DATA_URI_IMAGE = "data:image/";
  788. const PREFIX_DATA_URI_AUDIO = "data:audio/";
  789. const PREFIX_DATA_URI_VIDEO = "data:video/";
  790. const PREFIX_DATA_URI_IMAGE_SVG = "data:image/svg+xml";
  791. const PREFIX_DATA_URI_NO_MIMETYPE = "data:;";
  792. class DomProcessorHelper {
  793. static async getFilename(options, content) {
  794. let filename = options.filenameTemplate;
  795. const date = new Date();
  796. const url = new URL(options.url);
  797. filename = await DomUtil.evalTemplateVariable(filename, "page-title", () => options.title || "No title");
  798. filename = await DomUtil.evalTemplateVariable(filename, "page-language", () => options.info.lang || "No language");
  799. filename = await DomUtil.evalTemplateVariable(filename, "page-description", () => options.info.description || "No description");
  800. filename = await DomUtil.evalTemplateVariable(filename, "page-author", () => options.info.author || "No author");
  801. filename = await DomUtil.evalTemplateVariable(filename, "page-creator", () => options.info.creator || "No creator");
  802. filename = await DomUtil.evalTemplateVariable(filename, "page-publisher", () => options.info.publisher || "No publisher");
  803. filename = await DomUtil.evalTemplateVariable(filename, "datetime-iso", () => date.toISOString());
  804. filename = await DomUtil.evalTemplateVariable(filename, "date-iso", () => date.toISOString().split("T")[0]);
  805. filename = await DomUtil.evalTemplateVariable(filename, "time-iso", () => date.toISOString().split("T")[1].split("Z")[0]);
  806. filename = await DomUtil.evalTemplateVariable(filename, "date-locale", () => date.toLocaleDateString());
  807. filename = await DomUtil.evalTemplateVariable(filename, "time-locale", () => date.toLocaleTimeString());
  808. filename = await DomUtil.evalTemplateVariable(filename, "day-locale", () => String(date.getDate()));
  809. filename = await DomUtil.evalTemplateVariable(filename, "month-locale", () => String(date.getMonth()));
  810. filename = await DomUtil.evalTemplateVariable(filename, "year-locale", () => String(date.getFullYear()));
  811. filename = await DomUtil.evalTemplateVariable(filename, "datetime-locale", () => date.toLocaleString());
  812. filename = await DomUtil.evalTemplateVariable(filename, "datetime-utc", () => date.toUTCString());
  813. filename = await DomUtil.evalTemplateVariable(filename, "day-utc", () => String(date.getUTCDate()));
  814. filename = await DomUtil.evalTemplateVariable(filename, "month-utc", () => String(date.getUTCMonth()));
  815. filename = await DomUtil.evalTemplateVariable(filename, "year-utc", () => String(date.getUTCFullYear()));
  816. filename = await DomUtil.evalTemplateVariable(filename, "url-hash", () => url.hash.substring(1));
  817. filename = await DomUtil.evalTemplateVariable(filename, "url-host", () => url.host.replace(/\/$/, ""));
  818. filename = await DomUtil.evalTemplateVariable(filename, "url-hostname", () => url.hostname.replace(/\/$/, ""));
  819. filename = await DomUtil.evalTemplateVariable(filename, "url-href", () => url.href);
  820. filename = await DomUtil.evalTemplateVariable(filename, "url-password", () => url.password);
  821. filename = await DomUtil.evalTemplateVariable(filename, "url-pathname", () => url.pathname.replace(/^\//, "").replace(/\/$/, ""), true);
  822. filename = await DomUtil.evalTemplateVariable(filename, "url-port", () => url.port);
  823. filename = await DomUtil.evalTemplateVariable(filename, "url-protocol", () => url.protocol);
  824. filename = await DomUtil.evalTemplateVariable(filename, "url-search", () => url.search.substring(1));
  825. filename = await DomUtil.evalTemplateVariable(filename, "url-username", () => url.username);
  826. filename = await DomUtil.evalTemplateVariable(filename, "tab-id", () => String(options.tabId || "No tab id"));
  827. filename = await DomUtil.evalTemplateVariable(filename, "url-last-segment", () => DomUtil.getLastSegment(url));
  828. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-256", async () => DOM.digest("SHA-256", content));
  829. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-384", async () => DOM.digest("SHA-384", content));
  830. filename = await DomUtil.evalTemplateVariable(filename, "digest-sha-512", async () => DOM.digest("SHA-512", content));
  831. filename = filename.replace(/[~\\?%*:|"<>\x00-\x1f\x7F]+/g, "_"); // eslint-disable-line no-control-regex
  832. filename = filename.replace(/\.\.\//g, "").replace(/^\/+/, "").replace(/\/+/g, "/").replace(/\/$/, "");
  833. if (!options.backgroundSave) {
  834. filename = filename.replace(/\//g, "_");
  835. }
  836. if (filename.length > 192) {
  837. const extensionMatch = filename.match(/(\.[^.]{3,4})$/);
  838. const extension = extensionMatch && extensionMatch[0] && extensionMatch[0].length > 1 ? extensionMatch[0] : "";
  839. filename = filename.substring(0, 192 - extension.length) + "…" + extension;
  840. }
  841. if (!filename) {
  842. filename = "Unnamed page";
  843. }
  844. return filename;
  845. }
  846. static setFrameEmptySrc(frameElement) {
  847. if (frameElement.tagName == "OBJECT") {
  848. frameElement.setAttribute("data", "data:text/html,");
  849. } else {
  850. frameElement.setAttribute("srcdoc", "");
  851. frameElement.removeAttribute("src");
  852. }
  853. }
  854. static setFrameContent(frameElement, content) {
  855. if (frameElement.tagName == "OBJECT") {
  856. frameElement.setAttribute("data", "data:text/html," + content);
  857. } else {
  858. frameElement.setAttribute("srcdoc", content);
  859. frameElement.removeAttribute("src");
  860. }
  861. }
  862. static isolateElements(rootElement) {
  863. rootElement.querySelectorAll("*").forEach(element => {
  864. if (element.getAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME) == "") {
  865. element.removeAttribute(SELECTED_CONTENT_ATTRIBUTE_NAME);
  866. } else if (!element.querySelector("[" + SELECTED_CONTENT_ATTRIBUTE_NAME + "]")) {
  867. element.remove();
  868. }
  869. });
  870. isolateParentElements(rootElement.parentElement, rootElement);
  871. function isolateParentElements(parentElement, element) {
  872. if (parentElement) {
  873. Array.from(parentElement.childNodes).forEach(node => {
  874. if (node != element && node.tagName != "HEAD" && node.tagName != "STYLE") {
  875. node.remove();
  876. }
  877. });
  878. }
  879. element = element.parentElement;
  880. if (element && element.parentElement) {
  881. isolateParentElements(element.parentElement, element);
  882. }
  883. }
  884. }
  885. static async resolveImportURLs(stylesheetContent, baseURI, options) {
  886. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  887. stylesheetContent = DomUtil.removeCssComments(stylesheetContent);
  888. const imports = DomUtil.getImportFunctions(stylesheetContent);
  889. await Promise.all(imports.map(async cssImport => {
  890. const match = DomUtil.matchImport(cssImport);
  891. if (match) {
  892. const resourceURL = DomUtil.normalizeURL(match.resourceURL);
  893. if (resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  894. const styleSheetUrl = new URL(match.resourceURL, baseURI).href;
  895. let importedStylesheetContent = await Download.getContent(styleSheetUrl, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  896. importedStylesheetContent = DomUtil.wrapMediaQuery(importedStylesheetContent, match.media);
  897. if (stylesheetContent.includes(cssImport)) {
  898. importedStylesheetContent = await DomProcessorHelper.resolveImportURLs(importedStylesheetContent, styleSheetUrl, options);
  899. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(cssImport), importedStylesheetContent);
  900. }
  901. }
  902. }
  903. }));
  904. return stylesheetContent;
  905. }
  906. static resolveStylesheetURLs(stylesheetContent, baseURI) {
  907. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  908. urlFunctions.map(urlFunction => {
  909. const originalResourceURL = DomUtil.matchURL(urlFunction);
  910. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  911. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  912. const resolvedURL = new URL(resourceURL, baseURI).href;
  913. if (resourceURL != resolvedURL && stylesheetContent.includes(urlFunction)) {
  914. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, resolvedURL));
  915. }
  916. } else {
  917. if (resourceURL.startsWith(DATA_URI_PREFIX)) {
  918. const escapedResourceURL = resourceURL.replace(REGEXP_AMP, "&amp;").replace(REGEXP_NBSP, "&nbsp;").replace(REGEXP_START_TAG, "&lt;").replace(REGEXP_END_TAG, "&gt;");
  919. if (escapedResourceURL != resourceURL && stylesheetContent.includes(urlFunction)) {
  920. stylesheetContent = stylesheetContent.replace(DomUtil.getRegExp(urlFunction), urlFunction.replace(originalResourceURL, escapedResourceURL));
  921. }
  922. }
  923. }
  924. });
  925. return stylesheetContent;
  926. }
  927. static async resolveLinkStylesheetURLs(resourceURL, baseURI, media, options) {
  928. resourceURL = DomUtil.normalizeURL(resourceURL);
  929. if (resourceURL && resourceURL != baseURI && resourceURL != ABOUT_BLANK_URI) {
  930. let stylesheetContent = await Download.getContent(resourceURL, { asDataURI: false, maxResourceSize: options.maxResourceSize, maxResourceSizeEnabled: options.maxResourceSizeEnabled });
  931. stylesheetContent = await DomProcessorHelper.resolveImportURLs(stylesheetContent, resourceURL, options);
  932. stylesheetContent = DomUtil.wrapMediaQuery(stylesheetContent, media);
  933. return stylesheetContent;
  934. }
  935. }
  936. static async processStylesheet(stylesheetContent, baseURI, options, inline, indexStyle, batchRequest) {
  937. stylesheetContent = DomProcessorHelper.resolveStylesheetURLs(stylesheetContent, baseURI);
  938. const urlFunctions = DomUtil.getUrlFunctions(stylesheetContent);
  939. let indexVariable = 0;
  940. await Promise.all(urlFunctions.map(async urlFunction => {
  941. const originalResourceURL = DomUtil.matchURL(urlFunction);
  942. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  943. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL) && stylesheetContent.includes(urlFunction)) {
  944. const dataURI = await batchRequest.addURL(resourceURL);
  945. const regExpUrlFunction = DomUtil.getRegExp(urlFunction);
  946. if (!inline && options.compressCSS && dataURI.startsWith(PREFIX_DATA_URI_IMAGE) && !dataURI.startsWith(PREFIX_DATA_URI_IMAGE_SVG)) {
  947. const functions = stylesheetContent.match(regExpUrlFunction);
  948. if (functions && functions.length > 1) {
  949. const variableName = "--single-file-" + indexStyle + "-" + indexVariable;
  950. stylesheetContent = variableName + ":url(\"" + dataURI + "\")" + (indexVariable ? ";" : "}") + stylesheetContent;
  951. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, "var(" + variableName + ")");
  952. indexVariable++;
  953. } else {
  954. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  955. }
  956. } else {
  957. stylesheetContent = stylesheetContent.replace(regExpUrlFunction, urlFunction.replace(originalResourceURL, dataURI));
  958. }
  959. }
  960. }));
  961. if (indexVariable) {
  962. stylesheetContent = ":root{" + stylesheetContent;
  963. }
  964. return stylesheetContent;
  965. }
  966. static async processAttribute(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  967. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  968. let resourceURL = resourceElement.getAttribute(attributeName);
  969. if (resourceURL) {
  970. resourceURL = DomUtil.normalizeURL(resourceURL);
  971. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  972. try {
  973. const dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  974. if (dataURI.startsWith(prefixDataURI) || dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE)) {
  975. resourceElement.setAttribute(attributeName, dataURI);
  976. } else {
  977. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  978. }
  979. } catch (error) {
  980. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  981. }
  982. }
  983. }
  984. }));
  985. }
  986. static async processXLinks(resourceElements, baseURI, batchRequest) {
  987. const attributeName = "xlink:href";
  988. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  989. const originalResourceURL = resourceElement.getAttribute(attributeName);
  990. if (originalResourceURL) {
  991. const resourceURL = DomUtil.normalizeURL(originalResourceURL);
  992. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  993. try {
  994. const content = await batchRequest.addURL(new URL(resourceURL, baseURI).href, false);
  995. const DOMParser = DOM.getParser();
  996. if (DOMParser) {
  997. const svgDoc = new DOMParser().parseFromString(content, "image/svg+xml");
  998. const hashMatch = originalResourceURL.match(REGEXP_URL_HASH);
  999. if (hashMatch && hashMatch[0]) {
  1000. const symbolElement = svgDoc.querySelector(hashMatch[0]);
  1001. if (symbolElement) {
  1002. resourceElement.setAttribute(attributeName, hashMatch[0]);
  1003. resourceElement.parentElement.appendChild(symbolElement);
  1004. }
  1005. } else {
  1006. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1007. }
  1008. } else {
  1009. resourceElement.setAttribute(attributeName, "data:image/svg+xml," + content);
  1010. }
  1011. } catch (error) {
  1012. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1013. }
  1014. }
  1015. }
  1016. }));
  1017. }
  1018. static async processSrcset(resourceElements, attributeName, prefixDataURI, baseURI, batchRequest) {
  1019. await Promise.all(Array.from(resourceElements).map(async resourceElement => {
  1020. const srcset = DOM.parseSrcset(resourceElement.getAttribute(attributeName));
  1021. const srcsetValues = await Promise.all(srcset.map(async srcsetValue => {
  1022. const resourceURL = DomUtil.normalizeURL(srcsetValue.url);
  1023. if (resourceURL && resourceURL != baseURI && DomUtil.testValidPath(resourceURL)) {
  1024. try {
  1025. let dataURI = await batchRequest.addURL(new URL(resourceURL, baseURI).href);
  1026. if (!dataURI.startsWith(prefixDataURI) && !dataURI.startsWith(PREFIX_DATA_URI_NO_MIMETYPE)) {
  1027. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1028. }
  1029. return dataURI + (srcsetValue.w ? " " + srcsetValue.w + "w" : srcsetValue.d ? " " + srcsetValue.d + "x" : "");
  1030. } catch (error) {
  1031. resourceElement.setAttribute(attributeName, EMPTY_IMAGE);
  1032. }
  1033. }
  1034. }));
  1035. resourceElement.setAttribute(attributeName, srcsetValues.join(", "));
  1036. }));
  1037. }
  1038. }
  1039. // -------
  1040. // DomUtil
  1041. // -------
  1042. const DATA_URI_PREFIX = "data:";
  1043. const BLOB_URI_PREFIX = "blob:";
  1044. const HTTP_URI_PREFIX = /^https?:\/\//;
  1045. const ABOUT_BLANK_URI = "about:blank";
  1046. const NOT_EMPTY_URL = /^https?:\/\/.+/;
  1047. const REGEXP_URL_FN = /(url\s*\(\s*'(.*?)'\s*\))|(url\s*\(\s*"(.*?)"\s*\))|(url\s*\(\s*(.*?)\s*\))/gi;
  1048. const REGEXP_URL_SIMPLE_QUOTES_FN = /^url\s*\(\s*'(.*?)'\s*\)$/i;
  1049. const REGEXP_URL_DOUBLE_QUOTES_FN = /^url\s*\(\s*"(.*?)"\s*\)$/i;
  1050. const REGEXP_URL_NO_QUOTES_FN = /^url\s*\(\s*(.*?)\s*\)$/i;
  1051. 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;
  1052. const REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN = /@import\s*url\s*\(\s*'(.*?)'\s*\)\s*(.*?)(;|$|})/i;
  1053. const REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN = /@import\s*url\s*\(\s*"(.*?)"\s*\)\s*(.*?)(;|$|})/i;
  1054. const REGEXP_IMPORT_URL_NO_QUOTES_FN = /@import\s*url\s*\(\s*(.*?)\s*\)\s*(.*?)(;|$|})/i;
  1055. const REGEXP_IMPORT_SIMPLE_QUOTES_FN = /@import\s*'(.*?)'\s*(.*?)(;|$|})/i;
  1056. const REGEXP_IMPORT_DOUBLE_QUOTES_FN = /@import\s*"(.*?)"\s*(.*?)(;|$|})/i;
  1057. const REGEXP_IMPORT_NO_QUOTES_FN = /@import\s*(.*?)\s*(.*?)(;|$|})/i;
  1058. const REGEXP_ESCAPE = /([{}()^$&.*?/+|[\\\\]|\]|-)/g;
  1059. class DomUtil {
  1060. static normalizeURL(url) {
  1061. if (url.startsWith(DATA_URI_PREFIX)) {
  1062. return url;
  1063. } else {
  1064. return url.split("#")[0];
  1065. }
  1066. }
  1067. static async evalTemplateVariable(template, variableName, valueGetter, dontReplaceSlash) {
  1068. const replaceRegExp = new RegExp("{\\s*" + variableName + "\\s*}", "g");
  1069. if (template.match(replaceRegExp)) {
  1070. let value = await valueGetter();
  1071. if (!dontReplaceSlash) {
  1072. value = value.replace(/\/+/g, "_");
  1073. }
  1074. return template.replace(replaceRegExp, value);
  1075. }
  1076. return template;
  1077. }
  1078. static getLastSegment(url) {
  1079. let lastSegmentMatch = url.pathname.match(/\/([^/]+)$/);
  1080. let lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1081. if (!lastSegment) {
  1082. lastSegmentMatch = url.href.match(/([^/]+)\/?$/);
  1083. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1084. }
  1085. if (!lastSegment) {
  1086. lastSegmentMatch = lastSegment.match(/(.*)<\.[^.]+$/);
  1087. lastSegment = lastSegmentMatch && lastSegmentMatch[0];
  1088. }
  1089. if (!lastSegment) {
  1090. lastSegment = url.hostname.replace(/\/+/g, "_").replace(/\/$/, "");
  1091. }
  1092. lastSegment.replace(/\/$/, "").replace(/^\//, "");
  1093. return lastSegment;
  1094. }
  1095. static getRegExp(string) {
  1096. return new RegExp(string.replace(REGEXP_ESCAPE, "\\$1"), "gi");
  1097. }
  1098. static getUrlFunctions(stylesheetContent) {
  1099. return Array.from(new Set(stylesheetContent.match(REGEXP_URL_FN) || []));
  1100. }
  1101. static getImportFunctions(stylesheetContent) {
  1102. return stylesheetContent.match(REGEXP_IMPORT_FN) || [];
  1103. }
  1104. static matchURL(stylesheetContent) {
  1105. const match = stylesheetContent.match(REGEXP_URL_SIMPLE_QUOTES_FN) ||
  1106. stylesheetContent.match(REGEXP_URL_DOUBLE_QUOTES_FN) ||
  1107. stylesheetContent.match(REGEXP_URL_NO_QUOTES_FN);
  1108. return match && match[1];
  1109. }
  1110. static testValidPath(resourceURL) {
  1111. return !resourceURL.startsWith(DATA_URI_PREFIX) && !resourceURL.startsWith(BLOB_URI_PREFIX) && resourceURL != ABOUT_BLANK_URI && (!resourceURL.match(HTTP_URI_PREFIX) || resourceURL.match(NOT_EMPTY_URL));
  1112. }
  1113. static matchImport(stylesheetContent) {
  1114. const match = stylesheetContent.match(REGEXP_IMPORT_URL_SIMPLE_QUOTES_FN) ||
  1115. stylesheetContent.match(REGEXP_IMPORT_URL_DOUBLE_QUOTES_FN) ||
  1116. stylesheetContent.match(REGEXP_IMPORT_URL_NO_QUOTES_FN) ||
  1117. stylesheetContent.match(REGEXP_IMPORT_SIMPLE_QUOTES_FN) ||
  1118. stylesheetContent.match(REGEXP_IMPORT_DOUBLE_QUOTES_FN) ||
  1119. stylesheetContent.match(REGEXP_IMPORT_NO_QUOTES_FN);
  1120. if (match) {
  1121. const [, resourceURL, media] = match;
  1122. return { resourceURL, media };
  1123. }
  1124. }
  1125. static removeCssComments(stylesheetContent) {
  1126. let start, end;
  1127. do {
  1128. start = stylesheetContent.indexOf("/*");
  1129. end = stylesheetContent.indexOf("*/", start + 2);
  1130. if (start != -1 && end != -1) {
  1131. stylesheetContent = stylesheetContent.substring(0, start) + stylesheetContent.substr(end + 2);
  1132. }
  1133. } while (start != -1 && end != -1);
  1134. return stylesheetContent;
  1135. }
  1136. static wrapMediaQuery(stylesheetContent, mediaQuery) {
  1137. if (mediaQuery) {
  1138. return "@media " + mediaQuery + "{ " + stylesheetContent + " }";
  1139. } else {
  1140. return stylesheetContent;
  1141. }
  1142. }
  1143. }
  1144. function log(...args) {
  1145. console.log("S-File <core> ", ...args); // eslint-disable-line no-console
  1146. }
  1147. // -----
  1148. // Stats
  1149. // -----
  1150. const STATS_DEFAULT_VALUES = {
  1151. discarded: {
  1152. "HTML bytes": 0,
  1153. "hidden elements": 0,
  1154. "HTML imports": 0,
  1155. scripts: 0,
  1156. objects: 0,
  1157. "audio sources": 0,
  1158. "video sources": 0,
  1159. frames: 0,
  1160. "CSS rules": 0,
  1161. canvas: 0,
  1162. stylesheets: 0,
  1163. resources: 0,
  1164. medias: 0
  1165. },
  1166. processed: {
  1167. "HTML bytes": 0,
  1168. "hidden elements": 0,
  1169. "HTML imports": 0,
  1170. scripts: 0,
  1171. objects: 0,
  1172. "audio sources": 0,
  1173. "video sources": 0,
  1174. frames: 0,
  1175. "CSS rules": 0,
  1176. canvas: 0,
  1177. stylesheets: 0,
  1178. resources: 0,
  1179. medias: 0
  1180. }
  1181. };
  1182. class Stats {
  1183. constructor(options) {
  1184. this.options = options;
  1185. if (options.displayStats) {
  1186. this.data = JSON.parse(JSON.stringify(STATS_DEFAULT_VALUES));
  1187. }
  1188. }
  1189. set(type, subType, value) {
  1190. if (this.options.displayStats) {
  1191. this.data[type][subType] = value;
  1192. }
  1193. }
  1194. add(type, subType, value) {
  1195. if (this.options.displayStats) {
  1196. this.data[type][subType] += value;
  1197. }
  1198. }
  1199. addAll(pageData) {
  1200. if (this.options.displayStats) {
  1201. Object.keys(this.data.discarded).forEach(key => this.add("discarded", key, pageData.stats.discarded[key] || 0));
  1202. Object.keys(this.data.processed).forEach(key => this.add("processed", key, pageData.stats.processed[key] || 0));
  1203. }
  1204. }
  1205. }
  1206. return { getClass };
  1207. })();