docprocessor.js 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463
  1. /*
  2. * Copyright 2011 Gildas Lormeau
  3. * contact : gildas.lormeau <at> gmail.com
  4. *
  5. * This file is part of SingleFile Core.
  6. *
  7. * SingleFile Core 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 Core 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 Core. If not, see <http://www.gnu.org/licenses/>.
  19. */
  20. (function() {
  21. var IMPORT_URL_VALUE_EXP = /(url\s*\(\s*(?:'|")?\s*([^('"\))]*)\s*(?:'|")?\s*\))|(@import\s*\(?\s*(?:'|")?\s*([^('"\))]*)\s*(?:'|")?\s*(?:\)|;))/i;
  22. var URL_VALUE_EXP = /url\s*\(\s*(?:'|")?\s*([^('"\))]*)\s*(?:'|")?\s*\)/i;
  23. var IMPORT_VALUE_ALT_EXP = /@import\s*\(?\s*(?:'|")?\s*([^('"\))]*)\s*(?:'|")?\s*(?:\)|;)/i;
  24. var URL_EXP = /url\s*\(([^\)]*)\)/gi;
  25. var IMPORT_EXP = /(@import\s*url\s*\([^\)]*\)\s*;?)|(@import\s*('|")?\s*[^\(;'"]*\s*('|")?\s*;)/gi;
  26. var IMPORT_ALT_EXP = /@import\s*('|")?\s*[^\(;'"]*\s*('|")?\s*;/gi;
  27. var EMPTY_PIXEL_DATA = "data:image/gif;base64,R0lGODlhAQABAIAAAP///////yH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==";
  28. function decodeDataURI(dataURI) {
  29. var content = dataURI.indexOf(","), meta = dataURI.substr(5, content).toLowerCase(), data = decodeURIComponent(dataURI.substr(content + 1));
  30. if (/;\s*base64\s*[;,]/.test(meta)) {
  31. data = atob(data);
  32. }
  33. if (/;\s*charset=[uU][tT][fF]-?8\s*[;,]/.test(meta)) {
  34. data = decodeURIComponent(escape(data));
  35. }
  36. return data;
  37. }
  38. function formatURL(link, host) {
  39. var i, newlinkparts, hparts, lparts;
  40. if (!link)
  41. return "";
  42. lparts = link.split('/');
  43. host = host.split("#")[0].split("?")[0];
  44. if (/http:|https:|ftp:|data:|javascript:/i.test(lparts[0]))
  45. return link.trim();
  46. hparts = host.split('/');
  47. newlinkparts = [];
  48. if (hparts.length > 3)
  49. hparts.pop();
  50. if (lparts[0] == '') {
  51. if (lparts[1] == '')
  52. host = hparts[0] + '//' + lparts[2];
  53. else
  54. host = hparts[0] + '//' + hparts[2];
  55. hparts = host.split('/');
  56. delete lparts[0];
  57. if (lparts[1] == '') {
  58. delete lparts[1];
  59. delete lparts[2];
  60. }
  61. }
  62. for (i = 0; i < lparts.length; i++) {
  63. if (lparts[i] == '..') {
  64. if (lparts[i - 1])
  65. delete lparts[i - 1];
  66. else if (hparts.length > 3)
  67. hparts.pop();
  68. delete lparts[i];
  69. }
  70. if (lparts[i] == '.')
  71. delete lparts[i];
  72. }
  73. for (i = 0; i < lparts.length; i++)
  74. if (lparts[i])
  75. newlinkparts[newlinkparts.length] = lparts[i];
  76. return (hparts.join('/') + '/' + newlinkparts.join('/')).trim();
  77. }
  78. function resolveURLs(content, host) {
  79. var ret = content.replace(URL_EXP, function(value) {
  80. var result = value.match(URL_VALUE_EXP);
  81. if (result)
  82. if (result[1].indexOf("data:") != 0)
  83. return value.replace(result[1], formatURL(result[1], host));
  84. return value;
  85. });
  86. return ret.replace(IMPORT_ALT_EXP, function(value) {
  87. var result = value.match(IMPORT_VALUE_ALT_EXP);
  88. if (result)
  89. if (result[1].indexOf("data:") != 0)
  90. return "@import \"" + formatURL(result[1], host) + "\";";
  91. return value;
  92. });
  93. }
  94. function getDataURI(data, defaultURL, woURL) {
  95. if (data.content)
  96. return (woURL ? "" : "url(") + "data:" + data.mediaType + ";" + data.mediaTypeParam + "," + data.content + (woURL ? "" : ")");
  97. else
  98. return woURL ? defaultURL : "url(" + defaultURL + ")";
  99. }
  100. function removeComments(content) {
  101. var start, end;
  102. do {
  103. start = content.indexOf("/*");
  104. end = content.indexOf("*/", start);
  105. if (start != -1 && end != -1)
  106. content = content.substring(0, start) + content.substr(end + 2);
  107. } while (start != -1 && end != -1);
  108. return content;
  109. }
  110. function replaceURLs(content, host, requestManager, callback) {
  111. var i, url, result, values = removeComments(content).match(URL_EXP), requestMax = 0, requestIndex = 0;
  112. function sendRequest(origUrl) {
  113. requestMax++;
  114. requestManager.send(url, function(data) {
  115. requestIndex++;
  116. if (content.indexOf(origUrl) != -1) {
  117. data.mediaType = data.mediaType ? data.mediaType.split(";")[0] : null;
  118. content = content.replace(new RegExp(origUrl.replace(/([{}\(\)\^$&.\*\?\/\+\|\[\\\\]|\]|\-)/g, "\\$1"), "gi"), getDataURI(data,
  119. EMPTY_PIXEL_DATA, true));
  120. }
  121. if (requestIndex == requestMax)
  122. callback(content);
  123. }, null, "base64");
  124. }
  125. if (values)
  126. for (i = 0; i < values.length; i++) {
  127. result = values[i].match(URL_VALUE_EXP);
  128. if (result && result[1]) {
  129. url = formatURL(result[1], host);
  130. if (url.indexOf("data:") != 0)
  131. sendRequest(result[1]);
  132. }
  133. }
  134. }
  135. // ----------------------------------------------------------------------------------------------
  136. function processStylesheets(doc, docElement, baseURI, requestManager) {
  137. Array.prototype.forEach.call(docElement.querySelectorAll('link[href][rel*="stylesheet"]'), function(node) {
  138. var href = node.getAttribute("href"), url = formatURL(href, baseURI);
  139. function createStyleNode(content) {
  140. var i, newNode, commentNode;
  141. newNode = doc.createElement("style");
  142. for (i = 0; i < node.attributes.length; i++)
  143. if (node.attributes[i].value)
  144. newNode.setAttribute(node.attributes[i].name, node.attributes[i].value);
  145. newNode.dataset.href = url;
  146. newNode.removeAttribute("href");
  147. newNode.textContent = resolveURLs(content, url);
  148. if (node.disabled) {
  149. commentNode = doc.createComment();
  150. commentNode.textContent = newNode.outerHTML.replace(/</g, "&lt;").replace(/>/g, "&gt;").replace(/--/g, "&minus;&minus;");
  151. node.parentElement.replaceChild(commentNode, node);
  152. } else
  153. node.parentElement.replaceChild(newNode, node);
  154. }
  155. if (href.indexOf("data:") != 0)
  156. requestManager.send(url, function(data) {
  157. if (data.status >= 400)
  158. node.parentElement.removeChild(node);
  159. else
  160. createStyleNode(data.content || "");
  161. });
  162. else
  163. createStyleNode(decodeDataURI(href));
  164. });
  165. }
  166. function processImports(docElement, baseURI, characterSet, requestManager) {
  167. var ret = true;
  168. Array.prototype.forEach.call(docElement.querySelectorAll("style"), function(styleSheet) {
  169. var imports = removeComments(styleSheet.textContent).match(IMPORT_EXP);
  170. if (imports)
  171. imports.forEach(function(imp) {
  172. var url, href, result = imp.match(IMPORT_URL_VALUE_EXP);
  173. function insertStylesheet(content) {
  174. styleSheet.textContent = styleSheet.textContent.replace(imp, resolveURLs(content, url));
  175. }
  176. if (result && (result[2] || result[4])) {
  177. href = result[2] || result[4];
  178. url = formatURL(href, styleSheet.dataset.href || baseURI);
  179. if (href.indexOf("data:") != 0) {
  180. requestManager.send(url, function(data) {
  181. insertStylesheet(data.status < 400 && data.content ? data.content : "");
  182. }, null, characterSet);
  183. } else
  184. insertStylesheet(decodeDataURI(href));
  185. ret = false;
  186. }
  187. });
  188. });
  189. return ret;
  190. }
  191. function processStyleAttributes(docElement, baseURI, requestManager) {
  192. Array.prototype.forEach.call(docElement.querySelectorAll("*[style]"), function(node) {
  193. replaceURLs(node.getAttribute("style"), baseURI, requestManager, function(style) {
  194. node.setAttribute("style", style);
  195. });
  196. });
  197. }
  198. function processBgAttributes(docElement, baseURI, requestManager) {
  199. var backgrounds = docElement.querySelectorAll("*[background]");
  200. Array.prototype.forEach.call(backgrounds, function(node) {
  201. var url, value = node.getAttribute("background");
  202. if (value.indexOf(".") != -1) {
  203. url = formatURL(value, baseURI);
  204. if (url.indexOf("data:") != 0)
  205. requestManager.send(url, function(data) {
  206. node.setAttribute("background", getDataURI(data, EMPTY_PIXEL_DATA, true));
  207. }, null, "base64");
  208. }
  209. });
  210. }
  211. function insertDefaultFavico(doc, docElement, baseURI) {
  212. var node, docHead = docElement.querySelector("html > head"), favIcon = docElement
  213. .querySelector('link[href][rel="shortcut icon"], link[href][rel="apple-touch-icon"], link[href][rel="icon"]');
  214. if (!favIcon && docHead) {
  215. node = doc.createElement("link");
  216. node.setAttribute("type", "image/x-icon");
  217. node.setAttribute("rel", "shortcut icon");
  218. node.setAttribute("href", formatURL("/favicon.ico", baseURI));
  219. docHead.appendChild(node);
  220. }
  221. }
  222. function processImages(docElement, baseURI, requestManager) {
  223. var images;
  224. function process(attributeName) {
  225. Array.prototype.forEach.call(images, function(node) {
  226. var url = formatURL(node.getAttribute(attributeName), baseURI);
  227. if (url.indexOf("data:") != 0)
  228. requestManager.send(url, function(data) {
  229. node.setAttribute(attributeName, getDataURI(data, EMPTY_PIXEL_DATA, true));
  230. }, null, "base64");
  231. });
  232. }
  233. images = docElement.querySelectorAll('link[href][rel="shortcut icon"], link[href][rel="apple-touch-icon"], link[href][rel="icon"]');
  234. process("href");
  235. images = docElement.querySelectorAll('img[src], input[src][type="image"]');
  236. process("src");
  237. images = docElement.querySelectorAll('video[poster]');
  238. process("poster");
  239. }
  240. function processSVGs(docElement, baseURI, requestManager) {
  241. var images = docElement.querySelectorAll('object[type="image/svg+xml"], object[type="image/svg-xml"], embed[src*=".svg"]');
  242. Array.prototype.forEach.call(images, function(node) {
  243. var data = node.getAttribute("data"), src = node.getAttribute("src"), url = formatURL(data || src, baseURI);
  244. if (url.indexOf("data:") != 0)
  245. requestManager.send(url, function(data) {
  246. node.setAttribute(data ? "data" : "src", getDataURI(data, "data:text/xml,<svg></svg>", true));
  247. }, null, null);
  248. });
  249. }
  250. function processStyles(docElement, baseURI, requestManager) {
  251. Array.prototype.forEach.call(docElement.querySelectorAll("style"), function(styleSheet) {
  252. replaceURLs(styleSheet.textContent, styleSheet.dataset.href || baseURI, requestManager, function(textContent) {
  253. styleSheet.textContent = textContent;
  254. });
  255. });
  256. }
  257. function processScripts(docElement, baseURI, characterSet, requestManager) {
  258. Array.prototype.forEach.call(docElement.querySelectorAll("script[src]"), function(node) {
  259. var src = node.getAttribute("src");
  260. if (src.indexOf("data:") != 0)
  261. requestManager.send(formatURL(src, baseURI), function(data) {
  262. if (data.status < 400) {
  263. data.content = data.content.replace(/"([^"]*)<\/\s*script\s*>([^"]*)"/gi, '"$1<"+"/script>$2"');
  264. data.content = data.content.replace(/'([^']*)<\/\s*script\s*>([^']*)'/gi, "'$1<'+'/script>$2'");
  265. node.textContent = "\n" + data.content + "\n";
  266. }
  267. node.removeAttribute("src");
  268. }, characterSet);
  269. });
  270. }
  271. function processCanvas(doc, docElement, canvasData) {
  272. var index = 0;
  273. Array.prototype.forEach.call(docElement.querySelectorAll("canvas"), function(node) {
  274. var i, data = canvasData[index], newNode = doc.createElement("img");
  275. if (data) {
  276. newNode.setAttribute("src", data);
  277. for (i = 0; i < node.attributes.length; i++)
  278. if (node.attributes[i].value)
  279. newNode.setAttribute(node.attributes[i].name, node.attributes[i].value);
  280. if (!newNode.width)
  281. newNode.style.pixelWidth = node.clientWidth;
  282. if (!newNode.height)
  283. newNode.style.pixelHeight = node.clientHeight;
  284. node.parentElement.replaceChild(newNode, node);
  285. }
  286. index++;
  287. });
  288. }
  289. function removeScripts(docElement) {
  290. Array.prototype.forEach.call(docElement.querySelectorAll("script"), function(node) {
  291. node.parentElement.removeChild(node);
  292. });
  293. Array.prototype.forEach.call(docElement.querySelectorAll("*[onload]"), function(node) {
  294. node.removeAttribute("onload");
  295. });
  296. }
  297. function removeObjects(docElement) {
  298. var objects = docElement.querySelectorAll('applet, object:not([type="image/svg+xml"]):not([type="image/svg-xml"]), embed:not([src*=".svg"])');
  299. Array.prototype.forEach.call(objects, function(node) {
  300. node.parentElement.removeChild(node);
  301. });
  302. objects = docElement.querySelectorAll('audio[src], video[src]');
  303. Array.prototype.forEach.call(objects, function(node) {
  304. node.removeAttribute("src");
  305. });
  306. }
  307. function removeBlockquotesCite(docElement) {
  308. Array.prototype.forEach.call(docElement.querySelectorAll("blockquote[cite]"), function(node) {
  309. node.removeAttribute("cite");
  310. });
  311. }
  312. function removeFrames(docElement) {
  313. Array.prototype.forEach.call(docElement.querySelectorAll("iframe, frame"), function(node) {
  314. node.parentElement.removeChild(node);
  315. });
  316. }
  317. function removeMetaRefresh(docElement) {
  318. Array.prototype.forEach.call(docElement.querySelectorAll("meta[http-equiv=refresh]"), function(node) {
  319. node.parentElement.removeChild(node);
  320. });
  321. }
  322. function resetFrames(docElement, baseURI) {
  323. Array.prototype.forEach.call(docElement.querySelectorAll("iframe, frame"), function(node) {
  324. var src = formatURL(node.getAttribute("src"), baseURI);
  325. if (src.indexOf("data:") != 0)
  326. node.setAttribute("src", "about:blank");
  327. });
  328. }
  329. function setAbsoluteLinks(docElement, baseURI) {
  330. Array.prototype.forEach.call(docElement.querySelectorAll("a:not([href^='#'])"), function(link) {
  331. var fullHref = formatURL(link.getAttribute("href"), baseURI);
  332. if (fullHref && (!(fullHref.indexOf(baseURI.split("#")[0]) == 0) || fullHref.indexOf("#") == -1))
  333. link.setAttribute("href", fullHref);
  334. });
  335. }
  336. // ----------------------------------------------------------------------------------------------
  337. singlefile.initProcess = function(doc, docElement, addDefaultFavico, baseURI, characterSet, config, canvasData, requestManager, onInit, onProgress, onEnd) {
  338. var initManager = new RequestManager(), manager = new RequestManager(onProgress);
  339. function RequestManager(onProgress) {
  340. var that = this, currentCount = 0, requests = [];
  341. this.requestCount = 0;
  342. this.send = function(url, responseHandler, characterSet, mediaTypeParam) {
  343. this.requestCount++;
  344. requests.push({
  345. url : url,
  346. responseHandler : responseHandler,
  347. characterSet : characterSet,
  348. mediaTypeParam : mediaTypeParam
  349. });
  350. };
  351. this.doSend = function() {
  352. requests.forEach(function(request) {
  353. requestManager.send(request.url, function(response) {
  354. request.responseHandler(response);
  355. currentCount++;
  356. if (onProgress)
  357. onProgress(currentCount, that.requestCount);
  358. if (currentCount == that.requestCount) {
  359. that.requestCount = 0;
  360. currentCount = 0;
  361. if (that.onEnd)
  362. that.onEnd();
  363. }
  364. }, request.characterSet, request.mediaTypeParam);
  365. });
  366. requests = [];
  367. };
  368. }
  369. function cbImports() {
  370. if (config.removeScripts)
  371. removeScripts(docElement);
  372. if (config.removeObjects)
  373. removeObjects(docElement);
  374. if (config.removeFrames || config.getRawDoc)
  375. removeFrames(docElement);
  376. resetFrames(docElement, baseURI);
  377. removeBlockquotesCite(docElement);
  378. removeMetaRefresh(docElement);
  379. setAbsoluteLinks(docElement, baseURI);
  380. if (addDefaultFavico)
  381. insertDefaultFavico(doc, docElement, baseURI);
  382. processStyleAttributes(docElement, baseURI, manager);
  383. processBgAttributes(docElement, baseURI, manager);
  384. processImages(docElement, baseURI, manager);
  385. processSVGs(docElement, baseURI, manager);
  386. processStyles(docElement, baseURI, manager);
  387. processScripts(docElement, baseURI, characterSet, manager);
  388. processCanvas(doc, docElement, canvasData);
  389. if (onInit)
  390. setTimeout(function() {
  391. onInit(manager.requestCount);
  392. }, 1);
  393. }
  394. function cbStylesheets() {
  395. initManager.onEnd = function(noRequests) {
  396. if (noRequests)
  397. cbImports();
  398. else
  399. cbStylesheets();
  400. };
  401. processImports(docElement, baseURI, characterSet, initManager);
  402. initManager.doSend();
  403. if (initManager.requestCount == 0)
  404. cbImports();
  405. }
  406. manager.onEnd = onEnd;
  407. processStylesheets(doc, docElement, baseURI, initManager);
  408. initManager.onEnd = cbStylesheets;
  409. initManager.doSend();
  410. if (initManager.requestCount == 0)
  411. initManager.onEnd();
  412. return function() {
  413. manager.doSend();
  414. if (manager.onEnd && manager.requestCount == 0)
  415. manager.onEnd();
  416. };
  417. };
  418. })();