docprocessor.js 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465
  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()
  30. // 'data:'.length == 5
  31. , data = decodeURIComponent(dataURI.substr(content + 1));
  32. if (/;\s*base64\s*[;,]/.test(meta)) {
  33. data = atob(data); // decode base64
  34. }
  35. if (/;\s*charset=[uU][tT][fF]-?8\s*[;,]/.test(meta)) {
  36. data = decodeURIComponent(escape(data)); // decode UTF-8
  37. }
  38. return data;
  39. }
  40. function formatURL(link, host) {
  41. var i, newlinkparts, hparts, lparts;
  42. if (!link)
  43. return "";
  44. lparts = link.split('/');
  45. host = host.split("#")[0].split("?")[0];
  46. if (/http:|https:|ftp:|data:|javascript:/i.test(lparts[0]))
  47. return link.trim();
  48. hparts = host.split('/');
  49. newlinkparts = [];
  50. if (hparts.length > 3)
  51. hparts.pop();
  52. if (lparts[0] == '') {
  53. if (lparts[1] == '')
  54. host = hparts[0] + '//' + lparts[2];
  55. else
  56. host = hparts[0] + '//' + hparts[2];
  57. hparts = host.split('/');
  58. delete lparts[0];
  59. if (lparts[1] == '') {
  60. delete lparts[1];
  61. delete lparts[2];
  62. }
  63. }
  64. for (i = 0; i < lparts.length; i++) {
  65. if (lparts[i] == '..') {
  66. if (lparts[i - 1])
  67. delete lparts[i - 1];
  68. else if (hparts.length > 3)
  69. hparts.pop();
  70. delete lparts[i];
  71. }
  72. if (lparts[i] == '.')
  73. delete lparts[i];
  74. }
  75. for (i = 0; i < lparts.length; i++)
  76. if (lparts[i])
  77. newlinkparts[newlinkparts.length] = lparts[i];
  78. return (hparts.join('/') + '/' + newlinkparts.join('/')).trim();
  79. }
  80. function resolveURLs(content, host) {
  81. var ret = content.replace(URL_EXP, function(value) {
  82. var result = value.match(URL_VALUE_EXP);
  83. if (result)
  84. if (result[1].indexOf("data:") != 0)
  85. return value.replace(result[1], formatURL(result[1], host));
  86. return value;
  87. });
  88. return ret.replace(IMPORT_ALT_EXP, function(value) {
  89. var result = value.match(IMPORT_VALUE_ALT_EXP);
  90. if (result)
  91. if (result[1].indexOf("data:") != 0)
  92. return "@import \"" + formatURL(result[1], host) + "\";";
  93. return value;
  94. });
  95. }
  96. function getDataURI(data, defaultURL, woURL) {
  97. if (data.content)
  98. return (woURL ? "" : "url(") + "data:" + data.mediaType + ";" + data.mediaTypeParam + "," + data.content + (woURL ? "" : ")");
  99. else
  100. return woURL ? defaultURL : "url(" + defaultURL + ")";
  101. }
  102. function removeComments(content) {
  103. var start, end;
  104. do {
  105. start = content.indexOf("/*");
  106. end = content.indexOf("*/", start);
  107. if (start != -1 && end != -1)
  108. content = content.substring(0, start) + content.substr(end + 2);
  109. } while (start != -1 && end != -1);
  110. return content;
  111. }
  112. function replaceURLs(content, host, requestManager, callback) {
  113. var i, url, result, values = removeComments(content).match(URL_EXP), requestMax = 0, requestIndex = 0;
  114. function sendRequest(origUrl) {
  115. requestMax++;
  116. requestManager.send(url, function(data) {
  117. requestIndex++;
  118. if (content.indexOf(origUrl) != -1) {
  119. data.mediaType = data.mediaType ? data.mediaType.split(";")[0] : null;
  120. content = content.replace(new RegExp(origUrl.replace(/([{}\(\)\^$&.\*\?\/\+\|\[\\\\]|\]|\-)/g, "\\$1"), "gi"), getDataURI(data,
  121. EMPTY_PIXEL_DATA, true));
  122. }
  123. if (requestIndex == requestMax)
  124. callback(content);
  125. }, null, "base64");
  126. }
  127. if (values)
  128. for (i = 0; i < values.length; i++) {
  129. result = values[i].match(URL_VALUE_EXP);
  130. if (result && result[1]) {
  131. url = formatURL(result[1], host);
  132. if (url.indexOf("data:") != 0)
  133. sendRequest(result[1]);
  134. }
  135. }
  136. }
  137. // ----------------------------------------------------------------------------------------------
  138. function processStylesheets(doc, docElement, baseURI, requestManager) {
  139. Array.prototype.forEach.call(docElement.querySelectorAll('link[href][rel*="stylesheet"]'), function(node) {
  140. var href = node.getAttribute("href"), url = formatURL(href, baseURI);
  141. function createStyleNode(content) {
  142. var i, newNode, commentNode;
  143. newNode = doc.createElement("style");
  144. for (i = 0; i < node.attributes.length; i++)
  145. if (node.attributes[i].value)
  146. newNode.setAttribute(node.attributes[i].name, node.attributes[i].value);
  147. newNode.dataset.href = url;
  148. newNode.removeAttribute("href");
  149. newNode.textContent = resolveURLs(content, url);
  150. if (node.disabled) {
  151. commentNode = doc.createComment();
  152. commentNode.textContent = newNode.outerHTML.replace(/</g, "&lt;").replace(/>/g, "&gt;").replace(/--/g, "&minus;&minus;");
  153. node.parentElement.replaceChild(commentNode, node);
  154. } else
  155. node.parentElement.replaceChild(newNode, node);
  156. }
  157. if (href.indexOf("data:") != 0)
  158. requestManager.send(url, function(data) {
  159. if (data.status >= 400)
  160. node.parentElement.removeChild(node);
  161. else
  162. createStyleNode(data.content || "");
  163. });
  164. else
  165. createStyleNode(decodeDataURI(href));
  166. });
  167. }
  168. function processImports(docElement, baseURI, characterSet, requestManager) {
  169. var ret = true;
  170. Array.prototype.forEach.call(docElement.querySelectorAll("style"), function(styleSheet) {
  171. var imports = removeComments(styleSheet.textContent).match(IMPORT_EXP);
  172. if (imports)
  173. imports.forEach(function(imp) {
  174. var url, href, result = imp.match(IMPORT_URL_VALUE_EXP);
  175. function insertStylesheet(content) {
  176. styleSheet.textContent = styleSheet.textContent.replace(imp, resolveURLs(content, url));
  177. }
  178. if (result && (result[2] || result[4])) {
  179. href = result[2] || result[4];
  180. url = formatURL(href, styleSheet.dataset.href || baseURI);
  181. if (href.indexOf("data:") != 0) {
  182. requestManager.send(url, function(data) {
  183. insertStylesheet(data.status < 400 && data.content ? data.content : "");
  184. }, null, characterSet);
  185. } else
  186. insertStylesheet(decodeDataURI(href));
  187. ret = false;
  188. }
  189. });
  190. });
  191. return ret;
  192. }
  193. function processStyleAttributes(docElement, baseURI, requestManager) {
  194. Array.prototype.forEach.call(docElement.querySelectorAll("*[style]"), function(node) {
  195. replaceURLs(node.getAttribute("style"), baseURI, requestManager, function(style) {
  196. node.setAttribute("style", style);
  197. });
  198. });
  199. }
  200. function processBgAttributes(docElement, baseURI, requestManager) {
  201. var backgrounds = docElement.querySelectorAll("*[background]");
  202. Array.prototype.forEach.call(backgrounds, function(node) {
  203. var url, value = node.getAttribute("background");
  204. if (value.indexOf(".") != -1) {
  205. url = formatURL(value, baseURI);
  206. if (url.indexOf("data:") != 0)
  207. requestManager.send(url, function(data) {
  208. node.setAttribute("background", getDataURI(data, EMPTY_PIXEL_DATA, true));
  209. }, null, "base64");
  210. }
  211. });
  212. }
  213. function insertDefaultFavico(doc, docElement, baseURI) {
  214. var node, docHead = docElement.querySelector("html > head"), favIcon = docElement
  215. .querySelector('link[href][rel="shortcut icon"], link[href][rel="apple-touch-icon"], link[href][rel="icon"]');
  216. if (!favIcon && docHead) {
  217. node = doc.createElement("link");
  218. node.setAttribute("type", "image/x-icon");
  219. node.setAttribute("rel", "shortcut icon");
  220. node.setAttribute("href", formatURL("/favicon.ico", baseURI));
  221. docHead.appendChild(node);
  222. }
  223. }
  224. function processImages(docElement, baseURI, requestManager) {
  225. var images;
  226. function process(attributeName) {
  227. Array.prototype.forEach.call(images, function(node) {
  228. var url = formatURL(node.getAttribute(attributeName), baseURI);
  229. if (url.indexOf("data:") != 0)
  230. requestManager.send(url, function(data) {
  231. node.setAttribute(attributeName, getDataURI(data, EMPTY_PIXEL_DATA, true));
  232. }, null, "base64");
  233. });
  234. }
  235. images = docElement.querySelectorAll('link[href][rel="shortcut icon"], link[href][rel="apple-touch-icon"], link[href][rel="icon"]');
  236. process("href");
  237. images = docElement.querySelectorAll('img[src], input[src][type="image"]');
  238. process("src");
  239. images = docElement.querySelectorAll('video[poster]');
  240. process("poster");
  241. }
  242. function processSVGs(docElement, baseURI, requestManager) {
  243. var images = docElement.querySelectorAll('object[type="image/svg+xml"], object[type="image/svg-xml"], embed[src*=".svg"]');
  244. Array.prototype.forEach.call(images, function(node) {
  245. var data = node.getAttribute("data"), src = node.getAttribute("src"), url = formatURL(data || src, baseURI);
  246. if (url.indexOf("data:") != 0)
  247. requestManager.send(url, function(data) {
  248. node.setAttribute(data ? "data" : "src", getDataURI(data, "data:text/xml,<svg></svg>", true));
  249. }, null, null);
  250. });
  251. }
  252. function processStyles(docElement, baseURI, requestManager) {
  253. Array.prototype.forEach.call(docElement.querySelectorAll("style"), function(styleSheet) {
  254. replaceURLs(styleSheet.textContent, styleSheet.dataset.href || baseURI, requestManager, function(textContent) {
  255. styleSheet.textContent = textContent;
  256. });
  257. });
  258. }
  259. function processScripts(docElement, baseURI, characterSet, requestManager) {
  260. Array.prototype.forEach.call(docElement.querySelectorAll("script[src]"), function(node) {
  261. var src = node.getAttribute("src");
  262. if (src.indexOf("data:") != 0)
  263. requestManager.send(formatURL(src, baseURI), function(data) {
  264. if (data.status < 400) {
  265. data.content = data.content.replace(/"([^"]*)<\/\s*script\s*>([^"]*)"/gi, '"$1<"+"/script>$2"');
  266. data.content = data.content.replace(/'([^']*)<\/\s*script\s*>([^']*)'/gi, "'$1<'+'/script>$2'");
  267. node.textContent = "\n" + data.content + "\n";
  268. }
  269. node.removeAttribute("src");
  270. }, characterSet);
  271. });
  272. }
  273. function processCanvas(doc, docElement, canvasData) {
  274. var index = 0;
  275. Array.prototype.forEach.call(docElement.querySelectorAll("canvas"), function(node) {
  276. var i, data = canvasData[index], newNode = doc.createElement("img");
  277. if (data) {
  278. newNode.setAttribute("src", data);
  279. for (i = 0; i < node.attributes.length; i++)
  280. if (node.attributes[i].value)
  281. newNode.setAttribute(node.attributes[i].name, node.attributes[i].value);
  282. if (!newNode.width)
  283. newNode.style.pixelWidth = node.clientWidth;
  284. if (!newNode.height)
  285. newNode.style.pixelHeight = node.clientHeight;
  286. node.parentElement.replaceChild(newNode, node);
  287. }
  288. index++;
  289. });
  290. }
  291. function removeScripts(docElement) {
  292. Array.prototype.forEach.call(docElement.querySelectorAll("script"), function(node) {
  293. node.parentElement.removeChild(node);
  294. });
  295. Array.prototype.forEach.call(docElement.querySelectorAll("*[onload]"), function(node) {
  296. node.removeAttribute("onload");
  297. });
  298. }
  299. function removeObjects(docElement) {
  300. var objects = docElement.querySelectorAll('applet, object:not([type="image/svg+xml"]):not([type="image/svg-xml"]), embed:not([src*=".svg"])');
  301. Array.prototype.forEach.call(objects, function(node) {
  302. node.parentElement.removeChild(node);
  303. });
  304. objects = docElement.querySelectorAll('audio[src], video[src]');
  305. Array.prototype.forEach.call(objects, function(node) {
  306. node.removeAttribute("src");
  307. });
  308. }
  309. function removeBlockquotesCite(docElement) {
  310. Array.prototype.forEach.call(docElement.querySelectorAll("blockquote[cite]"), function(node) {
  311. node.removeAttribute("cite");
  312. });
  313. }
  314. function removeFrames(docElement) {
  315. Array.prototype.forEach.call(docElement.querySelectorAll("iframe, frame"), function(node) {
  316. node.parentElement.removeChild(node);
  317. });
  318. }
  319. function removeMetaRefresh(docElement) {
  320. Array.prototype.forEach.call(docElement.querySelectorAll("meta[http-equiv=refresh]"), function(node) {
  321. node.parentElement.removeChild(node);
  322. });
  323. }
  324. function resetFrames(docElement, baseURI) {
  325. Array.prototype.forEach.call(docElement.querySelectorAll("iframe, frame"), function(node) {
  326. var src = formatURL(node.getAttribute("src"), baseURI);
  327. if (src.indexOf("data:") != 0)
  328. node.setAttribute("src", "about:blank");
  329. });
  330. }
  331. function setAbsoluteLinks(docElement, baseURI) {
  332. Array.prototype.forEach.call(docElement.querySelectorAll("a:not([href^='#'])"), function(link) {
  333. var fullHref = formatURL(link.getAttribute("href"), baseURI);
  334. if (fullHref && (!(fullHref.indexOf(baseURI.split("#")[0]) == 0) || fullHref.indexOf("#") == -1))
  335. link.setAttribute("href", fullHref);
  336. });
  337. }
  338. // ----------------------------------------------------------------------------------------------
  339. singlefile.initProcess = function(doc, docElement, addDefaultFavico, baseURI, characterSet, config, canvasData, requestManager, onInit, onProgress, onEnd) {
  340. var initManager = new RequestManager(), manager = new RequestManager(onProgress);
  341. function RequestManager(onProgress) {
  342. var that = this, currentCount = 0, requests = [];
  343. this.requestCount = 0;
  344. this.send = function(url, responseHandler, characterSet, mediaTypeParam) {
  345. this.requestCount++;
  346. requests.push({
  347. url : url,
  348. responseHandler : responseHandler,
  349. characterSet : characterSet,
  350. mediaTypeParam : mediaTypeParam
  351. });
  352. };
  353. this.doSend = function() {
  354. requests.forEach(function(request) {
  355. requestManager.send(request.url, function(response) {
  356. request.responseHandler(response);
  357. currentCount++;
  358. if (onProgress)
  359. onProgress(currentCount, that.requestCount);
  360. if (currentCount == that.requestCount) {
  361. that.requestCount = 0;
  362. currentCount = 0;
  363. if (that.onEnd)
  364. that.onEnd();
  365. }
  366. }, request.characterSet, request.mediaTypeParam);
  367. });
  368. requests = [];
  369. };
  370. }
  371. function cbImports() {
  372. if (config.removeScripts)
  373. removeScripts(docElement);
  374. if (config.removeObjects)
  375. removeObjects(docElement);
  376. if (config.removeFrames || config.getRawDoc)
  377. removeFrames(docElement);
  378. resetFrames(docElement, baseURI);
  379. removeBlockquotesCite(docElement);
  380. removeMetaRefresh(docElement);
  381. setAbsoluteLinks(docElement, baseURI);
  382. if (addDefaultFavico)
  383. insertDefaultFavico(doc, docElement, baseURI);
  384. processStyleAttributes(docElement, baseURI, manager);
  385. processBgAttributes(docElement, baseURI, manager);
  386. processImages(docElement, baseURI, manager);
  387. processSVGs(docElement, baseURI, manager);
  388. processStyles(docElement, baseURI, manager);
  389. processScripts(docElement, baseURI, characterSet, manager);
  390. processCanvas(doc, docElement, canvasData);
  391. if (onInit)
  392. setTimeout(function() {
  393. onInit(manager.requestCount);
  394. }, 1);
  395. }
  396. function cbStylesheets() {
  397. initManager.onEnd = function(noRequests) {
  398. if (noRequests)
  399. cbImports();
  400. else
  401. cbStylesheets();
  402. };
  403. processImports(docElement, baseURI, characterSet, initManager);
  404. initManager.doSend();
  405. if (initManager.requestCount == 0)
  406. cbImports();
  407. }
  408. manager.onEnd = onEnd;
  409. processStylesheets(doc, docElement, baseURI, initManager);
  410. initManager.onEnd = cbStylesheets;
  411. initManager.doSend();
  412. if (initManager.requestCount == 0)
  413. initManager.onEnd();
  414. return function() {
  415. manager.doSend();
  416. if (manager.onEnd && manager.requestCount == 0)
  417. manager.onEnd();
  418. };
  419. };
  420. })();