Bug 1107588 - Use an xhr to download reader mode content instead of creating new browser elements. r=mfinkle

--HG--
extra : rebase_source : fe24bfefa2efc5500ca74a9be5f79a958bab32ee

mobile/android/base/tests/testReadingListCache.js

@@ -69,7 +69,7 @@ add_task(function* test_remove_article() {
 
 add_task(function* test_parse_articles() {
   for (let testcase of TEST_PAGES) {
-    let article = yield Reader._downloadAndParseDocument(testcase.url);
+    let article = yield ReaderMode.downloadAndParseDocument(testcase.url);
     checkArticle(article, testcase);
   }
 });

mobile/android/chrome/content/Reader.js

@@ -170,62 +170,7 @@ let Reader = {
 
     // Article hasn't been found in the cache, we need to
     // download the page and parse the article out of it.
-    return yield this._downloadAndParseDocument(url);
-  }),
-
-  _downloadDocument: function (url) {
-    return new Promise((resolve, reject) => {
-      // We want to parse those arbitrary pages safely, outside the privileged
-      // context of chrome. We create a hidden browser element to fetch the
-      // loaded page's document object then discard the browser element.
-      let browser = document.createElement("browser");
-      browser.setAttribute("type", "content");
-      browser.setAttribute("collapsed", "true");
-      browser.setAttribute("disablehistory", "true");
-
-      document.documentElement.appendChild(browser);
-      browser.stop();
-
-      browser.webNavigation.allowAuth = false;
-      browser.webNavigation.allowImages = false;
-      browser.webNavigation.allowJavascript = false;
-      browser.webNavigation.allowMetaRedirects = true;
-      browser.webNavigation.allowPlugins = false;
-
-      browser.addEventListener("DOMContentLoaded", event => {
-        let doc = event.originalTarget;
-
-        // ignore on frames and other documents
-        if (doc != browser.contentDocument) {
-          return;
-        }
-
-        if (doc.location.href == "about:blank") {
-          reject("about:blank loaded; aborting");
-
-          // Request has finished with error, remove browser element
-          browser.parentNode.removeChild(browser);
-          return;
-        }
-
-        resolve({ browser, doc });
-      });
-
-      browser.loadURIWithFlags(url, Ci.nsIWebNavigation.LOAD_FLAGS_NONE,
-                               null, null, null);
-    });
-  },
-
-  _downloadAndParseDocument: Task.async(function* (url) {
-    let { browser, doc } = yield this._downloadDocument(url);
-
-    try {
-      let uri = Services.io.newURI(url, null, null);
-      let article = yield ReaderMode.readerParse(uri, doc);
-      return article;
-    } finally {
-      browser.parentNode.removeChild(browser);
-    }
+    return yield ReaderMode.downloadAndParseDocument(url);
   }),
 
   /**

toolkit/components/reader/ReaderMode.jsm

@@ -61,9 +61,56 @@ let ReaderMode = {
     }
 
     let doc = browser.contentWindow.document;
-    return yield this.readerParse(uri, doc);
+    return yield this._readerParse(uri, doc);
   }),
 
+  /**
+   * Downloads and parses a document from a URL.
+   *
+   * @param url URL to download and parse.
+   * @return {Promise}
+   * @resolves JS object representing the article, or null if no article is found.
+   */
+  downloadAndParseDocument: Task.async(function* (url) {
+    let uri = Services.io.newURI(url, null, null);
+    let doc = yield this._downloadDocument(url);
+    return yield this._readerParse(uri, doc);
+  }),
+
+  _downloadDocument: function (url) {
+    return new Promise((resolve, reject) => {
+      let xhr = new XMLHttpRequest();
+      xhr.open("GET", url, true);
+      xhr.onerror = evt => reject(evt.error);
+      xhr.responseType = "document";
+      xhr.onload = evt => {
+        if (xhr.status !== 200) {
+          reject("Reader mode XHR failed with status: " + xhr.status);
+          return;
+        }
+
+        let doc = xhr.responseXML;
+
+        // Manually follow a meta refresh tag if one exists.
+        let meta = doc.querySelector("meta[http-equiv=refresh]");
+        if (meta) {
+          let content = meta.getAttribute("content");
+          if (content) {
+            let urlIndex = content.indexOf("URL=");
+            if (urlIndex > -1) {
+              let url = content.substring(urlIndex + 4);
+              this._downloadDocument(url).then((doc) => resolve(doc));
+              return;
+            }
+          }
+        }
+        resolve(doc);
+      }
+      xhr.send();
+    });
+  },
+
+
   /**
    * Retrieves an article from the cache given an article URI.
    *
@@ -138,7 +185,7 @@ let ReaderMode = {
    * @return {Promise}
    * @resolves JS object representing the article, or null if no article is found.
    */
-  readerParse: function (uri, doc) {
+  _readerParse: function (uri, doc) {
     return new Promise((resolve, reject) => {
       let numTags = doc.getElementsByTagName("*").length;
       if (numTags > this.MAX_ELEMS_TO_PARSE) {