No bug - update readability from github repo, includes fix for Bug 1173548 and Bug 1255978, r=Gijs
authorEvan Tseng <evan@tseng.io>
Mon, 28 Nov 2016 17:10:20 +0800
changeset 324621 e9ff40069326e4c5d5e72bc58e65bef0778ac67a
parent 324620 e355836470f90868b9bba2d15e2acc10f8b6ff4b
child 324622 357234ded571f22c456f3076b90f2c2cf5ca6950
push id24
push usermaklebus@msu.edu
push dateTue, 20 Dec 2016 03:11:33 +0000
reviewersGijs
bugs1173548, 1255978
milestone53.0a1
No bug - update readability from github repo, includes fix for Bug 1173548 and Bug 1255978, r=Gijs MozReview-Commit-ID: 1Hr9uUqZn5P
toolkit/components/reader/Readability.js
--- a/toolkit/components/reader/Readability.js
+++ b/toolkit/components/reader/Readability.js
@@ -114,17 +114,17 @@ Readability.prototype = {
   DEFAULT_MAX_PAGES: 5,
 
   // Element tags to score by default.
   DEFAULT_TAGS_TO_SCORE: "section,h2,h3,h4,h5,h6,p,td,pre".toUpperCase().split(","),
 
   // All of the regular expressions in use within readability.
   // Defined up here so we don't instantiate them repeatedly in loops.
   REGEXPS: {
-    unlikelyCandidates: /banner|combx|comment|community|disqus|extra|foot|header|menu|modal|related|remark|rss|share|shoutbox|sidebar|skyscraper|sponsor|ad-break|agegate|pagination|pager|popup/i,
+    unlikelyCandidates: /banner|combx|comment|community|disqus|extra|foot|header|legends|menu|modal|related|remark|rss|shoutbox|sidebar|skyscraper|sponsor|ad-break|agegate|pagination|pager|popup/i,
     okMaybeItsACandidate: /and|article|body|column|main|shadow/i,
     positive: /article|body|content|entry|hentry|h-entry|main|page|pagination|post|text|blog|story/i,
     negative: /hidden|^hid$| hid$| hid |^hid |banner|combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|modal|outbrain|promo|related|scroll|share|shoutbox|sidebar|skyscraper|sponsor|shopping|tags|tool|widget/i,
     extraneous: /print|archive|comment|discuss|e[\-]?mail|share|reply|all|login|sign|single|utility/i,
     byline: /byline|author|dateline|writtenby|p-author/i,
     replaceFonts: /<(\/?)font[^>]*>/gi,
     normalize: /\s{2,}/g,
     videos: /\/\/(www\.)?(dailymotion|youtube|youtube-nocookie|player\.vimeo)\.com/i,
@@ -466,16 +466,22 @@ Readability.prototype = {
 
     // Clean out junk from the article content
     this._cleanConditionally(articleContent, "form");
     this._clean(articleContent, "object");
     this._clean(articleContent, "embed");
     this._clean(articleContent, "h1");
     this._clean(articleContent, "footer");
 
+    // Clean out elements have "share" in their id/class combinations from final top candidates,
+    // which means we don't remove the top candidates even they have "share".
+    this._forEachNode(articleContent.children, function(topCandidate) {
+      this._cleanMatchedNodes(topCandidate, /share/);
+    });
+
     // If there is only one h2, they are probably using it as a header
     // and not a subheader, so remove it since we already have a header.
     if (articleContent.getElementsByTagName('h2').length === 1)
       this._clean(articleContent, "h2");
 
     this._clean(articleContent, "iframe");
     this._cleanHeaders(articleContent);
 
@@ -657,19 +663,16 @@ Readability.prototype = {
     // We can't grab an article if we don't have a page!
     if (!page) {
       this.log("No body found in document. Abort.");
       return null;
     }
 
     var pageCacheHtml = page.innerHTML;
 
-    // Check if any "dir" is set on the toplevel document element
-    this._articleDir = doc.documentElement.getAttribute("dir");
-
     while (true) {
       var stripUnlikelyCandidates = this._flagIsActive(this.FLAG_STRIP_UNLIKELYS);
 
       // First, node prepping. Trash nodes that look cruddy (like ones with the
       // class name "comment", etc), and turn divs into P tags where they have been
       // used inappropriately (as in, where they contain no other block level elements.)
       var elementsToScore = [];
       var node = this._doc.documentElement;
@@ -807,16 +810,17 @@ Readability.prototype = {
               topCandidates.pop();
             break;
           }
         }
       }
 
       var topCandidate = topCandidates[0] || null;
       var neededToCreateTopCandidate = false;
+      var parentOfTopCandidate;
 
       // If we still have no top candidate, just use the body as a last resort.
       // We also have to copy the body node so it is something we can modify.
       if (topCandidate === null || topCandidate.tagName === "BODY") {
         // Move all of the page's children into topCandidate
         topCandidate = doc.createElement("DIV");
         neededToCreateTopCandidate = true;
         // Move everything (not just elements, also text nodes etc.) into the container
@@ -833,17 +837,17 @@ Readability.prototype = {
       } else if (topCandidate) {
         // Because of our bonus system, parents of candidates might have scores
         // themselves. They get half of the node. There won't be nodes with higher
         // scores than our topCandidate, but if we see the score going *up* in the first
         // few steps up the tree, that's a decent sign that there might be more content
         // lurking in other places that we want to unify in. The sibling stuff
         // below does some of that - but only if we've looked high enough up the DOM
         // tree.
-        var parentOfTopCandidate = topCandidate.parentNode;
+        parentOfTopCandidate = topCandidate.parentNode;
         var lastScore = topCandidate.readability.contentScore;
         // The scores shouldn't get too low.
         var scoreThreshold = lastScore / 3;
         while (parentOfTopCandidate && parentOfTopCandidate.readability) {
           var parentScore = parentOfTopCandidate.readability.contentScore;
           if (parentScore < scoreThreshold)
             break;
           if (parentScore > lastScore) {
@@ -859,17 +863,19 @@ Readability.prototype = {
       // Now that we have the top candidate, look through its siblings for content
       // that might also be related. Things like preambles, content split by ads
       // that we removed, etc.
       var articleContent = doc.createElement("DIV");
       if (isPaging)
         articleContent.id = "readability-content";
 
       var siblingScoreThreshold = Math.max(10, topCandidate.readability.contentScore * 0.2);
-      var siblings = topCandidate.parentNode.children;
+      // Keep potential top candidate's parent node to try to get text direction of it later.
+      parentOfTopCandidate = topCandidate.parentNode;
+      var siblings = parentOfTopCandidate.children;
 
       for (var s = 0, sl = siblings.length; s < sl; s++) {
         var sibling = siblings[s];
         var append = false;
 
         this.log("Looking at sibling node:", sibling, sibling.readability ? ("with score " + sibling.readability.contentScore) : '');
         this.log("Sibling has score", sibling.readability ? sibling.readability.contentScore : 'Unknown');
 
@@ -963,16 +969,28 @@ Readability.prototype = {
         } else if (this._flagIsActive(this.FLAG_WEIGHT_CLASSES)) {
           this._removeFlag(this.FLAG_WEIGHT_CLASSES);
         } else if (this._flagIsActive(this.FLAG_CLEAN_CONDITIONALLY)) {
           this._removeFlag(this.FLAG_CLEAN_CONDITIONALLY);
         } else {
           return null;
         }
       } else {
+        // Find out text direction from ancestors of final top candidate.
+        var ancestors = [parentOfTopCandidate, topCandidate].concat(this._getNodeAncestors(parentOfTopCandidate));
+        this._someNode(ancestors, function(ancestor) {
+          if (!ancestor.tagName)
+            return false;
+          var articleDir = ancestor.getAttribute("dir");
+          if (articleDir) {
+            this._articleDir = articleDir;
+            return true;
+          }
+          return false;
+        });
         return articleContent;
       }
     }
   },
 
   /**
    * Check whether the input string could be a byline.
    * This verifies that the input is a string, and that the length
@@ -1691,16 +1709,35 @@ Readability.prototype = {
           ((embedCount === 1 && contentLength < 75) || embedCount > 1);
         return haveToRemove;
       }
       return false;
     });
   },
 
   /**
+   * Clean out elements whose id/class combinations match specific string.
+   *
+   * @param Element
+   * @param RegExp match id/class combination.
+   * @return void
+   **/
+  _cleanMatchedNodes: function(e, regex) {
+    var endOfSearchMarkerNode = this._getNextNode(e, true);
+    var next = this._getNextNode(e);
+    while (next && next != endOfSearchMarkerNode) {
+      if (regex.test(next.className + " " + next.id)) {
+        next = this._removeAndGetNext(next);
+      } else {
+        next = this._getNextNode(next);
+      }
+    }
+  },
+
+  /**
    * Clean out spurious headers from an Element. Checks things like classnames and link density.
    *
    * @param Element
    * @return void
   **/
   _cleanHeaders: function(e) {
     for (var headerIndex = 1; headerIndex < 3; headerIndex += 1) {
       this._removeNodes(e.getElementsByTagName('h' + headerIndex), function (header) {