Discard hidden articles when using fast path

java/org/chromium/distiller/ContentExtractor.java

 // Copyright 2014 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 package org.chromium.distiller;
 
 import org.chromium.distiller.document.TextDocument;
 import org.chromium.distiller.document.TextDocumentStatistics;
 import org.chromium.distiller.extractors.ArticleExtractor;
 import org.chromium.distiller.proto.DomDistillerProtos.StatisticsInfo;
@@ skipping 149 matching lines @@
     public List<String> getImageUrls() {
         return imageUrls;
     }
 
     /**
      * Get the element of the main article, if any.
      * @return An element of article (not necessarily the html5 article element).
      */
     private Element getArticleElement(Element root) {
         NodeList<Element> allArticles = root.getElementsByTagName("ARTICLE");
+        List<Element> visibleElements = getVisibleElements(allArticles);
         // Having multiple article elements usually indicates a bad case for this shortcut.
         // TODO(wychen): some sites exclude things like title and author in article element.
-        if (allArticles.getLength() == 1) {
-            return allArticles.getItem(0);
+        if (visibleElements.size() == 1) {
+            return visibleElements.get(0);
         }
         // Note that the CSS property matching is case sensitive, and "Article" is the correct
         // capitalization.
         String query = "[itemscope][itemtype*=\"Article\"],[itemscope][itemtype*=\"Post\"]";
         allArticles = DomUtil.querySelectorAll(root, query);
+        visibleElements = getVisibleElements(allArticles);
         // It is commonly seen that the article is wrapped separately or in multiple layers.
-        if (allArticles.getLength() > 0) {
-            return Element.as(DomUtil.getNearestCommonAncestor(allArticles));
+        if (visibleElements.size() > 0) {
+            return Element.as(DomUtil.getNearestCommonAncestor(visibleElements));
         }
         return null;
     }
 
     /**
+     * Get a list of visible elements.
+     * @return A list of visible elements.
+     */
+    private List<Element> getVisibleElements(NodeList<Element> nodeList) {
+        List<Element> visibleElements = new ArrayList<>();
+        for (int i = 0; i < nodeList.getLength(); i ++) {
+            Element element = nodeList.getItem(i);
+            if (DomUtil.isVisible(element)) {
+                visibleElements.add(element);
+            }
+        }
+        return visibleElements;
+    }
+
+    /**
      * Converts the original HTML page into a WebDocument for analysis.
      */
     private WebDocumentInfo createWebDocumentInfoFromPage() {
         WebDocumentInfo info = new WebDocumentInfo();
         WebDocumentBuilder documentBuilder = new WebDocumentBuilder();
         DomConverter converter = new DomConverter(documentBuilder);
         Element walkerRoot = getArticleElement(documentElement);
         if (walkerRoot == null) {
             walkerRoot = documentElement;
         }
@@ skipping 11 matching lines @@
      *
      * @param document the WebDocument representation of the page extracted from the DOM.
      */
     private void processDocument(WebDocument document) {
         TextDocument textDocument = document.createTextDocumentView();
         ArticleExtractor.INSTANCE.process(textDocument, candidateTitles);
         mStatisticsInfo.setWordCount(TextDocumentStatistics.countWordsInContent(textDocument));
         textDocument.applyToModel();
     }
 }