Extract image URLs in srcset as well

java/org/chromium/distiller/webdocument/WebImage.java

Index: java/org/chromium/distiller/webdocument/WebImage.java
diff --git a/java/org/chromium/distiller/webdocument/WebImage.java b/java/org/chromium/distiller/webdocument/WebImage.java
index ceb3a405279c3dbf95c723c6eb9ced12f7a095b6..df50a3587252627a5c89525eeaed13bed835f89c 100644
--- a/java/org/chromium/distiller/webdocument/WebImage.java
+++ b/java/org/chromium/distiller/webdocument/WebImage.java
@@ -10,6 +10,9 @@ import com.google.gwt.dom.client.ImageElement;
 
 import org.chromium.distiller.DomUtil;
 
+import java.util.ArrayList;
+import java.util.List;
+
 /**
  * WebImage represents an image in the WebDocument potentially needing extraction.
  */
@@ -22,6 +25,8 @@ public class WebImage extends WebElement {
     private int width;
     // The original height of the image in pixels.
     private int height;
+    // The list of image URLs.
+    private List<String> list;
 
     /**
      * Build an image element.
@@ -56,6 +61,10 @@ public class WebImage extends WebElement {
         DomUtil.makeSrcSetAbsolute(ie);
         DomUtil.stripImageElement(ie);
 
+        list = new ArrayList<>();
+        list.add(srcUrl);
+        list.addAll(DomUtil.getSrcSetUrls(ie));
+
         Element container = Document.get().createDivElement();
         container.appendChild(ie);
         return container.getInnerHTML();
@@ -86,10 +95,14 @@ public class WebImage extends WebElement {
     }
 
     /**
-     * Get the source URL of this image.
-     * @return Source URL or an empty string.
+     * Get the list of source URLs of this image.
+     * It's more efficient to call after generateOutput().
+     * @return Source URLs or an empty List.
      */
-    public String getSrc() {
-        return srcUrl;
+    public List<String> getUrlList() {
+        if (list == null) {
+            generateOutput(false);

[wychen, 2016/08/01 20:59:20]

This feels entangled. :(

[mdjones, 2016/08/01 21:14:35]

Can we instead cut the top part of generateOutput and put it into a function
called something like cloneAndProcessNode? We could then call that function
lazily in multiple places. Leaving this as-is will cause a lot of confusion
later on.

[wychen, 2016/08/01 21:35:15]

Done.

+        }
+        return list;
     }
 }