Fix for keeping lists structure

java/org/chromium/distiller/webdocument/filters/NestedElementRetainer.java

Index: java/org/chromium/distiller/webdocument/filters/NestedElementRetainer.java
diff --git a/java/org/chromium/distiller/webdocument/filters/NestedElementRetainer.java b/java/org/chromium/distiller/webdocument/filters/NestedElementRetainer.java
new file mode 100644
index 0000000000000000000000000000000000000000..293c766577c87df6ebcf5dc58ee710ef59193f7a
--- /dev/null
+++ b/java/org/chromium/distiller/webdocument/filters/NestedElementRetainer.java
@@ -0,0 +1,40 @@
+package org.chromium.distiller.webdocument.filters;
+
+import org.chromium.distiller.webdocument.WebDocument;
+import org.chromium.distiller.webdocument.WebElement;
+import org.chromium.distiller.webdocument.WebTag;
+
+import java.util.Stack;
+
+public class NestedElementRetainer {

[mdjones, 2015/08/05 15:44:35]

Add some documentation about what this class is and how it works.

+    public static void process(WebDocument document) {
+        boolean isContent = false;
+        int stackMark = -1;
+        Stack<WebTag> stack = new Stack<>();
+
+        for (WebElement e : document.getElements()) {
+            if (!(e instanceof WebTag)) {
+                if (!isContent) {
+                    isContent = e.getIsContent();
+                }
+            } else {
+                WebTag webTag = (WebTag) e;
+                if (webTag.isStartTag()) {
+                    webTag.setIsContent(isContent);
+                    stack.push(webTag);
+                    isContent = false;
+                } else {
+                    WebTag startWebTag = stack.pop();
+                    isContent |= stackMark >= stack.size();
+                    if (isContent) {
+                        stackMark = stack.size() - 1;
+                    }
+                    boolean wasContent = startWebTag.getIsContent();
+                    startWebTag.setIsContent(isContent);
+                    webTag.setIsContent(isContent);
+                    isContent = wasContent;
+                }
+            }
+        }
+    }
+}