Fix O(n!) tag name processing. Fixes #27. Also simplify constructor.

dkpro · Apr 3, 2016 · a9d101d · a9d101d
1 parent f8450d3
commit a9d101d
Show file tree

Hide file tree

Showing 2 changed files with 15 additions and 46 deletions.
diff --git a/...ilerplate/src/main/java/de/tudarmstadt/ukp/dkpro/c4corpus/boilerplate/impl/Paragraph.java b/...ilerplate/src/main/java/de/tudarmstadt/ukp/dkpro/c4corpus/boilerplate/impl/Paragraph.java
@@ -32,6 +32,7 @@
 public class Paragraph
         extends LinkedList<Node>
 {
+    private static final long serialVersionUID = 1L;
 
     //    private ArrayList<String> textNodes;
     int charsCountInLinks = 0;
@@ -40,32 +41,21 @@ public class Paragraph
     private String tagName = "";
     private String rawText = "";
 
-    public Paragraph(Node firstNode)
+    public Paragraph(Node firstNode, String path)
     {
         add(firstNode);
-        initRawInfo();
-    }
-
-    private void initRawInfo()
-    {
-        StringBuilder sb = new StringBuilder();
-        // This is only called at initialization time, so will never have more than a single node
-        for (Node n : this) {
-            //            NodeHelper.cleanEmptyElements(n);
-            if (n instanceof TextNode) {
-                this.setTagName(getPath(n));
-                String nodeRawText = ((TextNode) n).text();
-                sb.append(Utils.normalizeBreaks(nodeRawText).trim());
+        this.tagName = path;
+        if (firstNode instanceof TextNode) {
+            String nodeRawText = ((TextNode) firstNode).text();
+            this.rawText = Utils.normalizeBreaks(nodeRawText).trim();
 
-                if (NodeHelper.isLink(n)) {
-                    charsCountInLinks += nodeRawText.length();
-                }
+            if (NodeHelper.isLink(firstNode)) {
+                charsCountInLinks += nodeRawText.length();
             }
         }
-
-        rawText = sb.toString();
     }
 
+
     public int getLinksLength()
     {
         return this.charsCountInLinks;
@@ -96,30 +86,6 @@ public String getTagName()
         return this.tagName;
     }
 
-    public String getPath(Node n)
-    {
-        String nodePath = "";
-        // FIXME: This is pathologically recomputing things that we know because we're doing a depth-first traverse
-        while (n != null) {
-            if (n instanceof TextNode) {
-                n = n.parent();
-            }
-            if (NodeHelper.isInnerText(n)) {
-                n = n.parent();
-            }
-            String parentNodeName = n.nodeName();
-            nodePath = parentNodeName + "." + nodePath;
-
-            if (!parentNodeName.equalsIgnoreCase("html")) {
-                n = n.parent();
-            }
-            else {
-                break;
-            }
-        }
-
-        return nodePath;
-    }
 
     public void setTagName(String name)
     {

diff --git a/.../src/main/java/de/tudarmstadt/ukp/dkpro/c4corpus/boilerplate/impl/ParagraphsExplorer.java b/.../src/main/java/de/tudarmstadt/ukp/dkpro/c4corpus/boilerplate/impl/ParagraphsExplorer.java
@@ -41,6 +41,7 @@ public class ParagraphsExplorer
 
     private final LinkedList<Paragraph> paragraphs;
     private final LinkedList<Node> nodes;
+    private final LinkedList<String> tags;
 
     public enum AncestorState
     {
@@ -51,12 +52,14 @@ public enum AncestorState
     public ParagraphsExplorer()
     {
         this.paragraphs = new LinkedList<>();
-        nodes = new LinkedList<>();
+        this.nodes = new LinkedList<>();
+        this.tags = new LinkedList<String>();
     }
 
     @Override
     public void head(Node node, int depth)
     {
+        tags.add(node.nodeName());
         if (node.childNodeSize() == 0) {
             if (node instanceof TextNode && StringUtil.isBlank(node.outerHtml())) {
                 return;
@@ -69,7 +72,7 @@ public void head(Node node, int depth)
     @Override
     public void tail(Node node, int depth)
     {
-        //do nothing
+        tags.removeLast();
     }
 
     /**
@@ -147,7 +150,7 @@ public static AncestorState getAncestorState(Node lastNode, Node currentNode)
 
     private void insertAsNewParagraph(Node node)
     {
-        Paragraph p = new Paragraph(node);
+        Paragraph p = new Paragraph(node, String.join(".", tags));
         // if (!p.getRawText().isEmpty()) {
         paragraphs.add(p);
         // }