fixes to parser

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6323 6c8d7289-2bf4-0310-a012-ef5d649a1542

source/de/anomic/document/Document.java

@@ -555,12 +555,10 @@ dc_rights
     }
     
     public int inboundLinks() {
-        assert this.inboundLinks >= 0;
         return (this.inboundLinks < 0) ? 0 : this.inboundLinks;
     }
     
     public int outboundLinks() {
-        assert this.outboundLinks >= 0;
         return (this.outboundLinks < 0) ? 0 : this.outboundLinks;
     }
     

source/de/anomic/document/parser/tarParser.java

@@ -124,6 +124,7 @@ public class tarParser extends AbstractParser implements Idiom {
                 
                 // skip directories
                 if (entry.isDirectory()) continue;
+                if (entry.getSize() <= 0) continue;
                 
                 // Get the short entry name
                 final String entryName = entry.getName();

source/de/anomic/document/parser/zipParser.java

@@ -113,6 +113,7 @@ public class zipParser extends AbstractParser implements Idiom {
                 
                 // skip directories
                 if (entry.isDirectory()) continue;
+                if (entry.getSize() <= 0) continue;
                 
                 // Get the entry name
                 final String entryName = entry.getName();                

source/de/anomic/kelondro/text/DocumentIndex.java

@@ -82,7 +82,7 @@ public class DocumentIndex extends Segment {
             throw new IOException("cannot parse " + file.toString() + ": " + e.getMessage());
         }
         final Condenser condenser = new Condenser(document, true, true);
-    	return super.storeDocument(
+        return super.storeDocument(
                 url,
                 null,
                 new Date(file.lastModified()),
@@ -119,7 +119,7 @@ public class DocumentIndex extends Segment {
                     try {
                         add(w);
                     } catch (IOException e) {
-                        e.printStackTrace();
+                        if (e.getMessage().indexOf("cannot parse") < 0) e.printStackTrace();
                     }
                 }
             }
@@ -154,6 +154,10 @@ public class DocumentIndex extends Segment {
         return find(querystring, 0, 20);
     }
     
+    public void close() {
+        super.close();
+    }
+    
     public static void main(String[] args) {
         // first argument: path to segment
         // second argument: either 'add' or 'search'