[Archive-access-cvs] SF.net SVN: archive-access: [2067] trunk/archive-access/projects/nutchwax/ nutchwax-core/src/main/java/org/archive/access/nutch/jobs/ImportArcs.java

[<jle...@us...>]

Revision: 2067
          http://archive-access.svn.sourceforge.net/archive-access/?rev=2067&view=rev
Author:   jlee-archive
Date:     2007-11-02 19:55:42 -0700 (Fri, 02 Nov 2007)

Log Message:
-----------
Use the arcname from the ARCRecordMetaData instead of trying to parse it out of 
the filedesc.

Modified Paths:
--------------
    trunk/archive-access/projects/nutchwax/nutchwax-core/src/main/java/org/archive/access/nutch/jobs/ImportArcs.java

Modified: trunk/archive-access/projects/nutchwax/nutchwax-core/src/main/java/org/archive/access/nutch/jobs/ImportArcs.java
===================================================================
--- trunk/archive-access/projects/nutchwax/nutchwax-core/src/main/java/org/archive/access/nutch/jobs/ImportArcs.java	2007-11-03 01:29:12 UTC (rev 2066)
+++ trunk/archive-access/projects/nutchwax/nutchwax-core/src/main/java/org/archive/access/nutch/jobs/ImportArcs.java	2007-11-03 02:55:42 UTC (rev 2067)
@@ -306,7 +306,7 @@
     }
 
     checkCollectionName();
-
+    
     final ARCRecordMetaData arcData = rec.getMetaData();
     String oldUrl = url;
     
@@ -320,7 +320,7 @@
     {
       LOG.warn("Skipping record. Didn't pass normalization/filter " +
         oldUrl + ": " + e.toString());
-      
+
       return;
     }
 
@@ -546,7 +546,8 @@
   {
     if ((this.arcName == null) || this.arcName.length() <= 0)
     {
-      this.arcName = getARCName(rec.getMetaData());
+      this.arcName = rec.getMetaData().getArcFile().getName();
+      this.arcName = this.arcName.replace(".arc.gz", "");
     }
   }
 
@@ -571,47 +572,6 @@
       (rec.getStatusCode() < 400)));
   }
 
- /**
-  * @param firstARCRecordMeta The metadata record of the first record in an
-  * ARC (the filedesc record).
-  * @return Trimmed ARCName stripped of path preamble/prefix and suffix
-  *  (At least WERA expects an ARC name without scheme and suffix: i.e.
-  * IAH-20060315203614-00000-debord).
-  * @throws NullPointerException If unable to find an ARC name.
-  */
-  protected String getARCName(final ARCRecordMetaData firstARCRecordMeta)
-  {
-    String result = null;
-    
-    if (this.arcNameFromFirstRecord)
-    {
-      final Matcher m = FILEDESC_PATTERN.matcher(firstARCRecordMeta.getUrl());
-      
-      if ((m != null) && m.matches())
-      {
-        result = m.group(1);
-      }
-    }
-    else
-    {
-      final Matcher m = TAIL_PATTERN.matcher(
-        firstARCRecordMeta.getReaderIdentifier());
-      
-      if ((m != null) && m.matches())
-      {
-        result = m.group(1);
-      }
-    }
-    
-    if (result == null || result.length() <= 0)
-    {
-      throw new NullPointerException("Failed get of arcname: " +
-        firstARCRecordMeta);
-    }
-    
-    return result;
-  }
-
   protected String getStatus(final String url, String oldUrl,
     final String recordLengthAsStr, final String noSpacesMimetype)
   {


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.