[Archive-access-cvs] SF.net SVN: archive-access: [1745] trunk/archive-access/projects/nutchwax/ nutchwax-core/src/main/java/org/archive/access/nutch/jobs/ImportArcs.java

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

Revision: 1745
          http://archive-access.svn.sourceforge.net/archive-access/?rev=1745&view=rev
Author:   jlee-archive
Date:     2007-05-12 00:44:36 -0700 (Sat, 12 May 2007)

Log Message:
-----------
Added getArcName() and other small changes to allow subclasses of ImportArcs to specify a collectionName per AR
C, not just per import job.

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	2007-05-03 06:48:32 UTC (rev 1744)
+++ trunk/archive-access/projects/nutchwax/nutchwax-core/src/main/java/org/archive/access/nutch/jobs/ImportArcs.java	2007-05-12 07:44:36 UTC (rev 1745)
@@ -202,7 +202,7 @@
         
         job.setMapRunnerClass(job.getClass("wax.import.maprunner",
             ARCMapRunner.class));
-        job.setMapperClass(job.getClass("wax.import.mapper", ImportArcs.class));
+        job.setMapperClass(job.getClass("wax.import.mapper", this.getClass()));
         
         job.setInputFormat(TextInputFormat.class);
         
@@ -275,7 +275,7 @@
     public void onARCClose() {
         // Nothing to do.
     }   
-       
+      
     public void map(final WritableComparable key, final Writable value,
             final OutputCollector output, final Reporter r)
             throws IOException {
@@ -286,13 +286,13 @@
         ARCReporter reporter = (ARCReporter)r;
         
         // Its null first time map is called on an ARC.
-        if (this.arcName == null) {
-            this.arcName = getARCName(rec.getMetaData());
+        checkArcName(rec);
+
+        if (!isIndex(rec))
+        {
+            return;
         }
-        
-        if (!isIndex(rec)) {
-            return;
-        }        
+
         checkCollectionName();
 
         final ARCRecordMetaData arcData = rec.getMetaData();
@@ -462,7 +462,22 @@
             parse != null ? new ParseImpl(parse) : null);
         output.collect(Nutchwax.generateWaxKey(url, this.collectionName), v);
     }
-    
+ 
+    public void setCollectionName(String collectionName) {
+        this.collectionName = collectionName;
+        checkCollectionName();
+    }
+
+    public String getArcName() {
+        return this.arcName;
+    }
+
+    public void checkArcName(ARCRecord rec) {
+        if ((this.arcName == null) || this.arcName.length() <= 0) {
+            this.arcName = getARCName(rec.getMetaData());
+        }
+    }
+
     protected boolean checkCollectionName() {
         if ((this.collectionName != null) && this.collectionName.length() > 0) {
             return true;


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