Author: lewismc
Date: Tue Mar 18 22:07:25 2014
New Revision: 1579072
URL: http://svn.apache.org/r1579072
Log:
NUTCH-1738 Expose number of URLs generated per batch in GeneratorJob
Modified:
nutch/branches/2.x/CHANGES.txt
nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorJob.java
nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorReducer.java
Modified: nutch/branches/2.x/CHANGES.txt
URL:
http://svn.apache.org/viewvc/nutch/branches/2.x/CHANGES.txt?rev=1579072&r1=1579071&r2=1579072&view=diff
==============================================================================
--- nutch/branches/2.x/CHANGES.txt (original)
+++ nutch/branches/2.x/CHANGES.txt Tue Mar 18 22:07:25 2014
@@ -2,6 +2,8 @@ Nutch Change Log
Current Development
+* NUTCH-1738 Expose number of URLs generated per batch in GeneratorJob (Talat
UYARER via lewismc)
+
* NUTCH-1671 indexchecker to add digest field (snagel, lufeng)
* NUTCH-1645 Junit Test Case for Adaptive Fetch Schedule class (Yasin
Kılınç, lufeng, Sertac TURKEL via snagel)
Modified: nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorJob.java
URL:
http://svn.apache.org/viewvc/nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorJob.java?rev=1579072&r1=1579071&r2=1579072&view=diff
==============================================================================
--- nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorJob.java
(original)
+++ nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorJob.java Tue
Mar 18 22:07:25 2014
@@ -228,7 +228,7 @@ public class GeneratorJob extends NutchT
String batchId = getConf().get(BATCH_ID);
long finish = System.currentTimeMillis();
LOG.info("GeneratorJob: finished at " + sdf.format(finish) + ", time
elapsed: " + TimingUtil.elapsedTime(start, finish));
- LOG.info("GeneratorJob: generated batch id: " + batchId);
+ LOG.info("GeneratorJob: generated batch id: " + batchId + " containing " +
GeneratorReducer.count + " URLs");
return batchId;
}
Modified:
nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorReducer.java
URL:
http://svn.apache.org/viewvc/nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorReducer.java?rev=1579072&r1=1579071&r2=1579072&view=diff
==============================================================================
--- nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorReducer.java
(original)
+++ nutch/branches/2.x/src/java/org/apache/nutch/crawl/GeneratorReducer.java
Tue Mar 18 22:07:25 2014
@@ -42,7 +42,7 @@ extends GoraReducer<SelectorEntry, WebPa
private long limit;
private long maxCount;
- private long count = 0;
+ protected static long count = 0;
private boolean byDomain = false;
private Map<String, Integer> hostCountMap = new HashMap<String, Integer>();
private Utf8 batchId;