Author: jnioche
Date: Fri Apr 8 11:05:36 2011
New Revision: 1090199
URL: http://svn.apache.org/viewvc?rev=1090199&view=rev
Log:
NUTCH-972 : CrawlDbMerger doesn't break on non-existent input (Gabriele Kahlout
via jnioche)
Modified:
nutch/branches/branch-1.3/CHANGES.txt
nutch/branches/branch-1.3/src/java/org/apache/nutch/crawl/CrawlDbMerger.java
Modified: nutch/branches/branch-1.3/CHANGES.txt
URL:
http://svn.apache.org/viewvc/nutch/branches/branch-1.3/CHANGES.txt?rev=1090199&r1=1090198&r2=1090199&view=diff
==============================================================================
--- nutch/branches/branch-1.3/CHANGES.txt (original)
+++ nutch/branches/branch-1.3/CHANGES.txt Fri Apr 8 11:05:36 2011
@@ -2,6 +2,8 @@ Nutch Change Log
Release 1.3 - Current Development
+* NUTCH-972 CrawlDbMerger doesn't break on non-existent input (Gabriele
Kahlout via jnioche)
+
* NUTCH-967 Upgrade to Tika 0.9 (jnioche)
* NUTCH-975 Fix missing/wrong headers in source files (markus)
Modified:
nutch/branches/branch-1.3/src/java/org/apache/nutch/crawl/CrawlDbMerger.java
URL:
http://svn.apache.org/viewvc/nutch/branches/branch-1.3/src/java/org/apache/nutch/crawl/CrawlDbMerger.java?rev=1090199&r1=1090198&r2=1090199&view=diff
==============================================================================
---
nutch/branches/branch-1.3/src/java/org/apache/nutch/crawl/CrawlDbMerger.java
(original)
+++
nutch/branches/branch-1.3/src/java/org/apache/nutch/crawl/CrawlDbMerger.java
Fri Apr 8 11:05:36 2011
@@ -120,6 +120,7 @@ public class CrawlDbMerger extends Confi
JobConf job = createMergeJob(getConf(), output, normalize, filter);
for (int i = 0; i < dbs.length; i++) {
+ if (LOG.isInfoEnabled()) { LOG.info("Adding " + dbs[i]); }
FileInputFormat.addInputPath(job, new Path(dbs[i],
CrawlDb.CURRENT_NAME));
}
JobClient.runJob(job);
@@ -172,6 +173,7 @@ public class CrawlDbMerger extends Confi
ArrayList<Path> dbs = new ArrayList<Path>();
boolean filter = false;
boolean normalize = false;
+ FileSystem fs = FileSystem.get(getConf());
for (int i = 1; i < args.length; i++) {
if (args[i].equals("-filter")) {
filter = true;
@@ -180,7 +182,9 @@ public class CrawlDbMerger extends Confi
normalize = true;
continue;
}
- dbs.add(new Path(args[i]));
+ final Path dbPath = new Path(args[i]);
+ if(fs.exists(dbPath))
+ dbs.add(dbPath);
}
try {
merge(output, dbs.toArray(new Path[dbs.size()]), normalize, filter);