aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
diff options
context:
space:
mode:
Diffstat (limited to 'scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala10
1 files changed, 5 insertions, 5 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
index fd0b4e2..f79d672 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
@@ -18,15 +18,15 @@ class HBaseStatusCountJob(args: Args) extends JobBase(args) with HBasePipeConver
val source = HBaseCountJob.getHBaseSource(
args("hbase-table"),
args("zookeeper-hosts"),
- "grobid0:status_code")
+ "grobid0:status")
- val statusPipe : TypedPipe[Long] = source
+ val statusPipe : TypedPipe[String] = source
.read
- .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable)]('key, 'status_code)
- .map { case (key, raw_code) => Bytes.toLong(raw_code.copyBytes()) }
+ .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable)]('key, 'status)
+ .map { case (key, raw_status) => Bytes.toString(raw_status.copyBytes()) }
statusPipe.groupBy { identity }
.size
.debug
- .write(TypedTsv[(Long,Long)](args("output")))
+ .write(TypedTsv[(String,Long)](args("output")))
}