diff options
Diffstat (limited to 'scalding/src/main')
| -rw-r--r-- | scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala | 2 | ||||
| -rw-r--r-- | scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala | 6 |
2 files changed, 4 insertions, 4 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala index a007339..20cc7a1 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala @@ -17,7 +17,7 @@ class HBaseColCountJob(args: Args) extends JobBase(args) with HBasePipeConversio HBaseColCountJob.getHBaseSource( args("hbase-table"), args("zookeeper-hosts"), - args("column") + args("column")) .read .debug .groupAll { _.size('count) } diff --git a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala index 1635e03..f79d672 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala @@ -20,13 +20,13 @@ class HBaseStatusCountJob(args: Args) extends JobBase(args) with HBasePipeConver args("zookeeper-hosts"), "grobid0:status") - val statusPipe : TypedPipe[Long] = source + val statusPipe : TypedPipe[String] = source .read .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable)]('key, 'status) - .map { case (key, raw_status) => Bytes.toString(raw_code.copyBytes()) } + .map { case (key, raw_status) => Bytes.toString(raw_status.copyBytes()) } statusPipe.groupBy { identity } .size .debug - .write(TypedTsv[(Long,String)](args("output"))) + .write(TypedTsv[(String,Long)](args("output"))) } |