aboutsummaryrefslogtreecommitdiffstats
path: root/scalding
diff options
context:
space:
mode:
Diffstat (limited to 'scalding')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala27
-rw-r--r--scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala2
2 files changed, 3 insertions, 26 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
index 0675efc..27b3177 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
@@ -1,28 +1,5 @@
package sandcrawler
-import cascading.property.AppProps
-import cascading.tuple.Fields
-import com.twitter.scalding._
-import java.util.Properties
-import parallelai.spyglass.base.JobBase
-import parallelai.spyglass.hbase.{HBaseSource, HBasePipeConversions}
-import parallelai.spyglass.hbase.HBaseConstants.SourceMode
+import com.twitter.scalding.Args
-class HBaseStatusCountJob(args: Args) extends JobBase(args) with HBasePipeConversions {
- val output = args("output")
-
- HBaseStatusCountJob.getHBaseSource
- .read
- .fromBytesWritable(List('status))
-// .debug
- .groupBy('status){group => group.size('count)}
- .write(Tsv(output))
-}
-
-object HBaseStatusCountJob {
- def getHBaseSource = HBaseBuilder.build(
- "wbgrp-journal-extract-0-qa", // HBase Table Name
- "mtrcs-zk1.us.archive.org:2181", // HBase Zookeeper server (to get runtime config info; can be array?)
- List("grobid0:status"),
- SourceMode.SCAN_ALL)
-}
+class HBaseStatusCountJob(args: Args) extends HBaseCountJob(args, "grobid0:status")
diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
index 8b5c3d6..a0b313e 100644
--- a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
+++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
@@ -41,7 +41,7 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions {
.arg("app.conf.path", "app.conf")
.arg("output", output)
.arg("debug", "true")
- .source[Tuple](HBaseStatusCountJob.getHBaseSource,
+ .source[Tuple](HBaseCountJob.getHBaseSource("grobid0:status"),
sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*)))
.sink[Tuple](Tsv(output)) {
outputBuffer =>