aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorEllen Spertus <ellen.spertus@gmail.com>2018-06-18 11:28:22 -0700
committerEllen Spertus <ellen.spertus@gmail.com>2018-06-18 11:28:22 -0700
commit309f1fcb9dfe6968e7b1a0640280adf1b6c1ffc2 (patch)
treeaab70c271262d1c2f367878278977426399bf9ed
parent3e9b927ba668f13f818d20962a6fb6f5783407df (diff)
downloadsandcrawler-309f1fcb9dfe6968e7b1a0640280adf1b6c1ffc2.tar.gz
sandcrawler-309f1fcb9dfe6968e7b1a0640280adf1b6c1ffc2.zip
Changed grobid0:status to grobid0:status_code, per issue 1.
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala2
-rw-r--r--scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala2
2 files changed, 2 insertions, 2 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
index 27b3177..aabf9f8 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala
@@ -2,4 +2,4 @@ package sandcrawler
import com.twitter.scalding.Args
-class HBaseStatusCountJob(args: Args) extends HBaseCountJob(args, "grobid0:status")
+class HBaseStatusCountJob(args: Args) extends HBaseCountJob(args, "grobid0:status_code")
diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
index a0b313e..056347c 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](HBaseCountJob.getHBaseSource("grobid0:status"),
+ .source[Tuple](HBaseCountJob.getHBaseSource("grobid0:status_code"),
sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*)))
.sink[Tuple](Tsv(output)) {
outputBuffer =>