diff options
| author | bnewbold <bnewbold@archive.org> | 2018-07-13 22:02:59 +0000 |
|---|---|---|
| committer | bnewbold <bnewbold@archive.org> | 2018-07-13 22:02:59 +0000 |
| commit | 0e74b5f49950c5c46d9be5e99066bee71cc8828f (patch) | |
| tree | cc13851851bb3f20ba577881934942f54dc454c8 /scalding | |
| parent | b3149a911df8150056c54f86cd77c3516fc9838c (diff) | |
| parent | 309f1fcb9dfe6968e7b1a0640280adf1b6c1ffc2 (diff) | |
| download | sandcrawler-0e74b5f49950c5c46d9be5e99066bee71cc8828f.tar.gz sandcrawler-0e74b5f49950c5c46d9be5e99066bee71cc8828f.zip | |
Merge branch 'spertus-issue1' into 'master'
Changed grobid0:status to grobid0:status_code, per issue 1.
See merge request webgroup/sandcrawler!7
Diffstat (limited to 'scalding')
| -rw-r--r-- | scalding/src/main/scala/sandcrawler/HBaseStatusCountJob.scala | 2 | ||||
| -rw-r--r-- | scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala | 2 |
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 => |