diff options
| author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-06-01 14:54:17 -0700 |
|---|---|---|
| committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-06-01 14:54:17 -0700 |
| commit | 59c2827bcd682a6c78714438ef8d7aaca2ccf13b (patch) | |
| tree | 7685d226d56bc2c939665b9be6ebf3f94e2a7d63 /scalding/src/test | |
| parent | 8e3a5ba6f3bdfad32e9bfefeebcb85bbe254cbc9 (diff) | |
| download | sandcrawler-59c2827bcd682a6c78714438ef8d7aaca2ccf13b.tar.gz sandcrawler-59c2827bcd682a6c78714438ef8d7aaca2ccf13b.zip | |
Factored common code out of HBaseRowCountJob and its test into a new companion object.
Diffstat (limited to 'scalding/src/test')
| -rw-r--r-- | scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala | 14 |
1 files changed, 3 insertions, 11 deletions
diff --git a/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala index 94b3740..abb017c 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala @@ -12,6 +12,7 @@ import scala._ import com.twitter.scalding.Tsv import parallelai.spyglass.hbase.HBaseSource import parallelai.spyglass.hbase.HBaseConstants.SourceMode +import sandcrawler.HBaseRowCountJob /** * Example of how to define tests for HBaseSource @@ -39,18 +40,9 @@ class HBaseRowCountTest extends FunSpec with TupleConversions { .arg("app.conf.path", "app.conf") .arg("output", output) .arg("debug", "true") - .source[Tuple]( - new HBaseSource( - //"table_name", - //"quorum_name:2181", - "wbgrp-journal-extract-0-qa", - "mtrcs-zk1.us.archive.org:2181", - new Fields("key"), - List("file"), - List(new Fields("size", "mimetype")), - sourceMode = SourceMode.SCAN_ALL), + .source[Tuple](HBaseRowCountJob.getHBaseSource, sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*))) - .sink[Tuple](Tsv(output)) { + .sink[Tuple](Tsv(output)) { outputBuffer => it("should return the test data provided.") { |