diff options
author | Bryan Newbold <bnewbold@archive.org> | 2018-05-23 12:27:59 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2018-05-24 00:02:36 -0700 |
commit | 4ba428db30593b67283dd90b92141f99840dc78e (patch) | |
tree | f63c8e146e7f90a530abfebdb993ab45d57426d5 /scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala | |
parent | 29e4a83ff76da07bc6ad5d3f49d746ee0bc72023 (diff) | |
download | sandcrawler-4ba428db30593b67283dd90b92141f99840dc78e.tar.gz sandcrawler-4ba428db30593b67283dd90b92141f99840dc78e.zip |
rename jvm/scalding directories
Diffstat (limited to 'scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala')
-rw-r--r-- | scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala | 59 |
1 files changed, 59 insertions, 0 deletions
diff --git a/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala new file mode 100644 index 0000000..598f45d --- /dev/null +++ b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala @@ -0,0 +1,59 @@ +package example + +import org.junit.runner.RunWith +import com.twitter.scalding.{JobTest, TupleConversions} +import org.scalatest.FunSpec +import org.scalatest.junit.JUnitRunner +import org.slf4j.LoggerFactory +import org.apache.hadoop.hbase.io.ImmutableBytesWritable +import cascading.tuple.{Tuple, Fields} +import org.apache.hadoop.hbase.util.Bytes +import scala._ +import com.twitter.scalding.Tsv +import parallelai.spyglass.hbase.HBaseSource +import parallelai.spyglass.hbase.HBaseConstants.SourceMode + +/** + * Example of how to define tests for HBaseSource + */ +@RunWith(classOf[JUnitRunner]) +class HBaseRowCountTest extends FunSpec with TupleConversions { + + val output = "/tmp/testOutput" + + val log = LoggerFactory.getLogger(this.getClass.getName) + + val sampleData = List( + List("sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q", "a", "b"), + List("sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU", "a", "b") + ) + + JobTest("sandcrawler.HBaseRowCountJob") + .arg("test", "") + .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.GET_LIST, keyList = List("sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q", "sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU")), + sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*))) + .sink[Tuple](Tsv(output format "get_list")) { + outputBuffer => + log.debug("Output => " + outputBuffer) + + it("should return the test data provided.") { + println("outputBuffer.size => " + outputBuffer.size) + assert(outputBuffer.size === 2) + } + } + .run + .finish + +} |