diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2018-08-20 18:56:38 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2018-08-21 21:25:56 -0700 |
| commit | cce6047bdcf979729ec286d4c1f37676aeff2707 (patch) | |
| tree | f6774d360ad1767ddd1afc11e788b52b699eb75c /scalding/src/test | |
| parent | 9d9ff332fbbaf5194047bbd056d624a5e999b8c7 (diff) | |
| download | sandcrawler-cce6047bdcf979729ec286d4c1f37676aeff2707.tar.gz sandcrawler-cce6047bdcf979729ec286d4c1f37676aeff2707.zip | |
fix bugs/typos in HBaseColCountJob and HBaseStatusCountJob
Diffstat (limited to 'scalding/src/test')
| -rw-r--r-- | scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala | 6 | ||||
| -rw-r--r-- | scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala | 15 |
2 files changed, 7 insertions, 14 deletions
diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala index 3291670..d2cf9de 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCodeCountTest.scala @@ -17,7 +17,7 @@ import parallelai.spyglass.hbase.HBaseSource import scala._ @RunWith(classOf[JUnitRunner]) -class HBaseStatusCountTest extends FunSpec with TupleConversions { +class HBaseStatusCodeCountTest extends FunSpec with TupleConversions { val output = "/tmp/testOutput" val (testTable, testHost) = ("test-table", "dummy-host:2181") @@ -44,7 +44,7 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions { val statusType1Count = sampleData.count(lst => lst(1) == statusType1Bytes) val statusType2Count = sampleData.count(lst => lst(1) == statusType2Bytes) - JobTest("sandcrawler.HBaseStatusCountJob") + JobTest("sandcrawler.HBaseStatusCodeCountJob") .arg("test", "") .arg("app.conf.path", "app.conf") .arg("output", output) @@ -55,7 +55,7 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions { sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*))) .sink[Tuple](TypedTsv[(Long, Long)](output)) { outputBuffer => - it("should return a 2-element list.") { + it("should return a correct number of elements.") { assert(outputBuffer.size === 2) } diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala index b53abeb..7e91af3 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala @@ -17,7 +17,7 @@ import parallelai.spyglass.hbase.HBaseSource import scala._ @RunWith(classOf[JUnitRunner]) -class HBaseStatusCodeCountTest extends FunSpec with TupleConversions { +class HBaseStatusCountTest extends FunSpec with TupleConversions { val output = "/tmp/testOutput" val (testTable, testHost) = ("test-table", "dummy-host:2181") @@ -42,27 +42,20 @@ class HBaseStatusCodeCountTest extends FunSpec with TupleConversions { val statusType1Count = sampleData.count(lst => lst(1) == statusType1Bytes) val statusType2Count = sampleData.count(lst => lst(1) == statusType2Bytes) - JobTest("sandcrawler.HBaseStatusCodeCountJob") + JobTest("sandcrawler.HBaseStatusCountJob") .arg("test", "") .arg("app.conf.path", "app.conf") .arg("output", output) .arg("hbase-table", testTable) .arg("zookeeper-hosts", testHost) .arg("debug", "true") - .source[Tuple](HBaseCountJob.getHBaseSource(testTable, testHost, "grobid0:status_code"), + .source[Tuple](HBaseCountJob.getHBaseSource(testTable, testHost, "grobid0:status"), sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*))) - .sink[Tuple](TypedTsv[(Long, Long)](output)) { + .sink[Tuple](TypedTsv[(String, Long)](output)) { outputBuffer => it("should return a 2-element list.") { assert(outputBuffer.size === 2) } - - // Convert List[Tuple] to Map[Long, Long]. - val counts = outputBuffer.map(t => (t.getLong(0), t.getLong(1))).toMap - it("should have the appropriate number of each status type") { - assert(counts(statusType1) == statusType1Count) - assert(counts(statusType2) == statusType2Count) - } } .run .finish |