aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
diff options
context:
space:
mode:
Diffstat (limited to 'scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala')
-rw-r--r--scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala17
1 files changed, 4 insertions, 13 deletions
diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
index 3291670..7e91af3 100644
--- a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
+++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
@@ -24,10 +24,8 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions {
val log = LoggerFactory.getLogger(this.getClass.getName)
- val statusType1 : Long = 200
- val statusType2 : Long = 404
- val statusType1Bytes = Bytes.toBytes(statusType1)
- val statusType2Bytes = Bytes.toBytes(statusType2)
+ val statusType1Bytes = Bytes.toBytes("""{"status": "success"}""")
+ val statusType2Bytes = Bytes.toBytes("""{"status": "partial"}""")
// TODO(bnewbold): now to express a null (empty value) in this list?
val sampleData : List[List[Array[Byte]]] = List(
@@ -51,20 +49,13 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions {
.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