diff options
Diffstat (limited to 'scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala')
-rw-r--r-- | scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala | 41 |
1 files changed, 16 insertions, 25 deletions
diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala index 0da0b9c..7e91af3 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala @@ -24,22 +24,20 @@ 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 sampleData : List[List[Array[Byte]]] = List( - // TODO(bnewbold): now to express a null (empty value) in this list? - List(Bytes.toBytes("sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q"), statusType1Bytes), - List(Bytes.toBytes("sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU"), statusType1Bytes), - List(Bytes.toBytes("sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT"), statusType2Bytes), - List(Bytes.toBytes("sha1:35985C3YNNEGH5WAG5ZAAXWAEBNXJW56"), statusType2Bytes), - List(Bytes.toBytes("sha1:885C3YNNEGH5WAG5ZAAXWA8BNXJWT6CZ"), statusType2Bytes), - List(Bytes.toBytes("sha1:00904C3YNNEGH5WAG5ZA9XWAEBNXJWT6"), statusType2Bytes), - List(Bytes.toBytes("sha1:249C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ"), statusType1Bytes), - List(Bytes.toBytes("sha1:095893C3YNNEGH5WAG5ZAAXWAEBNXJWT"), statusType2Bytes) - ) + 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( + ("sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q", statusType1Bytes), + ("sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU", statusType1Bytes), + ("sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT", statusType2Bytes), + ("sha1:35985C3YNNEGH5WAG5ZAAXWAEBNXJW56", statusType2Bytes), + ("sha1:885C3YNNEGH5WAG5ZAAXWA8BNXJWT6CZ", statusType2Bytes), + ("sha1:00904C3YNNEGH5WAG5ZA9XWAEBNXJWT6", statusType2Bytes), + ("sha1:249C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ", statusType1Bytes), + ("sha1:095893C3YNNEGH5WAG5ZAAXWAEBNXJWT", statusType2Bytes)) + .map(pair => List(Bytes.toBytes(pair._1), pair._2)) val statusType1Count = sampleData.count(lst => lst(1) == statusType1Bytes) val statusType2Count = sampleData.count(lst => lst(1) == statusType2Bytes) @@ -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 |