diff options
author | Bryan Newbold <bnewbold@archive.org> | 2018-06-17 10:00:28 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2018-06-17 10:00:28 -0700 |
commit | 2a2921d82a5a46549e9be2c2eb5703ce28fa7579 (patch) | |
tree | 1cf775d36e590889128942711208e4b37022e73d /scalding/src/test | |
parent | 8cc18540e2663ab02428c15c2e1f17fe5624e1aa (diff) | |
download | sandcrawler-2a2921d82a5a46549e9be2c2eb5703ce28fa7579.tar.gz sandcrawler-2a2921d82a5a46549e9be2c2eb5703ce28fa7579.zip |
hbase table and zk host as arguments
Diffstat (limited to 'scalding/src/test')
3 files changed, 12 insertions, 3 deletions
diff --git a/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala index cc400c5..fcd9087 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala @@ -16,6 +16,7 @@ import scala._ class HBaseMimeCountTest extends FunSpec with TupleConversions { val output = "/tmp/testOutput" + val (test_table, test_host) = ("test-table", "dummy-host:2181") val log = LoggerFactory.getLogger(this.getClass.getName) @@ -40,8 +41,10 @@ class HBaseMimeCountTest extends FunSpec with TupleConversions { .arg("test", "") .arg("app.conf.path", "app.conf") .arg("output", output) + .arg("hbase-table", test_table) + .arg("zookeeper-hosts", test_host) .arg("debug", "true") - .source[Tuple](HBaseCountJob.getHBaseSource("file:mime"), + .source[Tuple](HBaseCountJob.getHBaseSource(test_table, test_host, "file:mime"), sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*))) .sink[Tuple](Tsv(output)) { outputBuffer => diff --git a/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala index 6f61eb3..a5c35a2 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala @@ -19,6 +19,7 @@ import scala._ class HBaseRowCountTest extends FunSpec with TupleConversions { val output = "/tmp/testOutput" + val (test_table, test_host) = ("test-table", "dummy-host:2181") val log = LoggerFactory.getLogger(this.getClass.getName) @@ -37,8 +38,10 @@ class HBaseRowCountTest extends FunSpec with TupleConversions { .arg("test", "") .arg("app.conf.path", "app.conf") .arg("output", output) + .arg("hbase-table", test_table) + .arg("zookeeper-hosts", test_host) .arg("debug", "true") - .source[Tuple](HBaseRowCountJob.getHBaseSource, + .source[Tuple](HBaseRowCountJob.getHBaseSource(test_table, test_host), sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*))) .sink[Tuple](Tsv(output)) { outputBuffer => diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala index a0b313e..1877d6f 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala @@ -16,6 +16,7 @@ import scala._ class HBaseStatusCountTest extends FunSpec with TupleConversions { val output = "/tmp/testOutput" + val (test_table, test_host) = ("test-table", "dummy-host:2181") val log = LoggerFactory.getLogger(this.getClass.getName) @@ -40,8 +41,10 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions { .arg("test", "") .arg("app.conf.path", "app.conf") .arg("output", output) + .arg("hbase-table", test_table) + .arg("zookeeper-hosts", test_host) .arg("debug", "true") - .source[Tuple](HBaseCountJob.getHBaseSource("grobid0:status"), + .source[Tuple](HBaseCountJob.getHBaseSource(test_table, test_host, "grobid0:status"), sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*))) .sink[Tuple](Tsv(output)) { outputBuffer => |