diff options
author | Bryan Newbold <bnewbold@archive.org> | 2018-07-17 18:52:58 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2018-07-17 18:56:10 -0700 |
commit | bc10143c54c02d5e6a806f1d5f7bb326c24793f3 (patch) | |
tree | db1f03eef42b89faf9bcb30c5be08a8e92ef758f | |
parent | f6f2fcafb245101f75c9cda175891ef2391cda97 (diff) | |
download | sandcrawler-bc10143c54c02d5e6a806f1d5f7bb326c24793f3.tar.gz sandcrawler-bc10143c54c02d5e6a806f1d5f7bb326c24793f3.zip |
add buildSink() method for writing to HBase
-rw-r--r-- | scalding/src/main/scala/sandcrawler/HBaseBuilder.scala | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala b/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala index b271def..fd04f2e 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala @@ -1,6 +1,8 @@ package sandcrawler import cascading.tuple.Fields +import parallelai.spyglass.base.JobBase +import cascading.tap.SinkMode import parallelai.spyglass.hbase.HBaseConstants.SourceMode import parallelai.spyglass.hbase.HBaseSource import scala._ @@ -48,4 +50,9 @@ object HBaseBuilder { val (families, fields) = parseColSpecs(colSpecs) new HBaseSource(table, server, new Fields("key"), families, fields, sourceMode = sourceMode, keyList = keyList) } + + def buildSink(table: String, server: String, colSpecs: List[String], sinkMode: SinkMode, keyList: List[String] = List("key")) : HBaseSource = { + val (families, fields) = parseColSpecs(colSpecs) + new HBaseSource(table, server, new Fields("key"), families, fields, sinkMode = sinkMode) + } } |