From 500525b82244151ed3e64d1cf31e96df394b5250 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Tue, 17 Jul 2018 18:52:58 -0700 Subject: add buildSink() method for writing to HBase --- scalding/src/main/scala/sandcrawler/HBaseBuilder.scala | 7 +++++++ 1 file changed, 7 insertions(+) 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) + } } -- cgit v1.2.3