diff options
Diffstat (limited to 'scalding/src/main')
3 files changed, 4 insertions, 4 deletions
| diff --git a/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala b/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala index 85766d6..b271def 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala @@ -44,7 +44,7 @@ object HBaseBuilder {      (families, groupedColNames.map({fields => new Fields(fields : _*)}))    } -  def build(table: String, server: String, colSpecs: List[String], sourceMode: SourceMode, keyList: List[String] = List("key")) = { +  def build(table: String, server: String, colSpecs: List[String], sourceMode: SourceMode, keyList: List[String] = List("key")) : HBaseSource = {      val (families, fields) = parseColSpecs(colSpecs)      new HBaseSource(table, server, new Fields("key"), families, fields, sourceMode = sourceMode, keyList = keyList)    } diff --git a/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala index 863c16f..1ebc261 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala @@ -24,7 +24,7 @@ class HBaseCountJob(args: Args, colSpec: String) extends JobBase(args) with HBas  }  object HBaseCountJob { -  def getHBaseSource(hbaseTable: String, zookeeperHosts: String, colSpec: String) = HBaseBuilder.build( +  def getHBaseSource(hbaseTable: String, zookeeperHosts: String, colSpec: String) : HBaseSource = HBaseBuilder.build(      hbaseTable,      // HBase Table Name      zookeeperHosts,  // HBase Zookeeper server (to get runtime config info; can be array?)      List(colSpec), diff --git a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala index 334b7a6..ba3b9cd 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala @@ -22,9 +22,9 @@ class HBaseRowCountJob(args: Args) extends JobBase(args) with HBasePipeConversio  object HBaseRowCountJob { -  // eg, "wbgrp-journal-extract-0-qa", "mtrcs-zk1.us.archive.org:2181" +  // eg, "wbgrp-journal-extract-0-qa",7 "mtrcs-zk1.us.archive.org:2181"    def getHBaseSource(hbaseTable: String, zookeeperHosts: String) : HBaseSource = { -    return HBaseBuilder.build( +    HBaseBuilder.build(        hbaseTable,        zookeeperHosts,        List("file:size"), | 
