aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2018-07-13 18:42:33 -0400
committerBryan Newbold <bnewbold@archive.org>2018-07-13 18:42:33 -0400
commitf955fef045d3c78e351a639546a27ab0a53fc0aa (patch)
tree5d25aa7e44d8bc196ac20f2e3a187bd0798ce134 /scalding/src/main/scala
parent0e74b5f49950c5c46d9be5e99066bee71cc8828f (diff)
parent071f4ff633eaa2669c4afcfc421f95cdc0a6765f (diff)
downloadsandcrawler-f955fef045d3c78e351a639546a27ab0a53fc0aa.tar.gz
sandcrawler-f955fef045d3c78e351a639546a27ab0a53fc0aa.zip
Merge branch 'bnewbold-args'
Diffstat (limited to 'scalding/src/main/scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseCountJob.scala10
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala21
2 files changed, 18 insertions, 13 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala
index 815478b..863c16f 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala
@@ -13,7 +13,9 @@ class HBaseCountJob(args: Args, colSpec: String) extends JobBase(args) with HBas
HBaseBuilder.parseColSpec(colSpec)
val Col: String = colSpec.split(":")(1)
- HBaseCountJob.getHBaseSource(colSpec)
+ HBaseCountJob.getHBaseSource(args("hbase-table"),
+ args("zookeeper-hosts"),
+ colSpec)
.read
.fromBytesWritable(Symbol(Col))
.debug
@@ -22,9 +24,9 @@ class HBaseCountJob(args: Args, colSpec: String) extends JobBase(args) with HBas
}
object HBaseCountJob {
- def getHBaseSource(colSpec: String) = HBaseBuilder.build(
- "wbgrp-journal-extract-0-qa", // HBase Table Name
- "mtrcs-zk1.us.archive.org:2181", // HBase Zookeeper server (to get runtime config info; can be array?)
+ def getHBaseSource(hbaseTable: String, zookeeperHosts: String, colSpec: String) = HBaseBuilder.build(
+ hbaseTable, // HBase Table Name
+ zookeeperHosts, // HBase Zookeeper server (to get runtime config info; can be array?)
List(colSpec),
SourceMode.SCAN_ALL)
}
diff --git a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
index 0c8e615..334b7a6 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
@@ -10,12 +10,10 @@ import parallelai.spyglass.hbase.HBaseConstants.SourceMode
class HBaseRowCountJob(args: Args) extends JobBase(args) with HBasePipeConversions {
-
- // For now doesn't actually count, just dumps a "word count"
-
val output = args("output")
- HBaseRowCountJob.getHBaseSource
+ HBaseRowCountJob.getHBaseSource(args("hbase-table"),
+ args("zookeeper-hosts"))
.read
.debug
.groupAll { _.size('count) }
@@ -23,9 +21,14 @@ class HBaseRowCountJob(args: Args) extends JobBase(args) with HBasePipeConversio
}
object HBaseRowCountJob {
- def getHBaseSource = HBaseBuilder.build(
- "wbgrp-journal-extract-0-qa", // HBase Table Name
- "mtrcs-zk1.us.archive.org:2181", // HBase Zookeeper server (to get runtime config info; can be array?)
- List("file:size", "file:mime"),
- SourceMode.SCAN_ALL)
+
+ // eg, "wbgrp-journal-extract-0-qa", "mtrcs-zk1.us.archive.org:2181"
+ def getHBaseSource(hbaseTable: String, zookeeperHosts: String) : HBaseSource = {
+ return HBaseBuilder.build(
+ hbaseTable,
+ zookeeperHosts,
+ List("file:size"),
+ SourceMode.SCAN_ALL)
+ }
+
}