aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2018-06-17 10:00:28 -0700
committerBryan Newbold <bnewbold@archive.org>2018-06-17 10:00:28 -0700
commit2a2921d82a5a46549e9be2c2eb5703ce28fa7579 (patch)
tree1cf775d36e590889128942711208e4b37022e73d /scalding/src/main
parent8cc18540e2663ab02428c15c2e1f17fe5624e1aa (diff)
downloadsandcrawler-2a2921d82a5a46549e9be2c2eb5703ce28fa7579.tar.gz
sandcrawler-2a2921d82a5a46549e9be2c2eb5703ce28fa7579.zip
hbase table and zk host as arguments
Diffstat (limited to 'scalding/src/main')
-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..53ce325 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(hbase_table: String, zookeeper_hosts: String, colSpec: String) = HBaseBuilder.build(
+ hbase_table, // HBase Table Name
+ zookeeper_hosts, // 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..7f5c84a 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(hbase_table: String, zookeeper_hosts: String) : HBaseSource = {
+ return HBaseBuilder.build(
+ hbase_table,
+ zookeeper_hosts,
+ List("file:size"),
+ SourceMode.SCAN_ALL)
+ }
+
}