aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2018-06-08 15:06:39 -0700
committerBryan Newbold <bnewbold@archive.org>2018-06-08 16:21:01 -0700
commit82e24b8de7776e355a15afb5f442356bc4ad2d81 (patch)
tree5f86ec4cda158361f8903262ce23c715fd979db3 /scalding/src/main/scala
parent410c48faf2099de74292e8583fcd2524d6fd1b7c (diff)
downloadsandcrawler-bnewbold-args.tar.gz
sandcrawler-bnewbold-args.zip
make hbase_table and zookeeper_hosts CLI argsbnewbold-args
Diffstat (limited to 'scalding/src/main/scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala18
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala21
2 files changed, 24 insertions, 15 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
index 819a652..a6ab53a 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
@@ -9,9 +9,11 @@ import parallelai.spyglass.hbase.{HBaseSource, HBasePipeConversions}
import parallelai.spyglass.hbase.HBaseConstants.SourceMode
class HBaseMimeCountJob(args: Args) extends JobBase(args) with HBasePipeConversions {
+
val output = args("output")
- HBaseMimeCountJob.getHBaseSource
+ HBaseMimeCountJob.getHBaseSource(args("hbase-table"),
+ args("zookeeper-hosts"))
.read
.fromBytesWritable(List('mime))
.debug
@@ -20,9 +22,13 @@ class HBaseMimeCountJob(args: Args) extends JobBase(args) with HBasePipeConversi
}
object HBaseMimeCountJob {
- 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:mime"),
- SourceMode.SCAN_ALL)
+
+ def getHBaseSource(hbase_table: String, zookeeper_hosts: String) : HBaseSource = {
+ return HBaseBuilder.build(
+ hbase_table,
+ zookeeper_hosts,
+ List("file:mime"),
+ 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)
+ }
+
}