aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
diff options
context:
space:
mode:
authorEllen Spertus <ellen.spertus@gmail.com>2018-06-08 17:15:02 -0700
committerEllen Spertus <ellen.spertus@gmail.com>2018-06-08 17:15:02 -0700
commit8d8ef0d8d19350228ae4d58267b156519ad1fc35 (patch)
tree026493d44050206eb208f9c6b79e8ef490eff4f6 /scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
parent410c48faf2099de74292e8583fcd2524d6fd1b7c (diff)
downloadsandcrawler-8d8ef0d8d19350228ae4d58267b156519ad1fc35.tar.gz
sandcrawler-8d8ef0d8d19350228ae4d58267b156519ad1fc35.zip
Tests pass.
Diffstat (limited to 'scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala18
1 files changed, 1 insertions, 17 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
index 819a652..911ccb8 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
@@ -8,21 +8,5 @@ import parallelai.spyglass.base.JobBase
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")
+class HBaseMimeCountJob(args: Args) extends HBaseCountJob(args, "file:mime") {}
- HBaseMimeCountJob.getHBaseSource
- .read
- .fromBytesWritable(List('mime))
- .debug
- .groupBy('mime){group => group.size('count)}
- .write(Tsv(output))
-}
-
-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)
-}