From cc54dea9c77f8386bafd71977b2be1b3b21f0aae Mon Sep 17 00:00:00 2001
From: Ellen Spertus <ellen.spertus@gmail.com>
Date: Wed, 6 Jun 2018 09:40:04 -0700
Subject: Added job and test for counting mime types.

---
 .../main/scala/sandcrawler/HBaseMimeCountJob.scala | 28 ++++++++++++++++++++++
 1 file changed, 28 insertions(+)
 create mode 100644 scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala

(limited to 'scalding/src/main')

diff --git a/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
new file mode 100644
index 0000000..819a652
--- /dev/null
+++ b/scalding/src/main/scala/sandcrawler/HBaseMimeCountJob.scala
@@ -0,0 +1,28 @@
+package sandcrawler
+
+import cascading.property.AppProps
+import cascading.tuple.Fields
+import com.twitter.scalding._
+import java.util.Properties
+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")
+
+  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)
+}
-- 
cgit v1.2.3