From 7c49fecd01cbd89bb5987442b89a4aafc1186ff9 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Mon, 20 Aug 2018 18:50:07 -0700
Subject: make col counter generic

---
 .../scala/sandcrawler/GrobidMetadataCountJob.scala | 36 ---------------------
 .../main/scala/sandcrawler/HBaseColCountJob.scala  | 37 ++++++++++++++++++++++
 2 files changed, 37 insertions(+), 36 deletions(-)
 delete mode 100644 scalding/src/main/scala/sandcrawler/GrobidMetadataCountJob.scala
 create mode 100644 scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala

(limited to 'scalding')

diff --git a/scalding/src/main/scala/sandcrawler/GrobidMetadataCountJob.scala b/scalding/src/main/scala/sandcrawler/GrobidMetadataCountJob.scala
deleted file mode 100644
index 08f3340..0000000
--- a/scalding/src/main/scala/sandcrawler/GrobidMetadataCountJob.scala
+++ /dev/null
@@ -1,36 +0,0 @@
-package sandcrawler
-
-import java.util.Properties
-
-import cascading.property.AppProps
-import cascading.tuple.Fields
-import com.twitter.scalding._
-import parallelai.spyglass.base.JobBase
-import parallelai.spyglass.hbase.HBaseConstants.SourceMode
-import parallelai.spyglass.hbase.HBasePipeConversions
-import parallelai.spyglass.hbase.HBaseSource
-
-class GrobidMetadataCountJob(args: Args) extends JobBase(args) with HBasePipeConversions {
-
-  val output = args("output")
-
-  GrobidMetadataCountJob.getHBaseSource(
-    args("hbase-table"),
-    args("zookeeper-hosts"))
-    .read
-    .debug
-    .groupAll { _.size('count) }
-    .write(Tsv(output))
-}
-
-object GrobidMetadataCountJob {
-
-  // eg, "wbgrp-journal-extract-0-qa",7 "mtrcs-zk1.us.archive.org:2181"
-  def getHBaseSource(hbaseTable: String, zookeeperHosts: String) : HBaseSource = {
-    HBaseBuilder.build(
-      hbaseTable,
-      zookeeperHosts,
-      List("grobid0:metadata"),
-      SourceMode.SCAN_ALL)
-  }
-}
diff --git a/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala
new file mode 100644
index 0000000..a007339
--- /dev/null
+++ b/scalding/src/main/scala/sandcrawler/HBaseColCountJob.scala
@@ -0,0 +1,37 @@
+package sandcrawler
+
+import java.util.Properties
+
+import cascading.property.AppProps
+import cascading.tuple.Fields
+import com.twitter.scalding._
+import parallelai.spyglass.base.JobBase
+import parallelai.spyglass.hbase.HBaseConstants.SourceMode
+import parallelai.spyglass.hbase.HBasePipeConversions
+import parallelai.spyglass.hbase.HBaseSource
+
+class HBaseColCountJob(args: Args) extends JobBase(args) with HBasePipeConversions {
+
+  val output = args("output")
+
+  HBaseColCountJob.getHBaseSource(
+    args("hbase-table"),
+    args("zookeeper-hosts"),
+    args("column")
+    .read
+    .debug
+    .groupAll { _.size('count) }
+    .write(Tsv(output))
+}
+
+object HBaseColCountJob {
+
+  // eg, "wbgrp-journal-extract-0-qa",7 "mtrcs-zk1.us.archive.org:2181"
+  def getHBaseSource(hbaseTable: String, zookeeperHosts: String, col: String) : HBaseSource = {
+    HBaseBuilder.build(
+      hbaseTable,
+      zookeeperHosts,
+      List(col),
+      SourceMode.SCAN_ALL)
+  }
+}
-- 
cgit v1.2.3