diff options
author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-06-08 17:15:02 -0700 |
---|---|---|
committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-06-08 17:15:02 -0700 |
commit | 8d8ef0d8d19350228ae4d58267b156519ad1fc35 (patch) | |
tree | 026493d44050206eb208f9c6b79e8ef490eff4f6 /scalding/src/main/scala/sandcrawler/HBaseCountJob.scala | |
parent | 410c48faf2099de74292e8583fcd2524d6fd1b7c (diff) | |
download | sandcrawler-8d8ef0d8d19350228ae4d58267b156519ad1fc35.tar.gz sandcrawler-8d8ef0d8d19350228ae4d58267b156519ad1fc35.zip |
Tests pass.
Diffstat (limited to 'scalding/src/main/scala/sandcrawler/HBaseCountJob.scala')
-rw-r--r-- | scalding/src/main/scala/sandcrawler/HBaseCountJob.scala | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala new file mode 100644 index 0000000..250f175 --- /dev/null +++ b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala @@ -0,0 +1,30 @@ +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 HBaseCountJob(args: Args, colSpec: String) extends JobBase(args) with HBasePipeConversions { + val output = args("output") + // TODO: Add error checking. + val Col: String = colSpec.split(":")(1) + + HBaseCountJob.getHBaseSource(colSpec) + .read + .fromBytesWritable(Symbol(Col)) + .debug + .groupBy(Col){group => group.size('count)} + .write(Tsv(output)) +} + +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?) + List(colSpec), + SourceMode.SCAN_ALL) +} |