diff options
Diffstat (limited to 'scalding/src/main/scala/sandcrawler')
| -rw-r--r-- | scalding/src/main/scala/sandcrawler/DumpUnGrobidedJob.scala (renamed from scalding/src/main/scala/sandcrawler/UnGrobidedDumpJob.scala) | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/scalding/src/main/scala/sandcrawler/UnGrobidedDumpJob.scala b/scalding/src/main/scala/sandcrawler/DumpUnGrobidedJob.scala index 0ce9167..7fd3ce0 100644 --- a/scalding/src/main/scala/sandcrawler/UnGrobidedDumpJob.scala +++ b/scalding/src/main/scala/sandcrawler/DumpUnGrobidedJob.scala @@ -15,19 +15,19 @@ import parallelai.spyglass.hbase.HBaseSource // full CDX metadata, and dumps to a TSV for later extraction by the // "extraction-ungrobided" job. // -// Does the same horrible join thing that UnGrobidedDumpJob does. -class UnGrobidedDumpJob(args: Args) extends JobBase(args) with HBasePipeConversions { +// Does the same horrible join thing that DumpUnGrobidedJob does. +class DumpUnGrobidedJob(args: Args) extends JobBase(args) with HBasePipeConversions { val output = args("output") - val allKeys : TypedPipe[(String,String,String,String)] = UnGrobidedDumpJob.getHBaseKeySource( + val allKeys : TypedPipe[(String,String,String,String)] = DumpUnGrobidedJob.getHBaseKeySource( args("hbase-table"), args("zookeeper-hosts")) .read .fromBytesWritable('key, 'c, 'mime, 'cdx) .toTypedPipe[(String,String,String,String)]('key, 'c, 'mime, 'cdx) - val existingKeys : TypedPipe[(String,Boolean)] = UnGrobidedDumpJob.getHBaseColSource( + val existingKeys : TypedPipe[(String,Boolean)] = DumpUnGrobidedJob.getHBaseColSource( args("hbase-table"), args("zookeeper-hosts")) .read @@ -46,7 +46,7 @@ class UnGrobidedDumpJob(args: Args) extends JobBase(args) with HBasePipeConversi } -object UnGrobidedDumpJob { +object DumpUnGrobidedJob { // eg, "wbgrp-journal-extract-0-qa",7 "mtrcs-zk1.us.archive.org:2181" def getHBaseColSource(hbaseTable: String, zookeeperHosts: String) : HBaseSource = { |