From 51ca75189b7c36577f8e80b9db1f66259f0f6178 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Sat, 22 Sep 2018 20:31:39 -0700 Subject: new DumpGrobidMetaInsertableJob --- .../sandcrawler/DumpGrobidMetaInsertableJob.scala | 38 ++++++++++++++++++++++ 1 file changed, 38 insertions(+) create mode 100644 scalding/src/main/scala/sandcrawler/DumpGrobidMetaInsertableJob.scala (limited to 'scalding/src/main/scala') diff --git a/scalding/src/main/scala/sandcrawler/DumpGrobidMetaInsertableJob.scala b/scalding/src/main/scala/sandcrawler/DumpGrobidMetaInsertableJob.scala new file mode 100644 index 0000000..ee2b7c2 --- /dev/null +++ b/scalding/src/main/scala/sandcrawler/DumpGrobidMetaInsertableJob.scala @@ -0,0 +1,38 @@ +package sandcrawler + +import java.util.Properties + +import cascading.property.AppProps +import cascading.tuple.Fields +import com.twitter.scalding._ +import com.twitter.scalding.typed.TDsl._ +import org.apache.hadoop.hbase.io.ImmutableBytesWritable +import org.apache.hadoop.hbase.util.Bytes +import parallelai.spyglass.base.JobBase +import parallelai.spyglass.hbase.HBaseConstants.SourceMode +import parallelai.spyglass.hbase.HBasePipeConversions +import parallelai.spyglass.hbase.HBaseSource + +// Dumps the SHA1 key and grobid0:metadata columns, plus file metadata needed +// to insert into fatcat. Used, eg, as part of long-tail mellon pipeline. +class DumpGrobidMetaInsertableJob(args: Args) extends JobBase(args) with HBasePipeConversions { + + val metaPipe : TypedPipe[(String, String, String, Long, String)] = HBaseBuilder.build(args("hbase-table"), + args("zookeeper-hosts"), + List("file:cdx", "file:mime", "file:size", "grobid0:metadata"), + SourceMode.SCAN_ALL) + .read + .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "cdx", "mime", "size", "metadata")) + .filter { case (_, cdx, mime, size, metadata) => cdx != null && mime != null && size != null && metadata != null } + .map { case (key, cdx, mime, size, metadata) => + (Bytes.toString(key.copyBytes()), + Bytes.toString(cdx.copyBytes()), + Bytes.toString(mime.copyBytes()), + Bytes.toLong(size.copyBytes()), + Bytes.toString(metadata.copyBytes()) + ) + }; + + metaPipe.write(TypedTsv[(String,String,String,Long,String)](args("output"))) + +} -- cgit v1.2.3