diff options
author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-08-06 16:38:46 -0700 |
---|---|---|
committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-08-06 16:38:46 -0700 |
commit | 308b33d889d804380427d2aa112efec77b3e1770 (patch) | |
tree | e0181fd4eb2d311bf45827447afe5d93291f931f /scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala | |
parent | b1d8a72a5cc469b5139d9a976ccfa9b4b3eea61d (diff) | |
download | sandcrawler-308b33d889d804380427d2aa112efec77b3e1770.tar.gz sandcrawler-308b33d889d804380427d2aa112efec77b3e1770.zip |
New code compiles. Old tests pass. New tests not yet written.
Diffstat (limited to 'scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala')
-rw-r--r-- | scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala b/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala index 01d852e..2fbb19f 100644 --- a/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala +++ b/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala @@ -27,8 +27,9 @@ class HBaseCrossrefScoreJob(args: Args) extends JobBase(args) with HBasePipeConv val grobidSource = HBaseCrossrefScore.getHBaseSource( args("hbase-table"), args("zookeeper-hosts")) - val grobidPipe : TypedPipe[(String, String, String)] = grobidSource - .read + + val pipe0 : cascading.pipe.Pipe = grobidSource.read + val grobidPipe : TypedPipe[(String, String, String)] = pipe0 .fromBytesWritable(new Fields("key", "tei_json")) // .debug // Should be 4 tuples for mocked data .toTypedPipe[(String, String)]('key, 'tei_json) @@ -78,7 +79,6 @@ class HBaseCrossrefScoreJob(args: Args) extends JobBase(args) with HBasePipeConv HBaseCrossrefScore.computeOutput(sha1, grobidJson, crossrefJson)} // Output: score, sha1, doi, grobid title, crossref title .write(TypedTsv[(Int, String, String, String, String)](args("output"))) - } object HBaseCrossrefScore { |