diff options
| author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-08-28 08:45:04 -0700 |
|---|---|---|
| committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-08-28 08:45:04 -0700 |
| commit | 98eebb1d9abc954c0fe1735bf5472c57e71dc7a8 (patch) | |
| tree | 344ec1bf43a945b8bd97e97fc5c1d9ef2b4a14df /scalding/src/main | |
| parent | 93c866293ccca23131224b47e868c632ad2de032 (diff) | |
| download | sandcrawler-98eebb1d9abc954c0fe1735bf5472c57e71dc7a8.tar.gz sandcrawler-98eebb1d9abc954c0fe1735bf5472c57e71dc7a8.zip | |
restored code I inadvertantly removed when merging
Diffstat (limited to 'scalding/src/main')
| -rw-r--r-- | scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala index d40410b..3146a6c 100644 --- a/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala +++ b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala @@ -41,7 +41,10 @@ class GrobidScorableDumpJob(args: Args) extends JobBase(args) { GrobidScorable.jsonToMapFeatures(entry._1, entry._2) } .filterNot { entry => entry.isEmpty } - .map { entry => entry.get } + .map { entry => { + validGrobidRows.inc + entry.get + }} .groupBy { case MapFeatures(slug, json) => slug } .map { tuple => val (slug : String, features : MapFeatures) = tuple |