diff options
| author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-08-28 08:45:04 -0700 |
|---|---|---|
| committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-09-04 14:59:29 -0700 |
| commit | 2c9172a3797dd218b5648c5390bf2dfb39d8e3a3 (patch) | |
| tree | 6bace5b3b3327b864bab04bba489f1dddd3c5365 /scalding/src/main | |
| parent | 3eb3e38daf4ad58b6da88d7abda222018e4a1ab5 (diff) | |
| download | sandcrawler-2c9172a3797dd218b5648c5390bf2dfb39d8e3a3.tar.gz sandcrawler-2c9172a3797dd218b5648c5390bf2dfb39d8e3a3.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 |