diff options
Diffstat (limited to 'scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala')
-rw-r--r-- | scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala index 19b257f..3146a6c 100644 --- a/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala +++ b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala @@ -40,11 +40,11 @@ class GrobidScorableDumpJob(args: Args) extends JobBase(args) { parsedGrobidRows.inc GrobidScorable.jsonToMapFeatures(entry._1, entry._2) } - .filter { entry => Scorable.isValidSlug(entry.slug) } - .map { entry => + .filterNot { entry => entry.isEmpty } + .map { entry => { validGrobidRows.inc - entry - } + entry.get + }} .groupBy { case MapFeatures(slug, json) => slug } .map { tuple => val (slug : String, features : MapFeatures) = tuple |