From 2c9172a3797dd218b5648c5390bf2dfb39d8e3a3 Mon Sep 17 00:00:00 2001
From: Ellen Spertus <ellen.spertus@gmail.com>
Date: Tue, 28 Aug 2018 08:45:04 -0700
Subject: restored code I inadvertantly removed when merging
---
scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
(limited to 'scalding/src')
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
--
cgit v1.2.3