aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala
diff options
context:
space:
mode:
Diffstat (limited to 'scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala18
1 files changed, 11 insertions, 7 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala b/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala
index a22af81..d3e78fe 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala
@@ -22,7 +22,7 @@ class HBaseCrossrefScoreJob(args: Args) extends JobBase(args) with HBasePipeConv
sourceMode = SourceMode.SCAN_ALL)
val grobidPipe = grobidSource
- .read
+ .read
.map('tei_json -> 'slug) {
json : String => HBaseCrossrefScore.grobidToSlug(json)}
@@ -42,17 +42,21 @@ class HBaseCrossrefScoreJob(args: Args) extends JobBase(args) with HBasePipeConv
}
object HBaseCrossrefScore {
- def grobidToSlug(json : String) = {
+ def grobidToSlug(json : String) : Option[String] = {
// https://stackoverflow.com/a/32717262/631051
val jsonObject = JSON.parseFull(json)
- val globalMap = jsonObject.get.asInstanceOf[Map[String, Any]]
- globalMap.get("title") match {
- case Some(title) => titleToSlug(title.asInstanceOf[String])
- case None => ""
+ if (jsonObject == None) {
+ None
+ } else {
+ val globalMap = jsonObject.get.asInstanceOf[Map[String, Any]]
+ globalMap.get("title") match {
+ case Some(title) => Some(titleToSlug(title.asInstanceOf[String]))
+ case None => None
+ }
}
}
- def titleToSlug(title : String) = {
+ def titleToSlug(title : String) : String = {
title.split(":")(0)
}
}