diff options
| author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-07-24 12:25:45 -0700 | 
|---|---|---|
| committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-07-24 12:25:45 -0700 | 
| commit | dae965840db388c53b969d76849e5e8e9569ceee (patch) | |
| tree | cf015ebe7d55ae6b9bc1a1272a57fd0d25e86014 /scalding/src/main/scala | |
| parent | 3e33d60aac9db78d0458876fbe987627db222bbb (diff) | |
| download | sandcrawler-dae965840db388c53b969d76849e5e8e9569ceee.tar.gz sandcrawler-dae965840db388c53b969d76849e5e8e9569ceee.zip | |
Changed return type of grobidToSlug() to Option[String].
Diffstat (limited to 'scalding/src/main/scala')
| -rw-r--r-- | scalding/src/main/scala/sandcrawler/HBaseCrossrefScoreJob.scala | 18 | 
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)    }  } | 
