aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala/sandcrawler/GrobidScorable.scala
diff options
context:
space:
mode:
Diffstat (limited to 'scalding/src/main/scala/sandcrawler/GrobidScorable.scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/GrobidScorable.scala7
1 files changed, 4 insertions, 3 deletions
diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala
index ba15f22..61055f2 100644
--- a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala
+++ b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala
@@ -16,10 +16,11 @@ class GrobidScorable extends Scorable with HBasePipeConversions {
GrobidScorable.getHBaseSource(args("hbase-table"), args("zookeeper-hosts"))
}
- def getFeaturesPipe(pipe : cascading.pipe.Pipe) : TypedPipe[MapFeatures] = {
- pipe
+ def getFeaturesPipe(args : Args)(implicit mode : Mode, flowDef : FlowDef) : TypedPipe[MapFeatures] = {
+ getSource(args)
+ .read
.fromBytesWritable(new Fields("key", "tei_json"))
- .toTypedPipe[(String, String)](new Fields('key, 'tei_json))
+ .toTypedPipe[(String, String)](new Fields("key", "tei_json"))
.map { entry =>
val (key : String, json : String) = (entry._1, entry._2)
GrobidScorable.grobidToSlug(json) match {