author parsing (and year, for crossref)

2 files changed, 6 insertions, 1 deletions

diff --git a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
index 3d18a21..ac7cc70 100644
--- a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
+++ b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
@@ -64,7 +64,7 @@ class CrossrefScorableTest extends FlatSpec with Matchers {
   "issn-type" : [ { "value" : "0987-7983", "type" : "print" } ],
   "subject" : [ "Pediatrics, Perinatology, and Child Health" ]
 }
-"""
+""".replace("<<DOI>>", "10.123/aBc")
   // scalastyle:on
   val CrossrefStringWithGoodTitle = CrossrefString.replace("<<TITLE>>", "Some Title")
   val CrossrefStringWithMaximumTitle = CrossrefString.replace("<<TITLE>>", "T" * Scorable.MaxTitleLength)
@@ -102,6 +102,10 @@ class CrossrefScorableTest extends FlatSpec with Matchers {
       case None => fail()
       case Some(map) => {
         map("title").asInstanceOf[String] shouldBe "Some Title"
+        map("doi").asInstanceOf[String] shouldBe "10.123/abc"
+        // TODO: full name? not just a string?
+        map("authors").asInstanceOf[List[String]] shouldBe List("Gaier")
+        map("year").asInstanceOf[Double].toInt shouldBe 2002
       }
     }
   }
diff --git a/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala b/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala
index 6c45cc5..119cf90 100644
--- a/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala
+++ b/scalding/src/test/scala/sandcrawler/GrobidScorableTest.scala
@@ -90,6 +90,7 @@ class GrobidScorableTest extends FlatSpec with Matchers {
       case Some(map) => {
         map should contain key "title"
         map("title").asInstanceOf[String] shouldBe "Dummy Example File"
+        map("authors").asInstanceOf[List[String]] shouldBe List("Brewster Kahle", "J Doe")
       }
     }
   }