From 9fb9a35b15ed9b553ad4f938dc4e636e5d91ac33 Mon Sep 17 00:00:00 2001
From: Ellen Spertus <ellen.spertus@gmail.com>
Date: Wed, 22 Aug 2018 13:17:49 -0700
Subject: Added title-length filtering to CrossrefScorable.

---
 .../scala/sandcrawler/CrossrefScorableTest.scala   | 36 ++++++++++++++++++++--
 1 file changed, 34 insertions(+), 2 deletions(-)

(limited to 'scalding/src/test/scala/sandcrawler')

diff --git a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
index 1789d1a..3d18a21 100644
--- a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
+++ b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
@@ -66,7 +66,10 @@ class CrossrefScorableTest extends FlatSpec with Matchers {
 }
 """
   // scalastyle:on
-  val CrossrefStringWithTitle = CrossrefString.replace("<<TITLE>>", "Some Title")
+  val CrossrefStringWithGoodTitle = CrossrefString.replace("<<TITLE>>", "Some Title")
+  val CrossrefStringWithMaximumTitle = CrossrefString.replace("<<TITLE>>", "T" * Scorable.MaxTitleLength)
+  val CrossrefStringWithExcessiveTitle = CrossrefString.replace("<<TITLE>>", "T" * Scorable.MaxTitleLength + "0")
+  val CrossrefStringWithNullTitle = CrossrefString.replace("\"<<TITLE>>\"", "null")
   val CrossrefStringWithEmptyTitle = CrossrefString.replace("<<TITLE>>", "")
   val CrossrefStringWithoutTitle = CrossrefString.replace("title", "nottitle")
   val MalformedCrossrefString = CrossrefString.replace("}", "")
@@ -82,13 +85,18 @@ class CrossrefScorableTest extends FlatSpec with Matchers {
     result.slug shouldBe Scorable.NoSlug
   }
 
+  it should "handle null title" in {
+    val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithNullTitle)
+    result.slug shouldBe Scorable.NoSlug
+  }
+
   it should "handle empty title" in {
     val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithEmptyTitle)
     result.slug shouldBe Scorable.NoSlug
   }
 
   it should "handle valid input" in {
-    val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithTitle)
+    val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithGoodTitle)
     result.slug shouldBe "sometitle"
     Scorable.jsonToMap(result.json) match {
       case None => fail()
@@ -97,4 +105,28 @@ class CrossrefScorableTest extends FlatSpec with Matchers {
       }
     }
   }
+
+  "CrossrefScorable.keepRecord()" should "return true for valid JSON with title" in {
+    CrossrefScorable.keepRecord(CrossrefStringWithGoodTitle) shouldBe true
+  }
+
+  it should "return true for valid JSON with a title of maximum permitted length" in {
+    CrossrefScorable.keepRecord(CrossrefStringWithMaximumTitle) shouldBe true
+  }
+
+  it should "return false for valid JSON with excessively long title" in {
+    CrossrefScorable.keepRecord(CrossrefStringWithExcessiveTitle) shouldBe false
+  }
+
+  it should "return false for valid JSON with null title" in {
+    CrossrefScorable.keepRecord(CrossrefStringWithNullTitle) shouldBe false
+  }
+
+  it should "return false for valid JSON with no title" in {
+    CrossrefScorable.keepRecord(CrossrefStringWithoutTitle) shouldBe false
+  }
+
+  it should "return false for invalid JSON" in {
+    CrossrefScorable.keepRecord(CrossrefStringWithoutTitle) shouldBe false
+  }
 }
-- 
cgit v1.2.3