aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
diff options
context:
space:
mode:
Diffstat (limited to 'scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala')
-rw-r--r--scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala57
1 files changed, 54 insertions, 3 deletions
diff --git a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
index 1789d1a..f598cae 100644
--- a/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
+++ b/scalding/src/test/scala/sandcrawler/CrossrefScorableTest.scala
@@ -64,12 +64,18 @@ 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 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("}", "")
+ val CrossrefStringWithNoAuthors = CrossrefString.replace("<<TITLE>>", "Some Valid Title").replace("author", "no-author")
+ val CrossrefStringWrongType = CrossrefString.replace("<<TITLE>>", "Some Valid Title").replace("journal-article", "other")
+ val CrossrefStringNoType = CrossrefString.replace("<<TITLE>>", "Some Valid Title").replace("type", "not-type")
// Unit tests
"CrossrefScorable.jsonToMapFeatures()" should "handle invalid JSON" in {
@@ -82,19 +88,64 @@ 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 missing authors" in {
+ val result = CrossrefScorable.jsonToMapFeatures(CrossrefStringWithNoAuthors)
+ 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()
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
}
}
}
+
+ "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
+ }
+
+ it should "handle content types" in {
+ val resultWrong = CrossrefScorable.jsonToMapFeatures(CrossrefStringWrongType)
+ resultWrong.slug shouldBe Scorable.NoSlug
+ val resultMissing = CrossrefScorable.jsonToMapFeatures(CrossrefStringNoType)
+ resultMissing.slug shouldBe Scorable.NoSlug
+ }
}