diff options
author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-08-13 10:27:48 -0700 |
---|---|---|
committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-08-13 10:27:48 -0700 |
commit | b4f1acce5eccbb56291f82906d9c01534c7f1506 (patch) | |
tree | 96ff33ed95a4eb9304280b1d5f1ccb269c0d0424 /scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala | |
parent | 1c6e1234974d8b6e4480a13ff5c4ff861c6d1deb (diff) | |
download | sandcrawler-b4f1acce5eccbb56291f82906d9c01534c7f1506.tar.gz sandcrawler-b4f1acce5eccbb56291f82906d9c01534c7f1506.zip |
Factored out ScorableFeatures.
Diffstat (limited to 'scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala')
-rw-r--r-- | scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala | 37 |
1 files changed, 37 insertions, 0 deletions
diff --git a/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala b/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala new file mode 100644 index 0000000..7ec0c4d --- /dev/null +++ b/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala @@ -0,0 +1,37 @@ +package sandcrawler + +import org.scalatest._ + +class ScorableFeaturesTest extends FlatSpec with Matchers { + private def titleToSlug(s : String) : String = { + new ScorableFeatures(title = s).toSlug + } + + "mapToSlug()" should "extract the parts of titles before a colon" in { + titleToSlug("HELLO:there") shouldBe "hello" + } + + it should "extract an entire colon-less string" in { + titleToSlug("hello THERE") shouldBe "hellothere" + } + + it should "return Scorable.NoSlug if given empty string" in { + titleToSlug("") shouldBe Scorable.NoSlug + } + + it should "return Scorable.NoSlug if given null" in { + titleToSlug(null) shouldBe Scorable.NoSlug + } + + it should "strip punctuation" in { + titleToSlug("HELLO!:the:re") shouldBe "hello" + titleToSlug("a:b:c") shouldBe "a" + titleToSlug( + "If you're happy and you know it, clap your hands!") shouldBe "ifyourehappyandyouknowitclapyourhands" + } + + it should "remove whitespace" in { + titleToSlug("foo bar : baz ::") shouldBe "foobar" + titleToSlug("\na\t:b:c") shouldBe "a" + } +} |