From 31354b1a6062c5c56a30610f68fa48c82a7e83f0 Mon Sep 17 00:00:00 2001 From: Ellen Spertus Date: Sun, 12 Aug 2018 18:08:51 -0700 Subject: Tests pass. --- .../src/test/scala/sandcrawler/ScorableTest.scala | 28 ++++++++++++++-------- 1 file changed, 18 insertions(+), 10 deletions(-) (limited to 'scalding/src/test/scala/sandcrawler/ScorableTest.scala') diff --git a/scalding/src/test/scala/sandcrawler/ScorableTest.scala b/scalding/src/test/scala/sandcrawler/ScorableTest.scala index 2f80492..95faacc 100644 --- a/scalding/src/test/scala/sandcrawler/ScorableTest.scala +++ b/scalding/src/test/scala/sandcrawler/ScorableTest.scala @@ -54,28 +54,36 @@ class ScorableTest extends FlatSpec with Matchers { "annex": null } """ + private def titleToSlug(s : String) : String = { + Scorable.mapToSlug(Scorable.toScorableMap(title = s)) + } - "titleToSlug()" should "extract the parts of titles before a colon" in { - Scorable.titleToSlug("HELLO:there") shouldBe "hello" + "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 { - Scorable.titleToSlug("hello THERE") shouldBe "hello there" + titleToSlug("hello THERE") shouldBe "hellothere" } it should "return Scorable.NoSlug if given empty string" in { - Scorable.titleToSlug("") shouldBe Scorable.NoSlug + titleToSlug("") shouldBe Scorable.NoSlug } it should "return Scorable.NoSlug if given null" in { - Scorable.titleToSlug(null) shouldBe Scorable.NoSlug + 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" } - "titleToSlug()" should "strip punctuation" in { - Scorable.titleToSlug("HELLO!:the:re") shouldBe "hello" - Scorable.titleToSlug("a:b:c") shouldBe "a" - Scorable.titleToSlug( - "If you're happy and you know it, clap your hands!") shouldBe "if youre happy and you know it clap your hands" + it should "remove whitespace" in { + titleToSlug("foo bar : baz ::") shouldBe "foobar" + titleToSlug("\na\t:b:c") shouldBe "a" } "jsonToMap()" should "return a map, given a legal JSON string" in { -- cgit v1.2.3