aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorEllen Spertus <ellen.spertus@gmail.com>2018-08-07 09:56:19 -0700
committerEllen Spertus <ellen.spertus@gmail.com>2018-08-07 09:56:19 -0700
commit7eed53615e3a106d1cbf7cc451b74674fd2c3daa (patch)
treeaae47e30d973317e3ea68c37a10089e3af2e22cd
parent713b8316d9170ec595f71d4f27df8d3184350921 (diff)
downloadsandcrawler-7eed53615e3a106d1cbf7cc451b74674fd2c3daa.tar.gz
sandcrawler-7eed53615e3a106d1cbf7cc451b74674fd2c3daa.zip
Added StringUtilitiesTest.scala, which passes.
-rw-r--r--scalding/src/test/scala/sandcrawler/StringUtilitiesTest.scala75
1 files changed, 75 insertions, 0 deletions
diff --git a/scalding/src/test/scala/sandcrawler/StringUtilitiesTest.scala b/scalding/src/test/scala/sandcrawler/StringUtilitiesTest.scala
new file mode 100644
index 0000000..2df5a22
--- /dev/null
+++ b/scalding/src/test/scala/sandcrawler/StringUtilitiesTest.scala
@@ -0,0 +1,75 @@
+package sandcrawler
+
+import org.scalatest._
+
+class StringUtilitiesTest extends FlatSpec with Matchers {
+ "removeAccents()" should "handle the empty string" in {
+ StringUtilities.removeAccents("") shouldBe ""
+ }
+
+ it should "not change a string with unaccented characters" in {
+ StringUtilities.removeAccents("abc123") shouldBe "abc123"
+ }
+
+ it should "remove accents from Ls" in {
+ StringUtilities.removeAccents("E\u0141\u0142en") shouldBe "ELlen"
+ }
+
+ it should "remove accents from Es without changing case" in {
+ val result = StringUtilities.removeAccents("\u00e9")
+ result should have length 1
+ result shouldBe "e"
+ }
+
+ it should "convert the ø in Soren" in {
+ StringUtilities.removeAccents("Søren") shouldBe "Soren"
+ StringUtilities.removeAccents("SØREN") shouldBe "SOREN"
+ }
+
+ // Tests adapted from https://oldfashionedsoftware.com/2009/11/19/string-distance-and-refactoring-in-scala/
+ "stringDistance" should "work on empty strings" in {
+ StringUtilities.stringDistance("", "") shouldBe 0
+ StringUtilities.stringDistance("a", "") shouldBe 1
+ StringUtilities.stringDistance("", "a") shouldBe 1
+ StringUtilities.stringDistance("abc", "") shouldBe 3
+ StringUtilities.stringDistance("", "abc") shouldBe 3
+ }
+
+ it should "work on equal strings" in {
+ StringUtilities.stringDistance("", "") shouldBe 0
+ StringUtilities.stringDistance("a", "a") shouldBe 0
+ StringUtilities.stringDistance("abc", "abc") shouldBe 0
+ }
+
+ it should "work where only inserts are needed" in {
+ StringUtilities.stringDistance("", "a") shouldBe 1
+ StringUtilities.stringDistance("a", "ab") shouldBe 1
+ StringUtilities.stringDistance("b", "ab") shouldBe 1
+ StringUtilities.stringDistance("ac", "abc") shouldBe 1
+ StringUtilities.stringDistance("abcdefg", "xabxcdxxefxgx") shouldBe 6
+ }
+
+ it should "work where only deletes are needed" in {
+ StringUtilities.stringDistance( "a", "") shouldBe 1
+ StringUtilities.stringDistance( "ab", "a") shouldBe 1
+ StringUtilities.stringDistance( "ab", "b") shouldBe 1
+ StringUtilities.stringDistance("abc", "ac") shouldBe 1
+ StringUtilities.stringDistance("xabxcdxxefxgx", "abcdefg") shouldBe 6
+ }
+
+ it should "work where only substitutions are needed" in {
+ StringUtilities.stringDistance( "a", "b") shouldBe 1
+ StringUtilities.stringDistance( "ab", "ac") shouldBe 1
+ StringUtilities.stringDistance( "ac", "bc") shouldBe 1
+ StringUtilities.stringDistance("abc", "axc") shouldBe 1
+ StringUtilities.stringDistance("xabxcdxxefxgx", "1ab2cd34ef5g6") shouldBe 6
+ }
+
+ it should "work where many operations are needed" in {
+ StringUtilities.stringDistance("example", "samples") shouldBe 3
+ StringUtilities.stringDistance("sturgeon", "urgently") shouldBe 6
+ StringUtilities.stringDistance("levenshtein", "frankenstein") shouldBe 6
+ StringUtilities.stringDistance("distance", "difference") shouldBe 5
+ StringUtilities.stringDistance("java was neat", "scala is great") shouldBe 7
+ }
+}