SI-8988 Escaping character in StringLike.split(c) prevents usage of optimized String.split code path

Escaping a char when calling split is slow.  We end up compiling a
Pattern to simply match a character literal.

Instead, we just use an loop with indexOf to construct our resulting
Array.

Current speed up over old behaviour is about 12-1

1 files changed, 37 insertions, 0 deletions

diff --git a/test/junit/scala/collection/immutable/StringLikeTest.scala b/test/junit/scala/collection/immutable/StringLikeTest.scala
new file mode 100644
index 0000000000..3722bdfe4d
--- /dev/null
+++ b/test/junit/scala/collection/immutable/StringLikeTest.scala
@@ -0,0 +1,37 @@
+package scala.collection.immutable
+
+import org.junit.Test
+import org.junit.runner.RunWith
+import org.junit.runners.JUnit4
+
+import scala.tools.testing.AssertUtil
+import scala.util.Random
+
+/* Test for SI-8988 */
+@RunWith(classOf[JUnit4])
+class StringLikeTest {
+  @Test
+  def testStringSplitWithChar: Unit = {
+    val chars = (0 to 255).map(_.toChar)
+    def randString = Random.nextString(30)
+
+    for (c <- chars) {
+      val s = randString
+      val jString = new java.lang.String(s)
+
+      // make sure we can match a literal character done by Java's split
+      val jSplit = jString.split("\\Q" + c.toString + "\\E")
+      val sSplit = s.split(c)
+      AssertUtil.assertSameElements(jSplit, sSplit, s"Not same result as Java split for char $c in string $s")
+    }
+  }
+
+  @Test
+  def testSplitEdgeCases: Unit = {
+    AssertUtil.assertSameElements("abcd".split('d'), Array("abc")) // not Array("abc", "")
+    AssertUtil.assertSameElements("abccc".split('c'), Array("ab")) // not Array("ab", "", "", "")
+    AssertUtil.assertSameElements("xxx".split('x'), Array[String]()) // not Array("", "", "", "")
+    AssertUtil.assertSameElements("".split('x'), Array("")) // not Array()
+    AssertUtil.assertSameElements("--ch--omp--".split("-"), Array("", "", "ch", "", "omp")) // All the cases!
+  }
+}