Merge pull request scala#4209 from kanielc/SI-8988

SI-8988 Escaping character in StringLike.split(c) is slow

Author: adriaanm

src/library/scala/collection/immutable/StringLike.scala

@@ -10,7 +10,7 @@ package scala
 package collection
 package immutable
 
-import mutable.Builder
+import mutable.{ ArrayBuilder, Builder }
 import scala.util.matching.Regex
 import scala.math.ScalaNumber
 import scala.reflect.ClassTag
@@ -203,8 +203,33 @@ self =>
 
   private def escape(ch: Char): String = "\\Q" + ch + "\\E"
 
-  @throws(classOf[java.util.regex.PatternSyntaxException])
-  def split(separator: Char): Array[String] = toString.split(escape(separator))
+  def split(separator: Char): Array[String] = {
+    val thisString = toString
+    var pos = thisString.indexOf(separator)
+
+    if (pos != -1) {
+      val res = new ArrayBuilder.ofRef[String]
+
+      var prev = 0
+      do {
+        res += thisString.substring(prev, pos)
+        prev = pos + 1
+        pos = thisString.indexOf(separator, prev)
+      } while (pos != -1)
+
+      if (prev != thisString.size)
+        res += thisString.substring(prev, thisString.size)
+
+      val initialResult = res.result()
+      pos = initialResult.length
+      while (pos > 0 && initialResult(pos - 1).isEmpty) pos = pos - 1
+      if (pos != initialResult.length) {
+        val trimmed = new Array[String](pos)
+        Array.copy(initialResult, 0, trimmed, 0, pos)
+        trimmed
+      } else initialResult
+    } else Array[String](thisString)
+  }
 
   @throws(classOf[java.util.regex.PatternSyntaxException])
   def split(separators: Array[Char]): Array[String] = {

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!
+  }
+}