Merge pull request #15666 from spray/w/15665-fix-BoyerMoore

=hco #15665 fix BoyerMoore for byte values > 0x7f

akka-http-core/src/main/scala/akka/http/parsing/BoyerMoore.scala

@@ -19,7 +19,7 @@ private class BoyerMoore(needle: Array[Byte]) {
     val table = Array.fill(256)(needle.length)
     @tailrec def rec(i: Int): Unit =
       if (i < nl1) {
-        table(needle(i)) = nl1 - i
+        table(needle(i) & 0xff) = nl1 - i
         rec(i + 1)
       }
     rec(0)
@@ -61,7 +61,7 @@ private class BoyerMoore(needle: Array[Byte]) {
       if (needle(j) == byte) {
         if (j == 0) i // found
         else rec(i - 1, j - 1)
-      } else rec(i + math.max(offsetTable(nl1 - j), charTable(byte)), nl1)
+      } else rec(i + math.max(offsetTable(nl1 - j), charTable(byte & 0xff)), nl1)
     }
     rec(offset + nl1, nl1)
   }

akka-http-core/src/test/scala/akka/http/parsing/BoyerMooreSpec.scala

@@ -4,6 +4,8 @@
 
 package akka.http.parsing
 
+import java.util.regex.Pattern
+
 import akka.parboiled2.CharPredicate
 
 import scala.annotation.tailrec
@@ -18,33 +20,42 @@ class BoyerMooreSpec extends WordSpec with Matchers {
   "The Boyer Moore implementation" should {
 
     "correctly find all matches of a few handwritten string-search examples" in {
-      find("foo", "the foe in moofoo is foobar") shouldEqual Seq(14, 21)
-      find("ana", "bananas") shouldEqual Seq(1, 3)
-      find("anna", "bananas") shouldEqual Seq()
+      findString("foo", "the foe in moofoo is foobar") shouldEqual Seq(14, 21)
+      findString("ana", "bananas") shouldEqual Seq(1, 3)
+      findString("anna", "bananas") shouldEqual Seq()
     }
 
     "perform identically to a regex search" in {
       val random = new Random()
       // the alphabet base is a random shuffle of 8 distinct alphanumeric chars
-      val alphabetBase = random.shuffle('0' to 'z' filter CharPredicate.AlphaNum) take 8
+      val alphabetBase: IndexedSeq[Byte] = random.shuffle(0 to 255).take(8).map(_.toByte)
+
       val haystackLen = 1000
       (0 to 9) foreach { run ⇒
-        val alphabet = alphabetBase.take(4 + random.nextInt(5)).mkString // 4 to 8 distinct alphanumeric chars
-        val randomAlphabetChars = Stream.continually(alphabet.charAt(random.nextInt(alphabet.length)))
-        val haystack = randomAlphabetChars.take(haystackLen).mkString
-        val needle = randomAlphabetChars.take(run / 3 + 3).mkString // 3 to 6 random alphabet chars
+        val alphabet = alphabetBase.take(4 + random.nextInt(5)) // 4 to 8 distinct alphanumeric chars
+        val randomAlphabetChars = Stream.continually(alphabet(random.nextInt(alphabet.length)))
+        def randomBytes(num: Int): ByteString = ByteString(randomAlphabetChars.take(num): _*)
+        val haystack = randomBytes(haystackLen)
+        val needle = randomBytes(run / 3 + 3) // 3 to 6 random alphabet chars
 
         val bmFinds = find(needle, haystack, skipFindsThatStartInFinds = true)
         val reFinds = findWithRegex(needle, haystack)
         if (bmFinds != reFinds) {
-          def showFind(ix: Int): String =
-            s"""...${haystack.substring(math.max(ix - 8, 0), math.min(ix + needle.length + 8, haystack.length))}...
-               |${" " * (3 + math.min(ix - 8, 8))}${"^" * needle.length}
+          def showBytes(bs: Seq[Byte]): String = bs.map(b ⇒ (b & 0xff).formatted("%02x")).mkString(" ")
+          def len(num: Int) = num * 2 + math.max(0, num - 1)
+
+          def showFind(ix: Int): String = {
+            val startIdx = math.max(ix - 8, 0)
+            val endIdx = math.min(ix + needle.length + 8, haystack.length)
+
+            s"""...${showBytes(haystack.drop(startIdx).take(endIdx - startIdx))}...
+               |${" " * (3 + math.min(8, ix) * 3)}${"^" * len(needle.length)}
                |""".stripMargin
+          }
           val foundOnlyByBM = bmFinds.filterNot(reFinds.contains).map(showFind).mkString
           val foundOnlyByRE = reFinds.filterNot(bmFinds.contains).map(showFind).mkString
-          fail(s"""alphabet: $alphabet
-                   |needle: $needle
+          fail(s"""alphabet: ${showBytes(alphabet)}
+                   |needle: ${showBytes(needle)}
                    |found only by boyer moore:
                    |$foundOnlyByBM
                    |found only by regex:
@@ -55,17 +66,20 @@ class BoyerMooreSpec extends WordSpec with Matchers {
     }
   }
 
-  def find(needle: String, haystack: String, skipFindsThatStartInFinds: Boolean = false): Seq[Int] = {
-    val boyerMoore = new BoyerMoore(needle.asciiBytes)
+  def findString(needle: String, haystack: String, skipFindsThatStartInFinds: Boolean = false): Seq[Int] =
+    find(ByteString(needle), ByteString(haystack), skipFindsThatStartInFinds)
+
+  def find(needle: ByteString, haystack: ByteString, skipFindsThatStartInFinds: Boolean = false): Seq[Int] = {
+    val boyerMoore = new BoyerMoore(needle.toArray[Byte])
     @tailrec def rec(offset: Int, result: Seq[Int]): Seq[Int] = {
       val ix =
-        try boyerMoore.nextIndex(ByteString(haystack), offset)
+        try boyerMoore.nextIndex(haystack, offset)
         catch { case NotEnoughDataException ⇒ -1 }
       if (ix >= 0) rec(if (skipFindsThatStartInFinds) ix + needle.length else ix + 1, result :+ ix) else result
     }
     rec(0, Seq.empty)
   }
 
-  def findWithRegex(needle: String, haystack: String): Seq[Int] =
-    needle.r.findAllMatchIn(haystack).map(_.start).toSeq
+  def findWithRegex(needle: ByteString, haystack: ByteString): Seq[Int] =
+    Pattern.quote(needle.map(_.toChar).mkString).r.findAllMatchIn(haystack.map(_.toChar).mkString).map(_.start).toSeq
 }