prevent panic with multi-byte unicode bmPrefix that partially match. fixes #6

Author: dlclark

regexp_test.go

@@ -747,6 +747,24 @@ func TestThreeByteUnicode_InputOnly(t *testing.T) {
 	}
 }
 
+func TestMultibyteUnicode_MatchPartialPattern(t *testing.T) {
+	re := MustCompile("猟な", 0)
+	if m, err := re.MatchString("なあ🍺な"); err != nil {
+		t.Fatal(err)
+	} else if m {
+		t.Fatal("Expected no match")
+	}
+}
+
+func TestMultibyteUnicode_Match(t *testing.T) {
+	re := MustCompile("猟な", 0)
+	if m, err := re.MatchString("なあ🍺猟な"); err != nil {
+		t.Fatal(err)
+	} else if !m {
+		t.Fatal("Expected match")
+	}
+}
+
 /*
 func TestPcreStuff(t *testing.T) {
 	re := MustCompile(`(?(?=(a))a)`, Debug)

syntax/prefix.go

@@ -704,9 +704,9 @@ func (b *BmPrefix) Scan(text []rune, index, beglimit, endlimit int) int {
 					advance = b.positive[match]
 					if (chTest & 0xFF80) == 0 {
 						test2 = (match - startmatch) + b.negativeASCII[chTest]
-					} else if nil != b.negativeUnicode {
+					} else if chTest < 0xffff && len(b.negativeUnicode) > 0 {
 						unicodeLookup = b.negativeUnicode[chTest>>8]
-						if unicodeLookup != nil {
+						if len(unicodeLookup) > 0 {
 							test2 = (match - startmatch) + unicodeLookup[chTest&0xFF]
 						} else {
 							test += advance