prevent panic when using 2-byte unicode bmPrefix against 3-byte input values. fixes #5

Author: dlclark

regexp_test.go

@@ -736,6 +736,17 @@ func TestECMANegateRange(t *testing.T) {
 	}
 }
 
+func TestThreeByteUnicode_InputOnly(t *testing.T) {
+	// confirm the bmprefix properly ignores 3-byte unicode in the input value
+	// this used to panic
+	re := MustCompile("高", 0)
+	if m, err := re.MatchString("📍Test高"); 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

@@ -666,9 +666,9 @@ func (b *BmPrefix) Scan(text []rune, index, beglimit, endlimit int) int {
 		if chTest != chMatch {
 			if chTest < 128 {
 				advance = 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 {
 					advance = unicodeLookup[chTest&0xFF]
 				} else {
 					advance = defadv