bytes, strings: allow -1 in Map to mean "drop this character".

xml: drop invalid characters in attribute names
    when constructing struct field names.

R=rsc
CC=r
http://codereview.appspot.com/157104

Committer: Russ Cox <rsc@golang.org>

2 files changed, 25 insertions, 9 deletions

diff --git a/src/pkg/bytes/bytes.go b/src/pkg/bytes/bytes.go
index 9ab199ceb..85d4f9fd7 100644
--- a/src/pkg/bytes/bytes.go
+++ b/src/pkg/bytes/bytes.go
@@ -207,7 +207,8 @@ func HasSuffix(s, suffix []byte) bool {
 }
 
 // Map returns a copy of the byte array s with all its characters modified
-// according to the mapping function.
+// according to the mapping function. If mapping returns a negative value, the character is
+// dropped from the string with no replacement.
 func Map(mapping func(rune int) int, s []byte) []byte {
 	// In the worst case, the array can grow when mapped, making
 	// things unpleasant.  But it's so rare we barge in assuming it's
@@ -222,16 +223,18 @@ func Map(mapping func(rune int) int, s []byte) []byte {
 			rune, wid = utf8.DecodeRune(s[i:])
 		}
 		rune = mapping(rune);
-		if nbytes+utf8.RuneLen(rune) > maxbytes {
-			// Grow the buffer.
-			maxbytes = maxbytes*2 + utf8.UTFMax;
-			nb := make([]byte, maxbytes);
-			for i, c := range b[0:nbytes] {
-				nb[i] = c
+		if rune >= 0 {
+			if nbytes+utf8.RuneLen(rune) > maxbytes {
+				// Grow the buffer.
+				maxbytes = maxbytes*2 + utf8.UTFMax;
+				nb := make([]byte, maxbytes);
+				for i, c := range b[0:nbytes] {
+					nb[i] = c
+				}
+				b = nb;
 			}
-			b = nb;
+			nbytes += utf8.EncodeRune(rune, b[nbytes:maxbytes]);
 		}
-		nbytes += utf8.EncodeRune(rune, b[nbytes:maxbytes]);
 		i += wid;
 	}
 	return b[0:nbytes];
diff --git a/src/pkg/bytes/bytes_test.go b/src/pkg/bytes/bytes_test.go
index 553ceb7c5..3f77e6e9f 100644
--- a/src/pkg/bytes/bytes_test.go
+++ b/src/pkg/bytes/bytes_test.go
@@ -365,6 +365,19 @@ func TestMap(t *testing.T) {
 	if string(m) != expect {
 		t.Errorf("rot13: expected %q got %q", expect, m)
 	}
+
+	// 5. Drop
+	dropNotLatin := func(rune int) int {
+		if unicode.Is(unicode.Latin, rune) {
+			return rune
+		}
+		return -1;
+	};
+	m = Map(dropNotLatin, Bytes("Hello, 세계"));
+	expect = "Hello";
+	if string(m) != expect {
+		t.Errorf("drop: expected %q got %q", expect, m)
+	}
 }
 
 func TestToUpper(t *testing.T)	{ runStringTests(t, ToUpper, "ToUpper", upperTests) }