mark unicode normalization with type

irc/strings.go

@@ -0,0 +1,66 @@
+package irc
+
+import (
+	"code.google.com/p/go.text/unicode/norm"
+	"regexp"
+	"strings"
+)
+
+var (
+	// regexps
+	ChannelNameExpr = regexp.MustCompile(`^[&!#+][\pL\pN]{1,63}$`)
+	NicknameExpr    = regexp.MustCompile("^[\\pL\\pN\\pP\\pS]{1,32}$")
+)
+
+// Names are normalized and canonicalized to remove formatting marks
+// and simplify usage. They are things like hostnames and usermasks.
+type Name string
+
+func NewName(str string) Name {
+	return Name(norm.NFKC.String(str))
+}
+
+func NewNames(strs []string) []Name {
+	names := make([]Name, len(strs))
+	for index, str := range strs {
+		names[index] = NewName(str)
+	}
+	return names
+}
+
+// tests
+
+func (name Name) IsChannel() bool {
+	return ChannelNameExpr.MatchString(name.String())
+}
+
+func (name Name) IsNickname() bool {
+	return NicknameExpr.MatchString(name.String())
+}
+
+// conversions
+
+func (name Name) String() string {
+	return string(name)
+}
+
+func (name Name) ToLower() Name {
+	return Name(strings.ToLower(name.String()))
+}
+
+// It's safe to coerce a Name to Text. Name is a strict subset of Text.
+func (name Name) Text() Text {
+	return Text(name)
+}
+
+// Text is PRIVMSG, NOTICE, or TOPIC data. It's canonicalized UTF8
+// data to simplify but keeps all formatting.
+type Text string
+
+func NewText(str string) Text {
+	return Text(norm.NFC.String(str))
+}
+
+func (text Text) String() string {
+	return string(text)
+}