fix #1003

irc/utils/glob.go

@@ -11,21 +11,53 @@ import (
 
 // yet another glob implementation in Go
 
-func CompileGlob(glob string) (result *regexp.Regexp, err error) {
-	var buf bytes.Buffer
-	buf.WriteByte('^')
+func addRegexp(buf *bytes.Buffer, glob string, submatch bool) (err error) {
 	for _, r := range glob {
 		switch r {
 		case '*':
-			buf.WriteString("(.*)")
+			if submatch {
+				buf.WriteString("(.*)")
+			} else {
+				buf.WriteString(".*")
+			}
 		case '?':
-			buf.WriteString("(.)")
+			if submatch {
+				buf.WriteString("(.)")
+			} else {
+				buf.WriteString(".")
+			}
 		case 0xFFFD:
-			return nil, &syntax.Error{Code: syntax.ErrInvalidUTF8, Expr: glob}
+			return &syntax.Error{Code: syntax.ErrInvalidUTF8, Expr: glob}
 		default:
 			buf.WriteString(regexp.QuoteMeta(string(r)))
 		}
 	}
+	return
+}
+
+func CompileGlob(glob string, submatch bool) (result *regexp.Regexp, err error) {
+	var buf bytes.Buffer
+	buf.WriteByte('^')
+	err = addRegexp(&buf, glob, submatch)
+	if err != nil {
+		return
+	}
 	buf.WriteByte('$')
 	return regexp.Compile(buf.String())
 }
+
+func CompileMasks(masks []string) (result *regexp.Regexp, err error) {
+	var buf bytes.Buffer
+	buf.WriteString("^(")
+	for i, mask := range masks {
+		err = addRegexp(&buf, mask, false)
+		if err != nil {
+			return
+		}
+		if i != len(masks)-1 {
+			buf.WriteByte('|')
+		}
+	}
+	buf.WriteString(")$")
+	return regexp.Compile(buf.String())
+}

irc/utils/glob_test.go

@@ -9,7 +9,7 @@ import (
 )
 
 func globMustCompile(glob string) *regexp.Regexp {
-	re, err := CompileGlob(glob)
+	re, err := CompileGlob(glob, false)
 	if err != nil {
 		panic(err)
 	}
@@ -46,3 +46,123 @@ func TestGlob(t *testing.T) {
 	assertMatches("S*e", "Skåne", true, t)
 	assertMatches("Sk?ne", "Skåne", true, t)
 }
+
+func BenchmarkGlob(b *testing.B) {
+	g := globMustCompile("https://*google.com")
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		g.MatchString("https://www.google.com")
+	}
+}
+
+func BenchmarkGlobCompilation(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		CompileGlob("https://*google.com", false)
+	}
+}
+
+// these are actual bans from my production network :-/
+var bans = []string{
+	"*!*@tor-network.onion",
+	"`!*@*",
+	"qanon!*@*",
+	"*!bibi@tor-network.onion",
+	"shivarm!*@*",
+	"8====d!*@*",
+	"shiviram!*@*",
+	"poop*!*@*",
+	"shivoram!*@*",
+	"shivvy!*@*",
+	"shavirim!*@*",
+	"shivarm_!*@*",
+	"_!*@*",
+}
+
+func TestMasks(t *testing.T) {
+	matcher, err := CompileMasks(bans)
+	if err != nil {
+		panic(err)
+	}
+
+	if !matcher.MatchString("evan!user@tor-network.onion") {
+		t.Errorf("match expected")
+	}
+	if !matcher.MatchString("`!evan@b9un4fv3he44q.example.com") {
+		t.Errorf("match expected")
+	}
+	if matcher.MatchString("horse!horse@t5dwi8vacg47y.example.com") {
+		t.Errorf("match not expected")
+	}
+	if matcher.MatchString("horse_!horse@t5dwi8vacg47y.example.com") {
+		t.Errorf("match not expected")
+	}
+	if matcher.MatchString("shivaram!shivaram@yrqgsrjy2p7my.example.com") {
+		t.Errorf("match not expected")
+	}
+}
+
+func BenchmarkMasksCompile(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		CompileMasks(bans)
+	}
+}
+
+func BenchmarkMasksMatch(b *testing.B) {
+	matcher, _ := CompileMasks(bans)
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		matcher.MatchString("evan!user@tor-network.onion")
+		matcher.MatchString("horse_!horse@t5dwi8vacg47y.example.com")
+		matcher.MatchString("shivaram!shivaram@yrqgsrjy2p7my.example.com")
+	}
+}
+
+// compare performance to compilation of the | clauses as separate regexes
+// first for compilation, then for matching
+
+func compileAll(masks []string) (result []*regexp.Regexp, err error) {
+	a := make([]*regexp.Regexp, 0, len(masks))
+	for _, mask := range masks {
+		m, err := CompileGlob(mask, false)
+		if err != nil {
+			return nil, err
+		}
+		a = append(a, m)
+	}
+	return a, nil
+}
+
+func matchesAny(masks []*regexp.Regexp, str string) bool {
+	for _, r := range masks {
+		if r.MatchString(str) {
+			return true
+		}
+	}
+	return false
+}
+
+func BenchmarkLinearCompile(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		compileAll(bans)
+	}
+}
+
+func BenchmarkLinearMatch(b *testing.B) {
+	a, err := compileAll(bans)
+	if err != nil {
+		panic(err)
+	}
+	if matchesAny(a, "horse_!horse@t5dwi8vacg47y.example.com") {
+		panic("incorrect match")
+	}
+	if !matchesAny(a, "evan!user@tor-network.onion") {
+		panic("incorrect match")
+	}
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		matchesAny(a, "horse_!horse@t5dwi8vacg47y.example.com")
+		matchesAny(a, "evan!user@tor-network.onion")
+		matchesAny(a, "shivaram!shivaram@yrqgsrjy2p7my.example.com")
+	}
+}