dbus: implement dbus_parse_address

This parses D-Bus addresses according to spec. It does significantly fewer copies than dbus_parse_address.

Signed-off-by: Ophestra <cat@gensokyo.uk>

3 files changed, 360 insertions, 0 deletions

diff --git a/dbus/address.go b/dbus/address.go
new file mode 100644
index 00000000..bec49e62
--- /dev/null
+++ b/dbus/address.go
@@ -0,0 +1,186 @@
+package dbus
+
+import (
+	"bytes"
+	"encoding/hex"
+	"errors"
+	"fmt"
+	"slices"
+)
+
+type AddrEntry struct {
+	Method string      `json:"method"`
+	Values [][2]string `json:"values"`
+}
+
+// Parse parses D-Bus address according to
+// https://dbus.freedesktop.org/doc/dbus-specification.html#addresses
+func Parse(addr []byte) ([]AddrEntry, error) {
+	// Look for a semicolon
+	address := bytes.Split(bytes.TrimSuffix(addr, []byte{';'}), []byte{';'})
+
+	// Allocate for entries
+	v := make([]AddrEntry, len(address))
+
+	for i, s := range address {
+		var pairs [][]byte
+
+		// Look for the colon :
+		if method, list, ok := bytes.Cut(s, []byte{':'}); !ok {
+			return v, &BadAddressError{ErrNoColon, i, s, -1, nil}
+		} else {
+			pairs = bytes.Split(list, []byte{','})
+			v[i].Method = string(method)
+			v[i].Values = make([][2]string, len(pairs))
+		}
+
+		for j, pair := range pairs {
+			key, value, ok := bytes.Cut(pair, []byte{'='})
+			if !ok {
+				return v, &BadAddressError{ErrBadPairSep, i, s, j, pair}
+			}
+			if len(key) == 0 {
+				return v, &BadAddressError{ErrBadPairKey, i, s, j, pair}
+			}
+			if len(value) == 0 {
+				return v, &BadAddressError{ErrBadPairVal, i, s, j, pair}
+			}
+			v[i].Values[j][0] = string(key)
+
+			if val, errno := unescapeValue(value); errno != errSuccess {
+				return v, &BadAddressError{errno, i, s, j, pair}
+			} else {
+				v[i].Values[j][1] = string(val)
+			}
+		}
+	}
+
+	return v, nil
+}
+
+func unescapeValue(v []byte) (val []byte, errno ParseError) {
+	if l := len(v) - (bytes.Count(v, []byte{'%'}) * 2); l < 0 {
+		errno = ErrBadValLength
+		return
+	} else {
+		val = make([]byte, l)
+	}
+
+	var i, skip int
+	for iu, b := range v {
+		if skip > 0 {
+			skip--
+			continue
+		}
+
+		if ib := bytes.IndexByte([]byte("-_/.\\*"), b); ib != -1 { // - // _/.\*
+			goto opt
+		} else if b >= '0' && b <= '9' { // 0-9
+			goto opt
+		} else if b >= 'A' && b <= 'Z' { // A-Z
+			goto opt
+		} else if b >= 'a' && b <= 'z' { // a-z
+			goto opt
+		}
+
+		if b != '%' {
+			errno = ErrBadValByte
+			break
+		}
+
+		skip += 2
+		if iu+2 >= len(v) {
+			errno = ErrBadValHexLength
+			break
+		}
+		if c, err := hex.Decode(val[i:i+1], v[iu+1:iu+3]); err != nil {
+			if errors.As(err, new(hex.InvalidByteError)) {
+				errno = ErrBadValHexByte
+				break
+			}
+			// unreachable
+			panic(err.Error())
+		} else if c != 1 {
+			// unreachable
+			panic(fmt.Sprintf("invalid decode length %d", c))
+		}
+		i++
+		continue
+
+	opt:
+		val[i] = b
+		i++
+	}
+
+	return
+}
+
+type ParseError uint8
+
+func (e ParseError) Error() string {
+	switch e {
+	case errSuccess:
+		panic("attempted to return success as error")
+	case ErrNoColon:
+		return "address does not contain a colon"
+	case ErrBadPairSep:
+		return "'=' character not found"
+	case ErrBadPairKey:
+		return "'=' character has no key preceding it"
+	case ErrBadPairVal:
+		return "'=' character has no value following it"
+	case ErrBadValLength:
+		return "unescaped value has impossible length"
+	case ErrBadValByte:
+		return "in D-Bus address, characters other than [-0-9A-Za-z_/.\\*] should have been escaped"
+	case ErrBadValHexLength:
+		return "in D-Bus address, percent character was not followed by two hex digits"
+	case ErrBadValHexByte:
+		return "in D-Bus address, percent character was followed by characters other than hex digits"
+
+	default:
+		return fmt.Sprintf("parse error %d", e)
+	}
+}
+
+const (
+	errSuccess ParseError = iota
+	ErrNoColon
+	ErrBadPairSep
+	ErrBadPairKey
+	ErrBadPairVal
+	ErrBadValLength
+	ErrBadValByte
+	ErrBadValHexLength
+	ErrBadValHexByte
+)
+
+type BadAddressError struct {
+	// error type
+	Type ParseError
+
+	// bad entry position
+	EntryPos int
+	// bad entry value
+	EntryVal []byte
+
+	// bad pair position
+	PairPos int
+	// bad pair value
+	PairVal []byte
+}
+
+func (a *BadAddressError) Is(err error) bool {
+	var b *BadAddressError
+	return errors.As(err, &b) && a.Type == b.Type &&
+		a.EntryPos == b.EntryPos && slices.Equal(a.EntryVal, b.EntryVal) &&
+		a.PairPos == b.PairPos && slices.Equal(a.PairVal, b.PairVal)
+}
+
+func (a *BadAddressError) Error() string {
+	return a.Type.Error()
+}
+
+func (a *BadAddressError) Unwrap() error {
+	return a.Type
+}
diff --git a/dbus/address_escape_test.go b/dbus/address_escape_test.go
new file mode 100644
index 00000000..3ea2cd26
--- /dev/null
+++ b/dbus/address_escape_test.go
@@ -0,0 +1,55 @@
+package dbus
+
+import (
+	"testing"
+)
+
+func TestUnescapeValue(t *testing.T) {
+	testCases := []struct {
+		value   string
+		want    string
+		wantErr ParseError
+	}{
+		// upstream test cases
+		{value: "abcde", want: "abcde"},
+		{value: "", want: ""},
+		{value: "%20%20", want: "  "},
+		{value: "%24", want: "$"},
+		{value: "%25", want: "%"},
+		{value: "abc%24", want: "abc$"},
+		{value: "%24abc", want: "$abc"},
+		{value: "abc%24abc", want: "abc$abc"},
+		{value: "/", want: "/"},
+		{value: "-", want: "-"},
+		{value: "_", want: "_"},
+		{value: "A", want: "A"},
+		{value: "I", want: "I"},
+		{value: "Z", want: "Z"},
+		{value: "a", want: "a"},
+		{value: "i", want: "i"},
+		{value: "z", want: "z"},
+		/* Bug: https://bugs.freedesktop.org/show_bug.cgi?id=53499 */
+		{value: "%c3%b6", want: "\xc3\xb6"},
+
+		{value: "%a", wantErr: ErrBadValHexLength},
+		{value: "%q", wantErr: ErrBadValHexLength},
+		{value: "%az", wantErr: ErrBadValHexByte},
+		{value: "%%", wantErr: ErrBadValLength},
+		{value: "%$$", wantErr: ErrBadValHexByte},
+		{value: "abc%a", wantErr: ErrBadValHexLength},
+		{value: "%axyz", wantErr: ErrBadValHexByte},
+		{value: "%", wantErr: ErrBadValLength},
+		{value: "$", wantErr: ErrBadValByte},
+		{value: " ", wantErr: ErrBadValByte},
+	}
+
+	for _, tc := range testCases {
+		t.Run("unescape "+tc.value, func(t *testing.T) {
+			if got, errno := unescapeValue([]byte(tc.value)); errno != tc.wantErr {
+				t.Errorf("unescapeValue() errno = %v, wantErr %v", errno, tc.wantErr)
+			} else if tc.wantErr == errSuccess && string(got) != tc.want {
+				t.Errorf("unescapeValue() = %q, want %q", got, tc.want)
+			}
+		})
+	}
+}
diff --git a/dbus/address_test.go b/dbus/address_test.go
new file mode 100644
index 00000000..a5d9ed91
--- /dev/null
+++ b/dbus/address_test.go
@@ -0,0 +1,119 @@
+package dbus_test
+
+import (
+	"errors"
+	"reflect"
+	"testing"
+
+	"git.gensokyo.uk/security/fortify/dbus"
+)
+
+func TestParse(t *testing.T) {
+	testCases := []struct {
+		name    string
+		addr    string
+		want    []dbus.AddrEntry
+		wantErr error
+	}{
+		{
+			name: "simple session unix",
+			addr: "unix:path=/run/user/1971/bus",
+			want: []dbus.AddrEntry{{
+				Method: "unix",
+				Values: [][2]string{{"path", "/run/user/1971/bus"}},
+			}},
+		},
+		{
+			name: "simple upper escape",
+			addr: "debug:name=Test,cat=cute,escaped=%c3%b6",
+			want: []dbus.AddrEntry{{
+				Method: "debug",
+				Values: [][2]string{
+					{"name", "Test"},
+					{"cat", "cute"},
+					{"escaped", "\xc3\xb6"},
+				},
+			}},
+		},
+		{
+			name: "simple bad escape",
+			addr: "debug:name=%",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrBadValLength,
+				EntryPos: 0, EntryVal: []byte("debug:name=%"), PairPos: 0, PairVal: []byte("name=%")},
+		},
+
+		// upstream test cases
+		{
+			name: "full address success",
+			addr: "unix:path=/tmp/foo;debug:name=test,sliff=sloff;",
+			want: []dbus.AddrEntry{
+				{Method: "unix", Values: [][2]string{{"path", "/tmp/foo"}}},
+				{Method: "debug", Values: [][2]string{{"name", "test"}, {"sliff", "sloff"}}},
+			},
+		},
+		{
+			name: "empty address",
+			addr: "",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrNoColon,
+				EntryVal: []byte{}, PairPos: -1},
+		},
+		{
+			name: "no body",
+			addr: "foo",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrNoColon,
+				EntryPos: 0, EntryVal: []byte("foo"), PairPos: -1},
+		},
+		{
+			name: "no pair separator",
+			addr: "foo:bar",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrBadPairSep,
+				EntryPos: 0, EntryVal: []byte("foo:bar"), PairPos: 0, PairVal: []byte("bar")},
+		},
+		{
+			name: "no pair separator multi pair",
+			addr: "foo:bar,baz",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrBadPairSep,
+				EntryPos: 0, EntryVal: []byte("foo:bar,baz"), PairPos: 0, PairVal: []byte("bar")},
+		},
+		{
+			name: "no pair separator single valid pair",
+			addr: "foo:bar=foo,baz",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrBadPairSep,
+				EntryPos: 0, EntryVal: []byte("foo:bar=foo,baz"), PairPos: 1, PairVal: []byte("baz")},
+		},
+		{
+			name: "no body single valid address",
+			addr: "foo:bar=foo;baz",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrNoColon,
+				EntryPos: 1, EntryVal: []byte("baz"), PairPos: -1},
+		},
+		{
+			name: "no key",
+			addr: "foo:=foo",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrBadPairKey,
+				EntryPos: 0, EntryVal: []byte("foo:=foo"), PairPos: 0, PairVal: []byte("=foo")},
+		},
+		{
+			name: "no value",
+			addr: "foo:foo=",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrBadPairVal,
+				EntryPos: 0, EntryVal: []byte("foo:foo="), PairPos: 0, PairVal: []byte("foo=")},
+		},
+		{
+			name: "no pair separator single valid pair trailing",
+			addr: "foo:foo,bar=baz",
+			wantErr: &dbus.BadAddressError{Type: dbus.ErrBadPairSep,
+				EntryPos: 0, EntryVal: []byte("foo:foo,bar=baz"), PairPos: 0, PairVal: []byte("foo")},
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got, err := dbus.Parse([]byte(tc.addr)); !errors.Is(err, tc.wantErr) {
+				t.Errorf("Parse() error = %v, wantErr %v", err, tc.wantErr)
+			} else if tc.wantErr == nil && !reflect.DeepEqual(got, tc.want) {
+				t.Errorf("Parse() = %#v, want %#v", got, tc.want)
+			}
+		})
+	}
+}