write output method, make tags order-preserving

message.go

@@ -1,12 +1,37 @@
 package main
 
 import (
+	"bytes"
 	"fmt"
 	"strings"
 )
 
+type Tag struct {
+	Key   string
+	Value string
+}
+
+// Tags is an key-order-preserving, last-insert-wins slice
+type Tags struct {
+	tags []Tag
+}
+
+func (t *Tags) Insert(new Tag) {
+	for i := range t.tags {
+		if t.tags[i].Key == new.Key {
+			t.tags[i] = new
+			return
+		}
+	}
+	t.tags = append(t.tags, new)
+}
+
+func (t *Tags) GetSlice() []Tag {
+	return t.tags
+}
+
 type Message struct {
-	Tags       map[string]string
+	Tags       []Tag
 	Source     string
 	Command    command
 	Parameters []string
@@ -32,8 +57,8 @@ func (m *Message) ParseMessage() error {
 	return nil
 }
 
-func (m *Message) parseTags() (map[string]string, error) {
-	tags := make(map[string]string)
+func (m *Message) parseTags() ([]Tag, error) {
+	tags := &Tags{}
 	for {
 		next := m.findNext("=", ";")
 		if next == -1 {
@@ -42,17 +67,17 @@ func (m *Message) parseTags() (map[string]string, error) {
 			key := m.Raw[m.parseIndex : m.parseIndex+next]
 			m.parseIndex += next + 1
 			eot := m.findNext(";", " ")
-			tags[key] = m.Raw[m.parseIndex : m.parseIndex+eot]
-			m.parseIndex += len(tags[key]) + 1
+			tags.Insert(Tag{Key: key, Value: unescapeTag(m.Raw[m.parseIndex : m.parseIndex+eot])})
+			m.parseIndex += len(m.Raw[m.parseIndex:m.parseIndex+eot]) + 1
 		} else if m.Raw[m.parseIndex+next] == ';' {
 			key := m.Raw[m.parseIndex : m.parseIndex+next]
 			m.parseIndex += next + 1
-			tags[key] = ""
+			tags.Insert(Tag{Key: key, Value: ""})
 		} else {
 			break
 		}
 	}
-	return tags, nil
+	return tags.GetSlice(), nil
 }
 
 func (m *Message) parseSource() (string, error) {
@@ -69,9 +94,10 @@ func (m *Message) parseCommand() (command, error) {
 	start := m.parseIndex
 	endofparse := strings.Index(m.Raw[m.parseIndex:], " ")
 	if endofparse == -1 {
-		return "", fmt.Errorf("end of string encountered while parsing command")
+		m.parseIndex = len(m.Raw) // a command with no parameters
+	} else {
+		m.parseIndex += endofparse
 	}
-	m.parseIndex += endofparse
 	return command(m.Raw[start:m.parseIndex]), nil
 }
 
@@ -131,6 +157,103 @@ func (m *Message) findNext(bs ...string) int {
 	return r
 }
 
-func (m *Message) ToBytes() []byte {
-	return nil
+func unescapeTag(s string) string {
+	sb := strings.Builder{}
+	escaping := false
+	i := 0
+	for i < len(s) {
+		if escaping {
+			if i+1 > len(s) {
+				break
+			}
+			switch s[i] {
+			case ':':
+				sb.WriteByte(';')
+			case '\\':
+				sb.WriteByte('\\')
+			case 's':
+				sb.WriteByte(' ')
+			case 'r':
+				sb.WriteByte('\r')
+			case 'n':
+				sb.WriteByte('\n')
+			default:
+				sb.WriteByte(s[i])
+			}
+			i++
+			escaping = false
+		} else if s[i] == '\\' {
+			if !escaping {
+				escaping = true
+				i++
+				continue
+			}
+		} else {
+			sb.WriteByte(s[i])
+			i++
+		}
+	}
+	return sb.String()
+}
+
+func escapeTag(s string) string {
+	sb := strings.Builder{}
+	for i := range s {
+		switch s[i] {
+		case ';':
+			sb.Write([]byte{'\\', ':'})
+		case ' ':
+			sb.Write([]byte{'\\', 's'})
+		case '\\':
+			sb.Write([]byte{'\\', '\\'})
+		case '\n':
+			sb.Write([]byte{'\\', 'n'})
+		case '\r':
+			sb.Write([]byte{'\\', 'r'})
+		default:
+			sb.WriteByte(s[i])
+		}
+	}
+	return sb.String()
+}
+
+func (m *Message) Canonicalize() []byte {
+	sb := bytes.NewBuffer([]byte{})
+	if m.Tags != nil && len(m.Tags) > 0 {
+		sb.WriteByte('@')
+		for i := range m.Tags {
+			sb.WriteString(m.Tags[i].Key)
+			if m.Tags[i].Value != "" {
+				sb.WriteByte('=')
+				sb.WriteString(escapeTag(m.Tags[i].Value))
+			}
+			if i+1 != len(m.Tags) {
+				sb.WriteByte(';')
+			}
+		}
+		sb.WriteByte(' ')
+	}
+
+	if m.Source != "" {
+		sb.WriteByte(':')
+		sb.WriteString(m.Source)
+		sb.WriteByte(' ')
+	}
+
+	sb.WriteString(string(m.Command))
+
+	if m.Parameters != nil && len(m.Parameters) > 0 {
+		sb.WriteByte(' ')
+		for i := range m.Parameters {
+			if i+1 == len(m.Parameters) && (strings.Contains(m.Parameters[i], " ") || m.Parameters[i] == "" || m.Parameters[i][0] == ':') {
+				sb.WriteByte(':')
+			}
+			sb.WriteString(m.Parameters[i])
+			if i+1 != len(m.Parameters) {
+				sb.WriteByte(' ')
+			}
+		}
+	}
+
+	return sb.Bytes()
 }

message_test.go

@@ -20,7 +20,7 @@ func TestParsing(t *testing.T) {
 		{
 			input: `@id=234AB :dan!d@localhost PRIVMSG #chan :Hey what's up!`,
 			output: &Message{
-				Tags:       map[string]string{"id": "234AB"},
+				Tags:       []Tag{{"id", "234AB"}},
 				Source:     "dan!d@localhost",
 				Command:    PRIVMSG,
 				Parameters: []string{"#chan", "Hey what's up!"},
@@ -28,19 +28,37 @@ func TestParsing(t *testing.T) {
 		{
 			input: `@a=b;c=32;k;rt=ql7 :dan!d@localhost PRIVMSG #chan :Hey what's up!`,
 			output: &Message{
-				Tags:       map[string]string{"a": "b", "c": "32", "k": "", "rt": "ql7"},
+				Tags:       []Tag{{"a", "b"}, {"c", "32"}, {"k", ""}, {"rt", "ql7"}},
 				Source:     "dan!d@localhost",
 				Command:    PRIVMSG,
 				Parameters: []string{"#chan", "Hey what's up!"},
 			}},
+		{
+			input: `@a=b\\and\nk;c=72\s45;d=gh\:764 CAP`,
+			output: &Message{
+				Tags:    []Tag{{"a", "b\\and\nk"}, {"c", "72 45"}, {"d", `gh;764`}},
+				Command: CAP,
+			}},
 		{
 			input: `@tag1=value1;tag2;vendor1/tag3=value2;vendor2/tag4= :dan!d@localhost PRIVMSG #chan :Hey what's up!`,
 			output: &Message{
-				Tags:       map[string]string{"tag1": "value1", "tag2": "", "vendor1/tag3": "value2", "vendor2/tag4": ""},
+				Tags:       []Tag{{"tag1", "value1"}, {"tag2", ""}, {"vendor1/tag3", "value2"}, {"vendor2/tag4", ""}},
 				Source:     "dan!d@localhost",
 				Command:    PRIVMSG,
 				Parameters: []string{"#chan", "Hey what's up!"},
 			}},
+		{
+			input: ":src AWAY",
+			output: &Message{
+				Source:  "src",
+				Command: AWAY,
+			}},
+		{
+			input: ":src AWAY ",
+			output: &Message{
+				Source:  "src",
+				Command: AWAY,
+			}},
 		{
 			input: `CAP REQ :sasl`,
 			output: &Message{
@@ -65,6 +83,13 @@ func TestParsing(t *testing.T) {
 				Command:    CAP,
 				Parameters: []string{"REQ", "    asdf qwer"},
 			}},
+		{
+			input: ":coolguy!ag@net\x035w\x03ork.admin PRIVMSG foo :bar baz",
+			output: &Message{
+				Source:     "coolguy!ag@net\x035w\x03ork.admin",
+				Command:    PRIVMSG,
+				Parameters: []string{"foo", "bar baz"},
+			}},
 		{
 			input: `:coolguy PRIVMSG bar :lol :) `,
 			output: &Message{
@@ -86,6 +111,57 @@ func TestParsing(t *testing.T) {
 				Command:    MODE,
 				Parameters: []string{"#tckk", "+n"},
 			}},
+		{
+			input: ":services.esper.net MODE #foo-bar +o foobar  ",
+			output: &Message{
+				Source:     "services.esper.net",
+				Command:    MODE,
+				Parameters: []string{"#foo-bar", "+o", "foobar"},
+			}},
+		{
+			input: `@tag1=value\\ntest COMMAND`,
+			output: &Message{
+				Tags:    []Tag{{"tag1", `value\ntest`}},
+				Command: command("COMMAND"),
+			}},
+		{
+			input: `@tag1=value\1 COMMAND`,
+			output: &Message{
+				Tags:    []Tag{{"tag1", `value1`}},
+				Command: command("COMMAND"),
+			}},
+		{
+			input: `@tag1=value\ COMMAND`,
+			output: &Message{
+				Tags:    []Tag{{"tag1", `value`}},
+				Command: command("COMMAND"),
+			}},
+		{
+			input: `@tag1=1;tag2=3;tag3=4;tag1=5 COMMAND`,
+			output: &Message{
+				Tags:    []Tag{{"tag1", "5"}, {"tag2", "3"}, {"tag3", "4"}},
+				Command: command("COMMAND"),
+			}},
+		{
+			input: `@tag1=1;tag2=3;tag3=4;tag1=5;vendor/tag2=8 COMMAND`,
+			output: &Message{
+				Tags:    []Tag{{"tag1", "5"}, {"tag2", "3"}, {"tag3", "4"}, {"vendor/tag2", "8"}},
+				Command: command("COMMAND"),
+			}},
+		{
+			input: `:SomeOp MODE #channel :+i`,
+			output: &Message{
+				Source:     "SomeOp",
+				Command:    MODE,
+				Parameters: []string{"#channel", "+i"},
+			}},
+		{
+			input: `:SomeOp MODE #channel +oo SomeUser :AnotherUser`,
+			output: &Message{
+				Source:     "SomeOp",
+				Command:    MODE,
+				Parameters: []string{"#channel", "+oo", "SomeUser", "AnotherUser"},
+			}},
 	}
 	for _, tc := range testcases {
 		m := &Message{Raw: tc.input}
@@ -94,6 +170,12 @@ func TestParsing(t *testing.T) {
 			t.Logf("tags: actual: %v, expected: %v", m.Tags, tc.output.Tags)
 			t.Fail()
 		}
+		for i := range tc.output.Tags {
+			if tc.output.Tags[i] != m.Tags[i] {
+				t.Logf("tags: actual: %+v, expected: %+v", m.Tags[i], tc.output.Tags[i])
+				t.Fail()
+			}
+		}
 		if tc.output.Source != m.Source {
 			t.Logf("source: actual: %v, expected: %v", m.Source, tc.output.Source)
 			t.Fail()
@@ -106,5 +188,129 @@ func TestParsing(t *testing.T) {
 			t.Logf("parameters: actual: %v (%d), expected: %v (%d)", m.Parameters, len(m.Parameters), tc.output.Parameters, len(tc.output.Parameters))
 			t.Fail()
 		}
+		for i := range tc.output.Parameters {
+			if tc.output.Parameters[i] != m.Parameters[i] {
+				t.Logf("parameters: actual: %v, expected: %v", m.Parameters[i], tc.output.Parameters[i])
+				t.Fail()
+			}
+		}
+	}
+}
+
+func TestCanonicalization(t *testing.T) {
+	testcases := []struct {
+		input  *Message
+		output string
+	}{
+		// these outputs are based on
+		// https://github.com/ircdocs/parser-tests/blob/master/tests/msg-split.yaml
+		// and https://modern.ircdocs.horse/#messages
+		{
+			input: &Message{
+				Source:     "irc.example.com",
+				Command:    CAP,
+				Parameters: []string{"LS", "*", "multi-prefix extended-join sasl"},
+			},
+			output: `:irc.example.com CAP LS * :multi-prefix extended-join sasl`,
+		},
+		{
+			input: &Message{
+				Tags:       []Tag{{"id", "234AB"}},
+				Source:     "dan!d@localhost",
+				Command:    PRIVMSG,
+				Parameters: []string{"#chan", "Hey what's up!"},
+			},
+			output: `@id=234AB :dan!d@localhost PRIVMSG #chan :Hey what's up!`,
+		},
+		{
+			input: &Message{
+				Tags:       []Tag{{"a", "b"}, {"c", "32"}, {"k", ""}, {"rt", "ql7"}},
+				Source:     "dan!d@localhost",
+				Command:    PRIVMSG,
+				Parameters: []string{"#chan", "Hey what's up!"},
+			},
+			output: `@a=b;c=32;k;rt=ql7 :dan!d@localhost PRIVMSG #chan :Hey what's up!`,
+		},
+		{
+			input: &Message{
+				Tags:    []Tag{{"a", "b\\and\nk"}, {"c", "72 45"}, {"d", `gh;764`}},
+				Command: CAP,
+			},
+			output: `@a=b\\and\nk;c=72\s45;d=gh\:764 CAP`,
+		},
+		{
+			input: &Message{
+				Tags:       []Tag{{"tag1", "value1"}, {"tag2", ""}, {"vendor1/tag3", "value2"}, {"vendor2/tag4", ""}},
+				Source:     "dan!d@localhost",
+				Command:    PRIVMSG,
+				Parameters: []string{"#chan", "Hey what's up!"},
+			},
+			output: `@tag1=value1;tag2;vendor1/tag3=value2;vendor2/tag4 :dan!d@localhost PRIVMSG #chan :Hey what's up!`,
+		},
+		{
+			input: &Message{
+				Source:  "src",
+				Command: AWAY,
+			},
+			output: ":src AWAY",
+		},
+		{
+			input: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", "sasl"},
+			},
+			output: `CAP REQ sasl`,
+		},
+		{
+			input: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", ""},
+			},
+			output: `CAP REQ :`,
+		},
+		{
+			input: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", ":asdf"},
+			},
+			output: `CAP REQ ::asdf`,
+		},
+		{
+			input: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", "    asdf qwer"},
+			},
+			output: `CAP REQ :    asdf qwer`,
+		},
+		{
+			input: &Message{
+				Source:     "coolguy",
+				Command:    PRIVMSG,
+				Parameters: []string{"bar", "lol :) "},
+			},
+			output: `:coolguy PRIVMSG bar :lol :) `,
+		},
+		{
+			input: &Message{
+				Source:     "coolguy!ag@net\x035w\x03ork.admin",
+				Command:    PRIVMSG,
+				Parameters: []string{"foo", "bar baz"},
+			},
+			output: ":coolguy!ag@net\x035w\x03ork.admin PRIVMSG foo :bar baz",
+		},
+		{
+			input: &Message{
+				Tags:    []Tag{{"tag1", `value\ntest`}},
+				Command: command("COMMAND"),
+			},
+			output: `@tag1=value\\ntest COMMAND`,
+		},
+	}
+	for _, tc := range testcases {
+		output := string(tc.input.Canonicalize())
+		if output != tc.output {
+			t.Logf("received '%s', expected '%s'", output, tc.output)
+			t.Fail()
+		}
 	}
 }