From 068ba8bfb0b48c7ecca87331faa2186108bf4a1e Mon Sep 17 00:00:00 2001
From: David Ashby <delta.mu.alpha@gmail.com>
Date: Thu, 7 Oct 2021 23:22:06 -0400
Subject: [PATCH] initial message-parsing work

---
 const.go        | 177 ++++++++++++++++++++++++++++++++++++++++++++++++
 go.mod          |   3 +
 message.go      | 137 +++++++++++++++++++++++++++++++++++++
 message_test.go | 110 ++++++++++++++++++++++++++++++
 4 files changed, 427 insertions(+)
 create mode 100644 const.go
 create mode 100644 go.mod
 create mode 100644 message.go
 create mode 100644 message_test.go

diff --git a/const.go b/const.go
new file mode 100644
index 0000000..77b1c71
--- /dev/null
+++ b/const.go
@@ -0,0 +1,177 @@
+package main
+
+type command string
+
+type channelType int
+
+const (
+	REGULAR channelType = iota
+	LOCAL
+)
+
+// client commands
+const (
+	CAP          command = "CAP"
+	AUTHENTICATE command = "AUTHENTICATE"
+	PASS         command = "PASS"
+	NICK         command = "NICK"
+	USER         command = "USER"
+	OPER         command = "OPER"
+	QUIT         command = "QUIT"
+
+	JOIN   command = "JOIN"
+	PART   command = "PART"
+	TOPIC  command = "TOPIC"
+	NAMES  command = "NAMES"
+	LIST   command = "LIST"
+	INVITE command = "INVITE"
+	KICK   command = "KICK"
+
+	MOTD    command = "MOTD"
+	VERSION command = "VERSION"
+	ADMIN   command = "ADMIN"
+	CONNECT command = "CONNECT"
+	TIME    command = "TIME"
+	STATS   command = "STATS"
+	INFO    command = "INFO"
+	MODE    command = "MODE"
+
+	PRIVMSG command = "PRIVMSG"
+	NOTICE  command = "NOTICE"
+
+	KILL command = "KILL"
+
+	AWAY     command = "AWAY"
+	USERHOST command = "USERHOST"
+)
+
+// numerics
+const (
+	RPL_WELCOME  command = "001"
+	RPL_YOURHOST command = "002"
+	RPL_CREATED  command = "003"
+	RPL_MYINFO   command = "004"
+	RPL_ISUPPORT command = "005"
+
+	RPL_BOUNCE command = "010"
+
+	RPL_UMODEIS       command = "221"
+	RPL_LUSERCLIENT   command = "251"
+	RPL_LUSEROP       command = "252"
+	RPL_LUSERUNKNOWN  command = "253"
+	RPL_LUSERCHANNELS command = "254"
+	RPL_LUSERME       command = "255"
+	RPL_ADMINME       command = "256"
+	RPL_ADMINLOC1     command = "257"
+	RPL_ADMINLOC2     command = "258"
+	RPL_ADMINEMAIL    command = "259"
+
+	RPL_TRYAGAIN command = "263"
+
+	RPL_LOCALUSERS  command = "265"
+	RPL_GLOBALUSERS command = "266"
+
+	RPL_WHOISCERTFP command = "276"
+
+	RPL_NONE command = "300"
+
+	RPL_AWAY     command = "301"
+	RPL_USERHOST command = "302"
+	RPL_ISON     command = "303"
+	RPL_UNAWAY   command = "305"
+	RPL_NOWAWAY  command = "306"
+
+	RPL_WHOISUSER     command = "311"
+	RPL_WHOISSERVER   command = "312"
+	RPL_WHOISOPERATOR command = "313"
+	RPL_WHOWASUSER    command = "314"
+	RPL_WHOISIDLE     command = "317"
+	RPL_ENDOFWHOIS    command = "318"
+	RPL_WHOISCHANNELS command = "319"
+
+	RPL_LISTSTART command = "321"
+	RPL_LIST      command = "322"
+	RPL_LISTEND   command = "323"
+
+	RPL_CHANNELMODEIS command = "324"
+	RPL_CREATIONTIME  command = "329"
+	RPL_NOTOPIC       command = "331"
+	RPL_TOPIC         command = "332"
+	RPL_TOPICWHOTIME  command = "333"
+
+	RPL_INVITING        command = "341"
+	RPL_INVITELIST      command = "346"
+	RPL_ENDOFINVITELIST command = "347"
+	RPL_EXCEPTLIST      command = "348"
+	RPL_ENDOFEXCEPTLIST command = "349"
+
+	RPL_VERSION command = "351"
+
+	RPL_NAMREPLY     command = "353"
+	RPL_ENDOFNAMES   command = "366"
+	RPL_BANLIST      command = "367"
+	RPL_ENDOFBANLIST command = "368"
+	RPL_ENDOFWHOWAS  command = "369"
+
+	RPL_MOTD      command = "372"
+	RPL_MOTDSTART command = "375"
+	RPL_ENDOFMOTD command = "376" // see also ERR_NOMOTD
+
+	RPL_YOUREOPER command = "381"
+
+	RPL_REHASHING command = "382"
+
+	ERR_UNKNOWNERROR     command = "400"
+	ERR_NOSUCHNICK       command = "401"
+	ERR_NOSUCHSERVER     command = "402"
+	ERR_NOSUCHCHANNEL    command = "403"
+	ERR_CANNOTSENDTOCHAN command = "404"
+	ERR_TOOMANYCHANNELS  command = "405"
+
+	ERR_UNKNOWNCOMMAND command = "421"
+
+	ERR_NOMOTD command = "422"
+
+	ERR_ERRONEUSNICKNAME command = "432"
+	ERR_NICKNAMEINUSE    command = "433"
+
+	ERR_USERNOTINCHANNEL command = "441"
+	ERR_NOTONCHANNEL     command = "442"
+	ERR_USERONCHANNEL    command = "443"
+
+	ERR_NOTREGISTERED command = "451"
+
+	ERR_NEEDMOREPARAMS    command = "461"
+	ERR_ALREADYREGISTERED command = "462"
+	ERR_PASSWDMISMATCH    command = "464"
+	ERR_YOUREBANNEDCREEP  command = "465"
+	ERR_CHANNELISFULL     command = "471"
+	ERR_UNKNOWNMODE       command = "472"
+	ERR_INVITEONLYCHAN    command = "473"
+	ERR_BANNEDFROMCHAN    command = "474"
+	ERR_BADCHANNELKEY     command = "475"
+	ERR_BADCHANMASK       command = "476"
+
+	ERR_NOPRIVILEGES     command = "481"
+	ERR_CHANOPRIVSNEEDED command = "482"
+	ERR_CANTKILLSERVER   command = "483"
+	ERR_NOOPERHOST       command = "491"
+
+	ERR_UMODEUNKNOWNFLAG command = "501"
+	ERR_USERSDONTMATCH   command = "502"
+
+	RPL_STARTTLS command = "670"
+	ERR_STARTTLS command = "691"
+
+	ERR_NOPRIVS command = "723"
+
+	RPL_LOGGEDIN    command = "900"
+	RPL_LOGGEDOUT   command = "901"
+	ERR_NICKLOCKED  command = "902"
+	RPL_SASLSUCCESS command = "903"
+	ERR_SASLFAIL    command = "904"
+	ERR_SASLTOOLONG command = "905"
+	ERR_SASLABORTED command = "906"
+	ERR_SASLALREADY command = "907"
+	RPL_SASLMECHS   command = "908"
+)
diff --git a/go.mod b/go.mod
new file mode 100644
index 0000000..e42f7dc
--- /dev/null
+++ b/go.mod
@@ -0,0 +1,3 @@
+module git.yetaga.in/alazyreader/irc
+
+go 1.17
diff --git a/message.go b/message.go
new file mode 100644
index 0000000..bc167b3
--- /dev/null
+++ b/message.go
@@ -0,0 +1,137 @@
+package main
+
+import (
+	"fmt"
+	"strings"
+)
+
+type Message struct {
+	Tags       map[string]string
+	Source     string
+	Command    command
+	Parameters []string
+	Raw        string
+
+	parseIndex int
+}
+
+func (m *Message) ParseMessage() error {
+	m.parseIndex = 0
+	m.consumeSpaces()
+	if m.consume('@') {
+		m.Tags, _ = m.parseTags()
+	}
+	m.consumeSpaces()
+	if m.consume(':') {
+		m.Source, _ = m.parseSource()
+	}
+	m.consumeSpaces()
+	m.Command, _ = m.parseCommand()
+	m.consumeSpaces()
+	m.Parameters, _ = m.parseParameters()
+	return nil
+}
+
+func (m *Message) parseTags() (map[string]string, error) {
+	tags := make(map[string]string)
+	for {
+		next := m.findNext("=", ";")
+		if next == -1 {
+			break
+		} else if m.Raw[m.parseIndex+next] == '=' {
+			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
+		} else if m.Raw[m.parseIndex+next] == ';' {
+			key := m.Raw[m.parseIndex : m.parseIndex+next]
+			m.parseIndex += next
+			tags[key] = ""
+			m.parseIndex++
+		} else {
+			break
+		}
+	}
+	return tags, nil
+}
+
+func (m *Message) parseSource() (string, error) {
+	start := m.parseIndex
+	endofparse := strings.Index(m.Raw[m.parseIndex:], " ")
+	if endofparse == -1 {
+		return "", fmt.Errorf("end of string encountered while parsing tags")
+	}
+	m.parseIndex += endofparse
+	return m.Raw[start:m.parseIndex], nil
+}
+
+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 += endofparse
+	return command(m.Raw[start:m.parseIndex]), nil
+}
+
+func (m *Message) parseParameters() ([]string, error) {
+	params := []string{}
+	for m.parseIndex <= len(m.Raw) {
+		m.consumeSpaces()
+		if m.consume(':') { // "run until end of line"
+			params = append(params, m.Raw[m.parseIndex:])
+			break
+		}
+		endofparse := strings.Index(m.Raw[m.parseIndex:], " ")
+		if endofparse == -1 { // no further params after this one
+			p := strings.TrimSuffix(m.Raw[m.parseIndex:], " ")
+			if len(p) > 0 { // reject empty trailing params
+				params = append(params, p)
+			}
+			break
+		}
+		params = append(params, m.Raw[m.parseIndex:m.parseIndex+endofparse])
+		m.parseIndex += endofparse
+	}
+	return params, nil
+}
+
+func (m *Message) consumeSpaces() {
+	if len(m.Raw) <= m.parseIndex {
+		return
+	}
+	for len(m.Raw) > m.parseIndex && m.Raw[m.parseIndex] == ' ' {
+		m.parseIndex++
+	}
+}
+
+func (m *Message) consume(r byte) bool {
+	if len(m.Raw) <= m.parseIndex {
+		return false
+	}
+	if m.Raw[m.parseIndex] == r {
+		m.parseIndex++
+		return true
+	}
+	return false
+}
+
+func (m *Message) findNext(bs ...string) int {
+	if len(bs) == 0 {
+		return 0
+	}
+	r := -1
+	for _, b := range bs {
+		i := strings.Index(m.Raw[m.parseIndex:], b)
+		if r == -1 || (i != -1 && i < r) {
+			r = i
+		}
+	}
+	return r
+}
+
+func (m *Message) ToBytes() []byte {
+	return nil
+}
diff --git a/message_test.go b/message_test.go
new file mode 100644
index 0000000..553d3f3
--- /dev/null
+++ b/message_test.go
@@ -0,0 +1,110 @@
+package main
+
+import "testing"
+
+func TestParsing(t *testing.T) {
+	testcases := []struct {
+		input  string
+		output *Message
+	}{
+		// these inputs are based on
+		// https://github.com/ircdocs/parser-tests/blob/master/tests/msg-split.yaml
+		// and https://modern.ircdocs.horse/#messages
+		{
+			input: `:irc.example.com CAP LS * :multi-prefix extended-join sasl`,
+			output: &Message{
+				Source:     "irc.example.com",
+				Command:    CAP,
+				Parameters: []string{"LS", "*", "multi-prefix extended-join sasl"},
+			}},
+		{
+			input: `@id=234AB :dan!d@localhost PRIVMSG #chan :Hey what's up!`,
+			output: &Message{
+				Tags:       map[string]string{"id": "234AB"},
+				Source:     "dan!d@localhost",
+				Command:    PRIVMSG,
+				Parameters: []string{"#chan", "Hey what's up!"},
+			}},
+		{
+			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"},
+				Source:     "dan!d@localhost",
+				Command:    PRIVMSG,
+				Parameters: []string{"#chan", "Hey what's up!"},
+			}},
+		{
+			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": ""},
+				Source:     "dan!d@localhost",
+				Command:    PRIVMSG,
+				Parameters: []string{"#chan", "Hey what's up!"},
+			}},
+		{
+			input: `CAP REQ :sasl`,
+			output: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", "sasl"},
+			}},
+		{
+			input: `CAP REQ :`,
+			output: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", ""},
+			}},
+		{
+			input: `CAP REQ ::asdf`,
+			output: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", ":asdf"},
+			}},
+		{
+			input: `CAP REQ :    asdf qwer`,
+			output: &Message{
+				Command:    CAP,
+				Parameters: []string{"REQ", "    asdf qwer"},
+			}},
+		{
+			input: `:coolguy PRIVMSG bar :lol :) `,
+			output: &Message{
+				Source:     "coolguy",
+				Command:    PRIVMSG,
+				Parameters: []string{"bar", "lol :) "},
+			}},
+		{
+			input: `:gravel.mozilla.org 432  #momo :Erroneous Nickname: Illegal characters`,
+			output: &Message{
+				Source:     "gravel.mozilla.org",
+				Command:    ERR_ERRONEUSNICKNAME,
+				Parameters: []string{"#momo", "Erroneous Nickname: Illegal characters"},
+			}},
+		{
+			input: `:gravel.mozilla.org MODE #tckk +n `,
+			output: &Message{
+				Source:     "gravel.mozilla.org",
+				Command:    MODE,
+				Parameters: []string{"#tckk", "+n"},
+			}},
+	}
+	for _, tc := range testcases {
+		m := &Message{Raw: tc.input}
+		m.ParseMessage()
+		if len(tc.output.Tags) != len(m.Tags) {
+			t.Logf("tags: actual: %v, expected: %v", m.Tags, tc.output.Tags)
+			t.Fail()
+		}
+		if tc.output.Source != m.Source {
+			t.Logf("source: actual: %v, expected: %v", m.Source, tc.output.Source)
+			t.Fail()
+		}
+		if tc.output.Command != m.Command {
+			t.Logf("command: actual: %v, expected: %v", m.Command, tc.output.Command)
+			t.Fail()
+		}
+		if len(tc.output.Parameters) != len(m.Parameters) {
+			t.Logf("parameters: actual: %v (%d), expected: %v (%d)", m.Parameters, len(m.Parameters), tc.output.Parameters, len(tc.output.Parameters))
+			t.Fail()
+		}
+	}
+}