Awesome

gochatbot

This is a chatops bot implemented in Go.

It features:

Support for Slack, Telegram and IRC
Non-durable and durable memory with BoltDB and Redis
Two ready to use rulesets: regex parsed messages and cron events
Easy integration with other programming languages through a plugin system (webservice JSON-RPC).
Container ready to use and easy to deploy
Possibility to deploy more than one bot with the same binary

Requirements: glide and Go 1.5 or newer.

Environmental Variables

Basic

GOCHATBOT_NAME - defines the name of the bot. It will honor this name to answer messages in chatrooms.

Durable Memory (BoltDB)

GOCHATBOT_BOLT_FILENAME - local file name of BoltDB file. Set it if you want BoltDB support activated.

Durable Memory (Redis)

GOCHATBOT_REDIS_DATABASE - Database index
GOCHATBOT_REDIS_HOST - Redis Hostname (with port)
GOCHATBOT_REDIS_PASSWORD - Redis database password

IRC message provider

GOCHATBOT_IRC_USER - IRC user to connect to the server
GOCHATBOT_IRC_NICK - IRC nickname for this account (will not handle nick renaming)
GOCHATBOT_IRC_SERVER - IRC server
GOCHATBOT_IRC_CHANNELS - comma-separated list of channels to connect to.
GOCHATBOT_IRC_PASSWORD - IRC password for user
GOCHATBOT_IRC_TLS - whether TLS connection must be used or not (0/1)

Slack message provider

GOCHATBOT_SLACK_TOKEN - Slack user token for the chatbot

Telegram message provider

GOCHATBOT_TELEGRAM_TOKEN - Telegram user token for the chatbot

RPC

GOCHATBOT_RPC_BIND - local IP address to bind the RPC HTTP server

Quick start (Compiled)

# go get -tags all cirello.io/gochatbot/...

Quick start (Docker version - Slack - Non-durable memory)

# docker run -e "GOCHATBOT_SLACK_TOKEN=xxxx-xxxx" -d ccirello/gochatbot

Quick start (Local Docker version - CLI - Non-durable memory)

# brew install glide
# make docker
# docker run -ti ccirello/gochatbot

(Type gochatbot jump and press enter. Do it twice. docker stop to exit.)

Quick start (Compiled version - Telegram - BoltDB memory)

# make quickstart
# GOCHATBOT_TELEGRAM_TOKEN=#####:... GOCHATBOT_BOLT_FILENAME=gochatbot.db ./gochatbot

Extending

A more thorough manual of how to extend the bot is yet to be written. This section will cover just adding, removing and modifying rules into the two shipped rulesets: regex and cron.

Extending regex ruleset

Edit the file rules.go in package root directory. Look for the line starting with: var regexRules = []regex.Rule{. This is the beginning of the data structure that comprises all regex rules.

Why regex rule? Because it uses regex to extract from the body of the message the parameters for later message parsing.

All regex rules have the following structure:

{
	`regex pattern (.*)`, `explanation of the rule`,
	func(bot bot.Self, msg string, args []string) []string {
		var ret []string

		// your logic comes here

		return ret
	},
},

regex pattern (.*) - is the regex match pattern to be used against the incoming message. Note that you can tell apart the messages that are sent in the room from those sent to the chatbot, by append a Go text/template variable.

`{{ .RobotName }} jump`, `tells the robot to jump`, // Only messages starting with bot's name will be parsed
// vs
`jump`, `tells the robot to jump`, // All messages whose content matches jump will be matched

explanation of the rule - is a human readable explanation of the rule, meant to be displayed with {{ .RobotName }} help command (if you haven't changed the bot's name, it will be gochatbot help).

func(bot bot.Self, msg string, args []string) []string { - is the function which parses and reacts on the incoming message. Few important as aspects:

Each incoming message is its own goroutine. It means you can execute blocking calls and the bot will keep working as usual.
Look at bot package documentation for what you can do with the bot. In the following example, we'll see how we can use the bot's brain to store state across messages.
msg is the raw version of the message.
args is the slice of strings that matched against the regex pattern (.*).
It must return a slice of strings, even if it is empty.

So, this is a practical and annotated version of the jump regex rule.

var regexRules = []regex.Rule{
	{
		`{{ .RobotName }} jump`, // Regex rule. No matching, so args will be empty
		`tells the robot to jump`, // Just a simple explanation of the rule
		func(bot bot.Self, msg string, args []string) []string {
			var ret []string

			ret = append(ret, "{{ .User }}, How high?") // In the outgoing messages, the text/template variable "{{ .User }}" is replaced with username.

			lastJumpTS := bot.MemoryRead("jump", "lastJump") // Reads from the bot's brain the last time this command was executed.
			ret = append(ret, fmt.Sprint("{{ .User }} (last time I jumped:", lastJumpTS, ")")) // Append this information to the outgoing messages slice.

			bot.MemorySave("jump", "lastJump", fmt.Sprint(time.Now())) // Saves into the bot's brain that it has just jumped.

			return ret
		},
	},
}

Extending cron ruleset

Edit the file rules.go in package root directory. Look for the line starting with: cronRules = map[string]cron.Rule{. This is the beginning of the data structure that comprises all cron rules.

Why cron rule? Because it uses crontab format to periodically execute tasks, that may, or may not yield messages onto a chat room.

All cron rules have the following structure:

"job name": {
	"crontab format",
	func() []messages.Message {
		return []messages.Message{}
	},
},

"job name" - the name of the cron task to be executed. Each cron task must be attached to a chatroom, otherwise they don't get executed. This attachment is done through the command "cron attach task-name" in the desired chatroom.

"crontab format" - the crontab-like set that tells how often this rule is executed. Refer to the cronexpr for the implementation.

func() []messages.Message { - the niladic function that's executed periodically. It returns a slice of messages.Message. But the only required value is the messages.Message.Message string field. It will overwrite the other values with the context information for correct delivery.

So, this is a practical and annotated version of the good morning cron rule.

var cronRules = map[string]cron.Rule{
	// name of the cron rule
	"message of the day": {
		"0 10 * * *", // every day at 10:00
		func() []messages.Message {
			return []messages.Message{
				{Message: "Good morning!"}, // Returns "Good Morning"
			}
		},
	},
}

Integrating with other languages (RPC)

If GOCHATBOT_RPC_BIND is set, gochatbot will open a HTTP server in the given address and it will expose four endpoints: /pop and /send for message handling; /memoryRead and /memorySave for memory manipulation.

Message endpoints use a JSON serialized version of the internal representation of messages. Thus if you get from /pop this:

{
	"Room":"room",
	"FromUserID":"fUID",
	"FromUserName":"fName",
	"ToUserID":"tUID",
	"ToUserName":"tName",
	"Message":"Message"
}

Probably you should be inverting From* with To* and returning something like this (note the inversion of "from" with "to" values):

{
	"Room":"room",
	"FromUserID":"tUID",
	"FromUserName":"tName",
	"ToUserID":"fUID",
	"ToUserName":"fName",
	"Message":"Message"
}

Check the rpc-example.php file for an implementation of an echo service in PHP.

For memory manipulation, /memoryRead needs to parameters in a GET request: GET /memoryRead?namespace=NS&key=K where NS is the namespace isolating the space of memory of this plugin from the others, and K the key name for the content. The response is a raw string to be parsed by the plugin internals.

/memorySave works similarly as /memoryRead, except it expects a POST call, and the body of the request is the raw string to be stored within the memory durable storage.

Multibot mode

You might find yourself in a situation where you want to deploy the same bot in two or more places at the same time. Perhaps, even sharing the same memory. It is possible to activate the multi-mode for gochatbot. Compile with multi build tag:

# go build -tags 'all multi' # 'multi' will use gochatbot-multi.go instead of gochatbot.go

Each configuration variable must be adapted with an index. Therefore:

# export GOCHATBOT_NAME="gochatbot"
# export GOCHATBOT_BOLT_FILENAME="gochatbot.db"
# export GOCHATBOT_SLACK_TOKEN="slack-token"

will need to be translated with an index number in the middle, like:

# export GOCHATBOT_0_NAME="gochatbot"
# export GOCHATBOT_0_BOLT_FILENAME="gochatbot.db"
# export GOCHATBOT_0_SLACK_TOKEN="slack-token"

# export GOCHATBOT_1_NAME="gochatbot"
# export GOCHATBOT_1_BOLT_FILENAME="gochatbot.db"
# export GOCHATBOT_1_TELEGRAM_TOKEN="telegram-token"

(The bot will connect to both Slack and Telegram)

Plugins

Gochatbot accept external plugins to its core. The plugin interface is actually a facade for the RPC ruleset and external files. If any file named beginning with gochatbot-plugin- and it has the execution bit set, it will be executed by main gochatbot process with an environmental variable GOCHATBOT_RPC_BIND containing an IPv4 address endpoint. All endpoints available for RPC ruleset are available in this exposition. Check example-gochatbot-plugin-logger.sh for an example of a plugin.

It also means that, as long as you are to execute the file, your plugin can be written in any language.

Reddit plugin

It is possible to track Reddit content with gochatbot. Useful when the team needs to monitor it for upcoming events, like release notifications.

reddit follow subreddit- follow one subreddit in a room
reddit unfollow subreddit - unfollow one subreddit in a room

Compile plugins/gochatbot-plugin-reddit, and place the resulting binary in the same working directory of main gochatbot binary.

Guarantees

I guarantee that I will maintain this chatops bot for the next 2 years, provide it with updates, Github Issues support and issue updates to keep it compatible with newer Go versions.

Also, I will work to my best to ensure a vibrant community around this bot, so even in the case I step down, I hope by the end of the guaranteed period, to have a project bigger than one man effort.

The last day of guaranteed action on this bot is 2017-12-05.