gempir / gempbot-node Goto Github PK

Logs should probably be not part of the standard bot anymore.
I want to move them over to a go application and my bot will be able to query data over an api

!countdown 10

and then count down

This is kinda difficult to achieve. My best idea would be a cross channel command like

!glogs nymn_hs gempir

Which would get the log file for gempir in #nymn_hs

currently all emotes are from a list via twitchemotes. The bot should see once a new ircv3 tag gets used and save the tag + the code in redis

the package.json file is growing and some packages are just not really essential

Currently everytime the bot restarts it has to make several big http requests. Redis should act has a cache of these emotes and then the bot itself caches from redis once it starts.

Only make make http requests every few hours or on command.

There is no need for globalcooldown anymore maybe add a per user cooldown otherwise just remove it.

Maybe consider monitoring the total output of the bot and make sure to not go over 100 messages / 30 secs

When caching emotes sometimes emotes from different channels get cached into redis in the wrong channel

Currently the whole bot gets killed when reconnecting

in an interval of choice random facts should be output.

!fact add my fact text (add a fact to the bot)
!fact set 300 (meaning every 300 seconds)

Squiddy requested these facts:
http://www.squid-world.com/facts-about-squids/

Some commands shouldn't be used so much, so instead of changing the global cooldown give each command atleast a specific cooldown for now.

The oddshot module is broken recording oddshots multiple times sometimes.

The whole module needs to be reworked

Combos should never display links

Kappa

Tmi.js was nice but relaybroker is its replacement. Everything is already handled over relaybroker now, tmi.js can be removed

The end goal of the bot is multichannel support. So it's important that every command will work independently in the channels and will not interfere with eachother

Alias for commands would be great, but they need to be persistent so that f.e. when you edit an alias you edit the originial command etc.

It coud look like this:

!alias add myAlias myCommand

The timeout manager will manage simple timeouts like for spam.

This will happen in 3 stages.

• first offense 10s
• second offense 30s
• third offense 90s

The next stage will trigger if a user gets flagged a second time in the last 300 seconds.

Minimum string length for quotes so it filters simple things MingLee

New command !8ball or something fitting. Use !8ball <Question for 8ball>. Command will pull a random quote from a random viewers logs as a respond for the question asked. Example, -Billy123: Will Forsen ever be lucky? -Gempbot: "Random quote" or -Gempbot: "-Billy123: Will Forsen ever be lucky?" "Random quote"

Output: number of people who are in chat and have written something within the last x minutes.
(maybe 10 minutes?)

This has nothing to do with github, but I want a more automatic way of deploying.

Banphrases seem to be slightly bugged.

And not timing out users. tested by directly adding a banphrase and the user only saying that phrase.

Splittings logs into months and years would make a lot of stuff a lot more perfomant and it will scale better in the future.

the structure could me like this

logs/gempir/2016/january/gempir.txt

maybe I can offer an API for it then aswell, because the files won't be as big.

• use .createStream() from irc-message

net.connect(6667, 'irc.freenode.net')
    .pipe(ircMsg.createStream())
    .on('data', function(message) {
        console.log(message)
    })

• setNoDelay is for disabling TCP buffering meaning more packets will be sent

Add a simple timer like !timer 600 Pizza so people can be reminded by it.

channels are controlled over redis so when redis changes it also needs to update the bot sometime so so the bot can join channels without restarting

• use regex for message parsing
• get rid of cluttered chat.js and create a real message Handler and parser etc.

At this scale TDD would probably be wiser, but first the existing code needs to be tested.