luni3359 / koa-bot Goto Github PK

It's plain text currently.

Currently on testing phase. Would be good to see it process youtube links.

Currently there's many instances of repeating code across the project, it'd be great if there was a system built for every of these to avoid repetition.

• Make a cog for networking
• Make a cog for handling OAuth2 (caching appropriate tokens accordingly)

More ideas to come later.

Very useful to know what version a bot is running on. Maybe include what commit it's in, along with its description?

This affects automatic galleries. This could technically be abused to show previews the user did not intend to. Nobody is going to do that, but it's still a regex bug.

Example of invalid matches:

• https://xyztwitter.com/user/status/0123456788
• https://twitter.com.xyz/user/status/0123456788

Very important.

The results may be status:safe, but they can include tags that aren't sfw. The post themselves aren't displayed on safebooru, and are marked with a warning that they're not safe.

There ought to be a way to search exclusively for posts meant to appear only in safebooru.

No idea how this happened, but it makes the bot stop posting altogether. Can still make queries and retrieve information. Will keep updated.

As per issue #30 I've come to realize the whole way the configuration file works is flawed. All of it is squashed within the bot variable. It's also not user-friendly.

Proposed changes:

• Put domains in their own file, and commit it (whether or not it should include data like favicons is to be discussed)
• Remove dependency on the bot variable to retrieve data from the configuration.
• Put quotes into their own file
  • ...with multi-lingual support!

Maybe not explicitly those options, as things could change in the future, but setting up the bot automatically is a good start.

Everything is kind of polluting the main file, we should make it more modular with the use of cogs.

Saw it happen with a danbooru gallery display with 9 images. More details later.

The current warning system serves the purpose of keeping media-based text channels from getting flooded with conversations that should go in different text channels (which are not entirely disallowed, but are discouraged).

How it currently behaves:

1. Counter keeps track of how many messages have been sent in any given text channel.
   • The counter resets when a message is sent in a different text channel.
   • Sending urls resets the counter.
2. If the counter surpasses the defined maximum, a warning is issued.

Current issues:

• Sending messages across multiple servers resets the counter, making the counter more useless the more channels the bot is in due to message frequency.
• Media-based commands don't reset the counter.

It shows nothing with the word "defurral". The main dictionary parser will have to be reworked.

Now that the project has been cogyfied it'll be extremely satisfying to have the cogs automatically reload on save.

Links such as:
https://www.website.com/directory/%3Bword%27s%20sand%20n%27%20fiction/a%20dog%20in%20the%20cm.jpg

...get converted as if they had a unit in them.
Possible fixes? Scratch out the dedicated unit parser, write a parser that checks for more things other than just units.

Unknown cause as of now.

Things to track:

• Users joined (and who)
• Users leaving (and who)

It should improve upload speeds to some degree. Testing needed.

In other words, make it so that only 1 picture is shown, but the user(s) can switch between pictures, and it updates in real time.

Issues I know we'll have firsthand:

• Uploaded pictures to embeds cannot be changed, nor you can upload a picture to an existing embed.

The error occurs on post #3808380.

They just don't convert

If a link is beyond obvious a danbooru post, automatically find it.

When there are not search matches Koa silently fails. Something should be said and, as a side feature, tell the user possible fixes to their tags (if the tag didn't exist).

It has been a problem since forever. Too often duplicates are posted.

Right now files have no limit as to how much space they can take up, and how it's handled is all separate from each other. Some central class should be in charge of doing this.

Right now it doesn't really work optimally, and it's kind of impossible to make multiple of the same domains get detected. I suggest:

• Wildcard domain support
• Capable to map multiple domains together to some extent
• No longer require the domains to be inside the configuration file

Multiple instances of ThreadPoolExecutor build up relatively quickly. This has to be investigated in-depth.

Bash doesn't work on Windows. There should be a functional Windows equivalent of run.sh.

To allow for more text, the bot should send separate embeds if necessary.

Nice feature to have as well!

Currently the user has to restart the whole bot to apply updates.

It doesn't recognize this and still says " has rolled the die." regardless of quantity.

Previously if you ran the bot in debug mode alongside the live one, the live bot would ignore all requests. Now the live bot keeps working for all intents and purposes, aside from ignoring the on_message tasks, because that has an explicit check preventing it.

I don't know how good this will be or whether or not we need this, but it'll be included, whatever it is.

Not sure what lib to use, but I suppose we could do it hard-coded as long as we cover yen and usd.

As per issue #25:

1. A lot of errors show up on the terminal each time the check doesn't pass, which is every time a command is used while the beta bot is running (it's annoying).
2. Because it only applies to commands, the old check on the on_message event was unaffected. It would be a good idea to "modernize" it by using actual discord.py checks.

Task: Get talking to the Twitch API working again.
Problem: User has no server of their own to authenticate with.

On the Twitter embeds there's an edge-case scenario, if you post two links (or more) that are valid galleries but one is blocked from embedding but the other isn't then both will be shown anyway.

Neat feature to have.

A couple issues to be noted:

1. Both galleries are shown as fast as possible and it's not very API-friendly. Maybe there's a way to make things more efficient.
2. Now the galleries are both showing out of order. Some action should occur, like removing the original post and replacing it with a special case message to show them in order.

100% sure this happened as a side effect of the link parsing update.

Currently the bash script works, but it just barely does so. It's also not intuitive and pretty much only I know how to use it. Here's a list of things pending to do on this:

• Add documentation on the script usage.
• Fix the hit-or-miss remote python dependencies updater.
• Add "update" and "update locally" options.
  • The current implementation updates locally. It breaks when the user isn't in the same network. It should be capable to scan across the network and find the host by its host name.
  • The new update option simply pulls from the GitHub repository.

Cogs are a real mess right now. Not only do they not communicate to each other properly, but the gallery command shouldn't be allowed to be called manually unless it was done by itself.

We probably have to make some commands only work if the invoker is the bot itself but the question is... how?

They can be exactly the same, yes, but I don't think there's a viable way to check if they're different.

A window or separate application to view content would come in handy compared to editing a json file.