Just an idea I've had,...

A lot of functions, methods, and variables are typed incorrectly throughout the project. This issue serves as a tracker for correcting those types.

The incorrect types can cause a number of problems, including:

• Errors when the code is run.
• Difficulties in debugging the code.
• Difficulties in understanding the code.

This issue should be fixed to improve the quality of the code and to make it more maintainable.

I recently ran into a multipost message when I was making an announcement over several channels

If a user pastes their code without it being formatted in a code block, Pax should detect it and respond with the following: /tip Format Your Code

• Change /see-apps subcommand:spam to match all other subcommands in appearance (when empty).
• Remove introductory embed (before clicking ➡️ after using see-apps) instead introduce dummy command with example applications.
  • Dummy command should mirror real command, only differentiated by prefix dummy- and manipulating test-data
  • Dummy command should have at least one application in every possible state
  • Dummy command should reset to original test data with subcommand
  • [Optional] Dummy entries should detail what to be done with them i.e. by putting "Mark me as Spam" in the name field.
  • [Optional] Dummy entries should include instructions / what is expected to happen, or how to undo an action

Renaming:

• Rename Applicant name → Applicant Name
• Rename first name → First Name
• Rename Second Opinion required → Second Opinion Required
• Rename Time zone → Time Zone
• Rename Hours available per week → Hours Available per Week
• Rename Why staff? → Why do you want to become a staff member?
• Rename Why do you want to become a staff member → Why do you want to become a staff member?
• Rename Contribute reason → How will you contribute if you become a staff member?
• Rename How can you contribute if you are given staff → How will you contribute if you become a staff member?
• Rename Submission time → Submission Time
• Rename Like / Dislikes → Likes / Dislikes

If introductory embed is not removed generally:

• Remove introductory embed when using /see-apps specific_id

Consider:

• Renaming /see-apps into /see-applications OR /see-sapps (see staff apps)
• Introducing subcommand list in command to group options all, spam accepted denied

Might have some more when the dummy command is here, since we now only have a real application now and I can't test anymore...

As a part of scouting for users that meet staff requirements, we must save the amount of messages the user has across HWH's help channels.

Implement a /view-logs command that retrieves a user-specified number of logs from the log.txt file, so that developers can actively test and debug on production code that is difficult to replicate in testing environments.

Users are able to use /translate for the same language, which should not be an option.

Not sure if this has been talked about, but I would think you shouldn't be able to get an alert from your own message. I already know that I said it, so it doesn't benefit me to get a ping for it.

/apply-for-staff raises an error when used in direct messages due to the bot trying to access the joined_at attribute, which is not present in DM's.

A check should be added to prevent users from being able to use the command in DM's

We need a script that allows Homework Help staff to turn Pax Academia on and off remotely rather than relying on a single point of contact.

• /correct must have a 15 secodn cooldown [Requirement from !SKA]
• Due date: 31/10/2022

Required by staff team: Create a system that detects user messages being deleted in quick succession.

• Should generate a report showing messages deleted in the last hour, every hour, in increments of ten minutes.
• Determine control limits for anomaly criteria
• When the algorithm detects that a user surpasses the control limit, their roles will be cached, and the user will be muted
• Create a button that allows staff to unmute the individual in cases of false positives

Some env variables are crucial for the bot to function correctly. (bot token, deepl token, etc...)
When loading the tokens through os, an assert statement should follow to verify whether or not a correct variable is loaded.

(a check can be by length, it must be an int, etc...)

Pax should monitor activity in a channel and should be able to change slowmode based on activity, number of users, or a schedule.

• Monitor #sensitive-topics activity (perhaps in DB)

• -->(Query) How many messages have been sent in the past 20 minutes.

• -->(Query) How many messages have been sent in the past 20 minutes.

• -->(Query) Any other metric that might be useful

• Command to set slowmode lower and upper range /slowmode range <range-min> <range-max> (max range would be OFF to 6h)

• Command to schedule slowmode: /slowmode schedule <slowmode-mode [OFF-6h]> <duration active>* <start-delay>
  *should return to default after duration. <start-delay> is an optional parameter everything else should be required.

• Command to set default slow mode: /slowmode default <mode>*
  * Modes: OFF (no slowmode), no-interact (Pax does nothing), auto (Pax does stuff).

• Command to set current slow mode: /slowmode mode <mode> (modes described above)

Optional:

• Command to set criteria for "Auto" mode to increase slowmode
• Command to set criteria for "Auto" mode to decrease slowmode

Error messages are currently displayed as embedded messages within the user interface. This can be confusing for users, as they may not immediately understand that the message is an error and how to address it. In addition, these error messages can be visually cluttered and disruptive to the user experience.

Make all error messages in the Discord bot ephemeral. This will help to ensure that users are not confused by the messages and can focus on the main content of the bot.

• Ability to set list of ignored channel (categories)
• Ability to set list of ignored roles
• Ability to set list of ignored users (at least ignore self as an option)
• Ability to pause/suspend alerts for a period of time

Questionable if we should implement or not:

• Above abilities, but whitelist instead of blacklist (setting target channels, target roles, target users)

An error occurs when trying to send a direct message to someone who has their DM's turned off

Ignoring exception in on_message
Traceback (most recent call last):
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\client.py", line 377, in _run_event
    await coro(*args, **kwargs)
  File "C:\Production code\Pax-Academia\cogs\Alerts.py", line 211, in on_message
    await user_alerts()
  File "C:\Production code\Pax-Academia\cogs\Alerts.py", line 171, in user_alerts
    await user.send(embed=embed)
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\abc.py", line 1580, in send
    data = await state.http.send_message(
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\http.py", line 353, in request
    raise Forbidden(response, data)
discord.errors.Forbidden: 403 Forbidden (error code: 50007): Cannot send messages to this user

As Pax commands start becoming more experimental, and a little more invasive with on_message events, Some commands that are in early production might give false positives/be sent incorrectly.

An easy way to decorate a command as experimental should be created to ask active, verified members of the community to share their feedback with experimental commands. This feedback will be stored in the database and will be used by developers to evaluate their code before making it invasive/forced

Add a wrapper and database entry for users that misbehave and try to abuse the bot so that they cannot use any commands anymore.
People like this are the reason

Builds on #45: User requires the following aspects for helper roles. [Staff-only command]

• Time in server

  • (alternative condition if user left and rejoined the server)

• #164

• Certain amount of messages in any channel in a certain time frame.

• Account age

The exact requirements for Valued Contributors must not be publicised!
The information in question can be found in #vc-protocols and should also not show up in source-code.

Role requirements may be subject to debate or change.

Currently, if a keyword is triggered in another guild, a user will be alerted even if they aren't in that guild.

"The chief utility of the feature is that it alerts helpers as to when (and where) a question has been posted elsewhere, so they’re put on notice and don’t unwittingly waste their time working on a question that already may have been answered (this is also why I don’t think auto-deletion is necessary, in addition to the issue with false-positive deletions, which may prove especially problematic when people delete and repost a question in the correct channel). But it is true that Pax’s warnings can add up and become an eyesore. So what if we take a middle-ground approach? Just as people have been doing for a while now, multiposted questions can be flagged with a reaction by the bot indicating so; this would put the helpers on notice. And the warnings can be deleted after some time has passed. Additionally, perhaps some threshold—such as the same content being posted in more than two channels—can be put in place wherein all multiposted messages are deleted except for the most recent one.

The "3+and-delete" rule would add an extra layer of protection against the automated deletion of nonoffending messages and, combined with the timed auto-deletion of warnings, would work to suppress extraneous messages all around." -@woseek

Example:

#math-a: Original Message - Flagged (Upon first multipost)
#math-b: First multipost - Flagged and Warned
#math-c: Second multipost - Deleted
#math-d: Third multipost - Deleted
-@arborym

Builds on #41 : Additionally SAC should check if the user has

• At least 500 messages in any channel.

Also optionally:

• Send ?modlogs log of user in #google-form-submissions channel automatically upon (valid*) received application.
  • Wouldn't work if someone submitted a troll application.
  • Shouldn't work with staff account names.

Count Messages in channels, which are in Channel Categories that end in "Help" (or idk you could do smth more clever than that).

Detail saves the entire link when downloading an attachment

This can be used directly to identify persons or download content. EVEN if it was posted in a channel you do not have access to.

We should consider masking this one way or another, so that the full link isn't saved. I recommend keeping the HTTP POST/ GET but instead of the full link, only saving the filetype extension.

https://github.com/Arborym/Pax-Academia/blob/8290648b0cdf200c9af1a881a1f693715e8fbd5f/util/bandwidth.py#L25

Bot blocks and fails to respond to heartbeat, resulting in the bot losing connection and thus going offline on discord's end

User command which gives user link to the staff application form, if they meet the following criteria:

• Discord account age minimum 1 year.
• Server member for at least 30 days.

Edit:

• There should be a log of who requested the staff application form.

When a user never sent a message before (aka, not present in user table), and tries to use /apply-for-staff, the command tries to access a non-existing row in the table and raises an error

Move rules out of the code into a separate file.

There are two issues with our current /ask command:

1. Sometimes, users think that that is the command to ask a question. So they do /ask What's 2+2?, then never get a response.
2. It isn't anonymous, so whoever uses the command will almost assuredly be pinged once the user finally does ask their question. Just because I want the user to ask their question, doesn't mean I want to commit to answering it.

Me: /ask
Bot: @gabe used the /ask command: ask your question
Them: Okay what is 2 + 2?
Them: @gabe why aren't you answering my question?

Related to this thread in Discord

Every message that is sent triggers the Surveys log.
f"$ sent a survey in {message.channel.name}, bot responded".

This clutters the logs immensely as there is a message sent every two seconds.

The bot isn't allowed to send messages in some channels, which results in the following error:

Ignoring exception in on_message
Traceback (most recent call last):
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\client.py", line 377, in _run_event
    await coro(*args, **kwargs)
  File "C:\Production code\Pax-Academia\cogs\Surveys.py", line 34, in on_message
    await message.channel.send(
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\abc.py", line 1580, in send
    data = await state.http.send_message(
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\http.py", line 353, in request
    raise Forbidden(response, data)
discord.errors.Forbidden: 403 Forbidden (error code: 50013): Missing Permissions

Some messages exceed the 2000 character length of discord.

The current database.sqlite should be reworked to be a proper, relationship based db.
(Which is what I'm working on)

When a user sends two separate multiposts (with differing fingerprints) in the same channel, we can delete the original multipost message such that only the most recent one remains.

Relevant discord thread here.

If outputing two files at once is a problem, add a parameter to /view-db to choose which file format you want to have.

Traceback (most recent call last):
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\client.py", line 377, in _run_event
    await coro(*args, **kwargs)
  File "C:\Production code\Pax-Academia\cogs\Alerts.py", line 258, in on_message
    await user_alerts()
  File "C:\Production code\Pax-Academia\cogs\Alerts.py", line 204, in user_alerts
    await user.send(embed=embed)
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\abc.py", line 1580, in send
    data = await state.http.send_message(
  File "C:\Production code\Pax-Academia\venv\lib\site-packages\discord\http.py", line 359, in request
    raise HTTPException(response, data)
discord.errors.HTTPException: 400 Bad Request (error code: 50035): Invalid Form Body
In embeds.0.fields.0.value: Must be 1024 or fewer in length.

Make sure that the value of this field does not exceed 1024 characters in length.

Especially in raid situations outright banning all accounts which joined during a certain time will inevitably lead to a small number of false positives. Having available a possibility to mass-mute and mass-kick suspected accounts allows legit users to come back and ask to be unmuted.

In order to streamline this process, both a command /mmute for mass-muting and /mkick for mass-kicking as well as a combination of both, /mmutekick that first mutes and then kicks all users could be implemented. The syntax of the command should be compatible with the one used for Aperture at the moment.

Every time a user sends a direct message to the bot (not a command), the bot raises an error in messagecounter.py

Traceback (most recent call last):
  File "xxx\client.py", line 378, in _run_event
    await coro(*args, **kwargs)
  File "xxx\MessageCounter.py", line 31, in on_message
    category_name = message.channel.category.name.lower()
                    ^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'DMChannel' object has no attribute 'category'

Fix: exit (return) event (on_message) if message is a direct message

The database is called alerts.db and should be changed to database.sqlite

• rename database
• update references in alerts.py

Problem

The deepl.translate function is synchronous. This means that when someone uses the command, the entire bot halts all other activities (i.e. other commands and background tasks like moderation) and waits for the DeepL API to return the result. This may take a while (especially with rate limiting!), so we would really like this to be asynchronous.

Potential Solutions

1. pass the call off into another thread and asynchronously busy-wait for it to be completed. For example:

   import threading
   import asyncio
   
   # this is psuedocode (don't want to make this issue comment too long)
   # - the arguments to deepl.translate are more nuanced than this
   # - thread.result() is not a real function, we would need to use a mutable container
   # - it would be nice to raise exceptions from the thread
   async def translate(text, src, dst):
       thread = threading.Thread(deepl.translate, args=(text, src, dst))
       thread.start()
       while thread.is_running():
           await asyncio.sleep(0.1)
       return thread.result()

2. Use a different library that supports async calls,. I found deepl-tr-async, but it's an extremely heavyweight library and I don't really think it makes sense to add all of those dependencies.

3. A better solution that would take a lot more time would be to reimplement the deepl api (or fork it) using aiohttp to make the call itself asynchronous.

• When user submits an application on google form run /check-requirements and return result in column as yes/no
• When user submits an application check if user ran /check-requirements previously and return result in column as yes/no.

Additional (Optional Tasks

• When user submits an application on the google form check if user is (still) valid applicant.
  • User is in server.
  • User ID is not already a Staff member.
  • User doesn't currently have the Muted role.

Questions for staff form in Discord.

With the new discord username system going into place, the code should be adapted to incorporate these new names and slowly remove the discriminators.

At the moment, there are no errors raised by the program, but that could change in the near future.
Most affected are logging and staffapps which should be looked at.

Move filter word list in separate file for alerts.