moonrise55 / mbot Goto Github PK

Change title, tab colors at bottom, and color header / add answer

We should consider integrating an API hook for a collaborate whiteboard software. This wouldn't need to create a whiteboard for every puzzle, but puzzles could have commands to generate and link them. This has a few benefits:

• Normalized whiteboard use: no more having to learn Magma/Miro/Google Drawings/whatever else depending on who creates the first link. However, this is also a potential downside, if the tool we choose here goes bad and we need to switch it out.
• Centralized information flow: one M-Bot command can tell solvers whether there is a whiteboard for the puzzle, and if so, where it is. The single link can also be automatically pinned and transferred to the puzzle sheet. Once we achieve M-Board integration, M-Board can link the whiteboard as well.

Components:

• Collaborative whiteboard connector module, which contains the API integration required for whatever software we choose. Miro and Magma both have REST APIs, so they may be good places to start. I believe the team currently favors Miro.
• !whiteboard command (in the hunt.py cog), which checks whether the puzzle has a whiteboard and links the whiteboard if it does, or creates and links one if not. Creation of the whiteboard should also pin the link message and copy the link to a predetermined field in the puzzle sheet.
• -recreate flag, which creates and links a second whiteboard in the event that the first one becomes unusable

Components

• Automatically insert name into puzzle sheet A1
• If a website is set, try to make the round/puzzle URL and insert it into puzzle sheet A2

Additionally, we also might want:

• Add a -slug= flag to !create and !round for puzzles with slugs that aren't just their title
• Insert a reminder for solvers to check the URL in case it's still weird

Currently, the !login update flags are a mix of somewhat inconsistent abbreviations. We should support more abbreviation options, as well as single-letter flags like -u and -p.

• Make a separate !helporg command to summarize all the random housekeeping commands floating around. This include how to update !login, how to archive chat history, how to delete categories, how to check server stats.

• Clean the normal !help command to just be stuff that is relevant to all solvers during a hunt

• Proposed fix: "if !nexus is run within a round category, just display that round." and "if any !nexus command is too large, just link the sheet."
• We could also consider a #nexus channel (for bighunt mode) that paginates the puzzles in each round

Currently, several easter egg parses (thanks, hello) only respond to mbot. They should respond to m-bot as well.

After the bot has been running for a bit and many puzzles have been created/solved, it starts hitting the 60-second request limit for the Google Sheets API extremely frequently, even in situations where it reasonably shouldn't (i.e. there has been excessive bot activity in the past 10 minutes, but not in the past 1 minute). Presumably this is due to some kind of hanging process or request that we're not closing out properly. We should figure out what's causing this and avoid it, if possible.

If an integration is specified for a hunt, M-Bot should be prepared to receive events from M-Board (over an API or a websocket) in order to update the Discord server as usual.

Interacts with azwong42/M-Board#5

Checklist for additions to the !graph command:

• Mark metas (points or lines)
• Include team name in graph title (will require name to be databased)
• Toggle flag for elapsed hours vs. clock time
• -stop-at flag to generate a graph from hunt start to N hours

• We should store links to puzzles in the Nexus sheet and the puzzle spreadsheet.
• Puzzle creation should attempt to guess the website slug based on the puzzle name, and should have an additional -slug=<slug> flag to set the slug if the name doesn't match.
• !update should also support -slug=<slug>

We can use !login to set the hunt role to none, but doing this in ~!createhunt` throws an error

There are archive commands for either a channel or a category to save all conversation to a google sheet linked. Currently they just send a link to any images found. I have a branch where the commands download the images to a local folder instead.

1. revive that branch and make the commands accept a local folder location to save images so anyone can use it
2. stop sending a message for every channel archived. Gets annoying for an entire category. Probably just pass a status message object around instead.

Solving puzzles drops them 2 spaces below the most recent green checkmark. Presumably this is either an issue with channel.move() or the way we're counting category channels.

This would be cool to have, so we don't have to make them manually. Possible options:

• Generates a graph for the current hunt
• Generates a graph for any hunt given a Nexus link
• Organiser only?

Goal is to add some features that might be useful for a multi-round hunt. Namely sorting puzzles into categories by round.

1. command to create a new round
   will make a new category, text and voice channel, add category id to a table in the nexus

2. command to create a new puzzle will now

• if -round given
  [if round exists] sort puzzle into correct category and order nexus appropriately
  [if round does not exist] fail

• if no -round given
  [if current category is a round] assume current category is the correct round and create puzzle there
  [if current category is not a round] fail

2.1) command to move puzzle across rounds?

3. a log channel that collects all puzzle creation/solve times

4. update help menu

As before to minimize black box database storage, the nexus google sheet will be the true state. Except hopefully you won't need to sort puzzles into rounds yourself. Theoretically no editing of the nexus sheet will be needed.

Round category ids will be stored on a new tab in the nexus file (or in the far right of the nexus sheet to minimize sheet importing)
Log channel will be hardcoded in the cog or maybe make another entry for the login db.

Since we might want to revert the changes back after mystery hunt, it's probably easiest to work from a copy of the puzzle.py cog and call it bighunt.py. Then should be a simple matter of swapping out the active cog to revert changes. Should also rename hunt.py -> login.py and puzzle.py -> hunt.py for accuracy.

Currently, every puzzle-related command needs to make:

• A database request (to get the nexus link)
• (if multihunt) another database request (to get the nexus link)
• A read request for the nexus sheet (to retrieve puzzle data for the channel)
• Whatever else the command is doing

If we cache puzzle data whenever we run an operation on a puzzle (creation, note, solve), we can avoid all three of the overhead requests for a reasonable number of puzzles. We already have LRU cache logic set up for voice channel IDs; this shouldn't be all that different.

Just in case

Add a -meta option to !create to automatically mark a puzzle as a meta in the nexus. This will also make the puzzle creation workflow integrate more easily with M-Board

If an integration is specified for a hunt, every puzzle-related M-Bot command should optionally call a corresponding M-Board API endpoint to ensure the information makes it to the website.

Interacts with azwong42/M-Board#4

• reduce likelihood of duplicate/lingering bugs
• easier to share new features across modes
• more flexibility for docs, etc

Note: this only occurs when not in bighunt mode.

Probably just add it to the !login database

Ignore sheet coloring if no MAIN

Potential issues: how to split the puzzle names?

A list of improvements for the multihunt management workflow

• Hide folder access in original command (or figure out a way to not require the folder to be linked)
• login update -role should update category permissions
• Store bighunt status in database per-hunt (and add a bighunt version of !createhunt)
• Update !help again