| Commit message (Collapse) | Author | Age | Lines |
|---|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
These 3 errors is:
- `ConversionError`
- `MaxConcurrencyReached`
- `ExtensionError`
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Added test for case when `Context.invoked_from_error_handler`
is `True`
|
| |
|
|
| |
Added `invoked_from_error_handler` attribute that is `False` default.
|
| |
|
|
|
| |
- Added `assertIsNone` to `test_error_handler_already_handled`.
- Removed `ErrorHandlerTests.cog`, moved it to each test recreation.
|
| | |
|
| |
|
|
|
|
| |
Created test that make sure when error is already handled in local error
handler, this don't await `ctx.send` to make sure that this don't move
forward.
|
| |
|
|
|
|
|
|
| |
As @mathsman5133 pointed out, it's better to use the `Command`-instance
we typically already have in the current context than to rely on parsing
the qualified name again.
The invocation is now done as: `await ctx.send_help(ctx.command)`
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
After the refactoring of the help command, we need to use the built-in
method of calling the help command: `Context.send_help`. As an argument,
the qualified name (a string containing the full command path, including
parents) of the command can be passed.
Examples:
- await ctx.send_help("reminders edit")
This would send a help embed with information on `!reminders edit` to
the Context.
- await ctx.send_help(ctx.command.qualified_name)
This would extract the qualified name of the command, which is the full
command path, and send a help embed to Context.
- await ctx.send_help()
This will send the main "root" help embed to the Context.
|
| | |
|
| |
|
|
|
|
| |
The `in_whitelist` decorator should not fail when a decorated command was called in a DMChannel; it should simply conclude that the user is not allowed to use the command. I've added a test case that uses a DMChannel context with User, not Member, objects.
In addition, I've opted to display a test case description in the `subTest`: Simply printing the actual arguments and context is messy and does not actually show you the information you'd like. This description is enough to figure out which test is failing and what the gist of the test is.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The API of the `in_whitelisted_context` decorator was a bit clunky:
- The long parameter names frequently required multiline decorators
- Despite `#bot-commands` being the defacto default, it needed to be passed
- The name of the function, `in_whitelisted_context` is fairly long in itself
To shorten the call length of the decorator, the parameter names were shortened by dropping the `whitelisted_` prefix. This means that the parameter names are now just `channels`, `categories`, and `roles`. This already means that all current usages of the decorator are reduced to one line.
In addition, `#bot-commands` has now been made the default redirect channel for the decorator. This means that if no `redirect` was passed, users will be redirected to `bot-commands` to use the command. If needed, `None` (or any falsey value) can be passed to disable redirection. Passing another channel id will trigger that channel to be used as the redirection target instead of bot-commands.
Finally, the name of the decorator was shortened to `in_whitelist`, which already communicates what it is supposed to do.
|
| |
|
|
| |
I have added tests for the new `in_whitelisted_context` decorator. They work by calling the decorator with different kwargs to generate a specific predicate callable. That callable is then called to assess if it comes to the right conclusion.
|
| | |
|
| |
|
|
|
|
|
|
| |
The `in_channel` decorator that served as a factory for `in_channel` checks was replaced by the broaded `in_whitelisted_context` decorator. This means that we can now whitelist commands using channel IDs, category IDs, and/or role IDs. The whitelists will be applied in an "OR" fashion, meaning that as soon as some part of the context happens to be whitelisted, the `predicate` check the decorator produces will return `True`.
To reflect that this is now a broader decorator that checks for a whitelisted *context* (as opposed to just whitelisted channels), the exception the predicate raises has been changed to `InWhitelistedContextCheckFailure` to reflect the broader scope of the decorator.
I've updated all the commands that used the previous version, `in_channel`, to use the replacement.
|
| |
|
| |
Replaced `TimeoutError` with `asyncio.TimeoutError`.
|
| |
|
|
|
|
| |
The "unsilence" action of the silence/hush command used `send_messages=True` when unsilencing a hushed channel. This had the side effect of also enabling send messages permissions for those with the Muted rule, as an explicit True permission apparently overwrites an explicit False permission, even if the latter was set for a higher top-role.
The solution is to revert back to the `Inherit` permission by assigning `None`. This is what we normally use when Developers are allowed to send messages to a channel.
|
| |\ |
|
| | | |
|
| | |
| |
| |
| |
| | |
Should return 1st arg (or None) if eval cmd in message, otherwise return
full content.
|
| | | |
|
| | |
| |
| |
| |
| | |
The tasks extensions loop requires an event loop to exist. To work
around this, it's been mocked.
|
| | | |
|
ks129
Sebastiaan Zeeff
MarkKoz
Sebastiaan Zeeff
Eivind Teig