Python typings of all payloads that Discord sends as TypedDict
s.
The library requires no setup except for installation. The easiest way to
install it is through PyPI where it is similarly named discord-typings
:
pip install discord-typings
Now, start importing the relevant typings directly from the library:
from discord_typings import UserData
def print_author(user: UserData) -> None:
print(f"Created by {user['username']}#{user['discriminator']}")
print_author({
'id': '344404945359077377',
'username': 'Bluenix',
'discriminator': '7543',
'avatar': None
})
It is also possible to import the library inside of a TYPE_CHECKING
-clause:
from typing import Any, TYPE_CHECKING
if TYPE_CHECKING:
from discord_typings import Snowflake
class Data:
id: 'Snowflake'
value: Any
Note
It is not recommended to import the library this way, as it does not allow you to introspect the annotations in other code. It is merely pointed out for completeness.
There is no documentation or API reference as this provides no value on top of the Discord developer documentation the types are based on. Typings are named after the object they represent, and there are little to no exceptions to this rule.
Typings use suffixes to differentiate between potential wrapping objects in
user code - such as importing UserData
to parse into a custom User
class.
These are Data
, Event
, and Command
. Event
refers to a gateway event
received over the gateway while Command
refers to a gateway command sent
over the gateway. Data
is used for any general objects like UserData
.
To differentiate between the data for complete application commands, and the
data Discord expects to receive to create an application command, there is
a special-cased ApplicationCommandPayload
.
Discord-typings is a relatively simple library with little code, but unfortunately needs to be constantly maintained with Discord's API. The purpose of this library is for this to be a community effort; any help with maintenance is greatly appreciated.
The library follows the API docs both in terms of naming, structure, and order inside of individual files. This should hopefully make it easier to find the code to change when you have the documentation entry at hand.
All internal imports should be done by importing the entire module, then
accessing the typing as an attribute. If from ... import ...
, or specific
things are imported, that is likely to create complicated circular imports.
Inside of annotations, wrap the attribute access (discord_typings.X
) in
quotes to make it a string and defer its evaluation.
All typings which directly represent a Discord payload (excluding typings
created for the purpose of a Union
) should be added to the file's __all__
.
Files should be created with an underscore prefix, as direct imports should be
avoided. Instead, only the top-level namespace discord_typings
should be
used. This means that each file should have a line in the module's __init__
.
Once discord-typings releases its v1.0
release, it will follow strict
semantic versioning guarantees. Keep in mind that this only applies
to the Python interface - such as TypeDict
s, top-level unions, and
other aliases.
Due to Discord's frequent API updates, it is not guaranteed that code which type-checks in one minor version will do so in another one. This is because code which does not type-check will not have an effect on runtime for users.
As a reminder of semantic versioning, and a summary of the above:
Release Type | Upgrade | Downgrade | Comment |
---|---|---|---|
Major | Only type of release with breaking Python changes | ||
Minor | X | May add new features, can break type-checking | |
Patch | X | X | Only intended for bugs |
Warning
Patch versions will be used to rectify any accidental breaking changes or unintended bugs / behaviour, therefore you should always use the latest patch version.
Note
Because patch versions may change previous behaviour, they could be considered breaking, however the intention is always to fix unintended behaviour or previous breaking changes which should not have been. In those cases, the previous version will be pulled from PyPI.