PySLOBS: A Python Wrapper for the StreamLabs Desktop API
-------------------------------------
## About the API
[Streamlabs Desktop](https://www.streamlabs.com) (formerly StreamLabs OBS) is a live streaming application
based on a fork of [Open Broadcaster Software](https://obsproject.com) with additional
features.
It offers the
[Streamlabs Desktop API](https://github.com/stream-labs/streamlabs-desktop-api-docs) to allow
third-party applications to interact with the application while it is running.
This includes selecting, editing and monitoring scenes, video sources and audio
sources. It includes monitoring performance.
This doesn't include chat or getting notifications about donations, subscriptions and
followers. You will need to look elsewhere for that.
Typically, it would be accessed from same computer that is running Streamlabs Desktop, or
from a computer on the same LAN.
The API is based on [WebSockets](https://en.wikipedia.org/wiki/WebSocket) so it can
be accessed from a browser in JavaScript. (Apparently, it can also be accessed through
a named pipe.)
## About the Python Wrapper
Rather than accessing it from Javascript, this Python wrapper allows access to
Streamlabs Desktop from a Python application. The details about websockets are hidden
and more Pythonic interfaces are provided.
### Versions
This API requires Python 3.9 or later. It is tested on Python 3.10.
(If there is a good reason you need an earlier version of Python,
please raise a GitHub issue.)
Streamlabs Desktop versions from 1.2.0-1.10.0 have been tested.
### Pythonic names and types
The Python interface is designed to allow you to write
[PEP8-compliant](https://www.python.org/dev/peps/pep-0008/) Python code.
Camel-case items in the original API are represented in their preferred PEP8 form -
i.e. underscores between words to avoid ambiguity and capitalised constants.
Enumerated types and named tuples are used in preference to JSON dictionaries and
numeric constants where possible.
Identifiers that clash with reserved words used in Python are suffixed with `_` - e.g.
`id_()`.
Date-times from notifications do not have a timezone associated with them. They are in
the timezone of the host running the API.
### Testing Progress.
The API is moderately large. In this version of the Python wrapper, not every method
offered has been tested. Testing has been focussed on the areas the developer actively
wanted to use first.
See `PROGRESS.md` for an idea of what is and isn't tested.
### Usage
#### Calling asyncio libraries
This Python wrapper is based on `asyncio`. If you have not used the `asyncio` features
of Python before, please consult a tutorial. In particular:
* You need to `await` all methods defined as async. Similarly, context managers defined
as async need to be called with `async with`.
* After creation, remember to await the `SlobsConnection.background_processing()`
coroutine.
* To shut-down cleanly, remember to close the connection.
* To avoid exceptions being swallowed, handle them in your top-level code.
See the example code.
#### Connection
First, you need a `SlobsConnection` instance.
Connections actively process received messages, so it is important that they
be included in the `asyncio` event loop.
So your main program might look like this:
import asyncio
import logging
from pyslobs import connection
async def main():
# Note: This assumes an INI file has been configured.
conn = connection.SlobsConnection()
# Give CPU to both your task and the connection instance.
await asyncio.gather(
conn.background_processing(),
do_your_thing(conn)
)
logging.basicConfig(level=logging.DEBUG)
asyncio.run(main())
Connections should be closed when complete, so the background processing can
know to shut-down.
#### Configuration and Authentication
The SlobsConnection needs details about how to connect to the Streamlabs Desktop
application.
It needs to know which host is running the application, which port it is listening to,
and the "API Token".
The API Token is like a password. It is generated by the application and
shared with any tools that want to talk to it. It should not be shared with others; if
it is, a new token should be generated.
##### Obtaining connection details
To obtain all these details, start Streamlabs Desktop, open the settings page, and select
"Remote Control". A (blurry) QR code will be displayed. Do not show that QR code on
your stream.
If you click on the QR code, and click "Show Details" underneath it, you will be given
the following details:
* API token
* Port: This should default to 59650. If it is a different value, your Python
application will need to know this.
* IP addresses: If your application is on a different host to Streamlabs Desktop, your
Python application will need to know one of the IP addresses offered.
##### Using connection details
These details can be provided by directly instantiating a `ConnectionConfig`, and
passing it to `SlobsConnection()`.
Alternatively, if none is passed, the SlobsConnection constructor will look for their
presence in INI files. The INI file should be stored in the current directory or the
user's home directory. They should be called `.pyslobs` or `pyslobs.ini`.
The content of the ini file should be:
[connection]
domain=localhost
port=59650
token=<your secret token>
When running the examples or exercises, if no ini file is found, it will assume defaults
and prompt for the user to type in the API token each time.
#### Services
Once you have a connection, it can be used to instantiate any of nine Services:
1. AudioService
1. NotificationsService
1. PerformanceService
1. SceneCollectionsService
1. ScenesService
1. SelectionService
1. SourcesService
1. StreamingService
1. TransitionsService
Services can be used:
* subscribe to events, such as when the user selects a new active scene.
* to make some limited direct changes, such as setting an active scene by
scene id.
* fetch SlobsClass instances (see below) that can be manipulated more directly.
#### Classes
In the original API they describe "Classes". These are represented by subclasses
of `SlobsClass` in the Python API.
SlobsClass subclasses include:
1. AudioSource
1. Scene
1. SceneItem
1. SceneItemFolder
1. SceneNode
1. Selection
1. Source
Instances of SlobsClass should only be constructed through Services methods and methods
on other instances.
Instances may have properties (such as names) that can be accessed. Be careful that the
values of these properties may be out-of-date if the value was changed within the app
or if it was changed through a different instance referencing the same StreamLabs Desktop resource.
Objects can be used to fetch other Objects or `namedtuple`s describing other records
in the API.
#### Subscriptions
Some `Services` offer the ability to subscribe to events. A list is provided
below.
To subscribe, create a callback function with a signature like:
async def on_event(key, message)
(A method could also be used, in which case there should be an additional `self`
parameter.)
Then call the event's `subscribe()` method, passing the callback function. e.g.
subscription = ScenesService(conn).scene_switched.subscribe(on_event)
Each time the event happens, the callback will be called. The `key` parameter
will indicate which sort of event has triggered the callback (which is useful if you
want to reuse a call-back function for several event types). The `message` parameter
is a dictionary containing details of the event. (The StreamLabs Desktop API is unclear about what
fields will be present, so this may require some experimentation to find out what
data is available.)
The result of subscribe is a `Subscription` object that supports `unsubscribe()`.
(It can also be used as an asynchronous context manager to ensure unsubscriptions
are not forgotten.)
The subscribe method also accepts a `subscribe_preferences` parameter which is an
instance of `SubscriptionPreferences`. It indicates whether the callback should
*also* be called when the subscription is ended due to a call to unsubscribe or
by the connection being closed. In these events, the value of key will be set to
special values: `UNSUBSCRIBED` and `CLOSED` as appropriate.
##### Subscribable Events by Service
* SceneCollectionService
* collection_added
* collection_removed
* collection_switched
* collection_updated
* collection_will_switch
* ScenesService
* item_added
* item_removed
* item_updated
* scene_added
* scene_removed
* scene_switched
* Sources Service
* source_added
* source_removed
* source_updated
* StreamingService
* recording_status_change
* replay_buffer_status_change
* streaming_status_change
* Transitions Service
* studio_mode_changed
## Examples:
The examples folder contains many small programs to demonstrate how to use the
API. These are not included with the package, but can be found on the GitHub
site. Install PySLOBS, copy the raw example files to your machine, start your
copy of Streamlabs Desktop, and run the examples (once PySLOBS is installed).
## Special cases:
* `Sources` have an additional field `configurable` which isn't documented.
* `Sources` have a method `refresh` which raises an Internal Server Error. This has been
reported to Streamlabs.
* `SceneCollectionsService` methods sometimes raise Internal Server Errors if called
too rapidly.
* `TransitionsService` methods sometimes raise Internal Server Errors if called
too rapidly.
* Notifications subtype may be `NEWS`, which is not mentioned in the documentation.
* When a source is added, the `source_updated` callback may also be triggered an arbitrary number of times.