Nextcloud Talk Bot API in Applications

The AppAPI is an excellent choice for developing and deploying bots for Nextcloud Talk.

Bots for Nextcloud Talk, in essence, don’t differ significantly from regular external applications. The functionality of an external application can include just the bot or provide additional functionalities as well.

Let’s consider a simple example of how to transform the skeleton of an external application into a Nextcloud Talk bot.

The first step is to add the TALK_BOT and TALK scopes to your info.xml file:

<scopes>
    <value>TALK</value>
    <value>TALK_BOT</value>
</scopes>

The TALK_BOT scope enables your application to register the bot within the Nextcloud system, while the TALK scope permits access to Talk’s endpoints.

In the global enabled_handler, you should include a call to your bot’s enabled_handler, as shown in the bot example:

def enabled_handler(enabled: bool, nc: NextcloudApp) -> str:
    try:
        CURRENCY_BOT.enabled_handler(enabled, nc)  # registering/unregistering the bot's stuff.
    except Exception as e:
        return str(e)
    return ""

Afterward, using FastAPI, you can define endpoints that will be invoked by Talk:

@APP.post("/currency_talk_bot")
async def currency_talk_bot(
    message: Annotated[talk_bot.TalkBotMessage, Depends(atalk_bot_msg)],
    background_tasks: BackgroundTasks,
):
    return Response()

Note

You must include to each endpoint your bot provides the Depends(nc_app).

Depending on nc_app serves as an automatic authentication handler for messages from the cloud.

message: Annotated[talk_bot.TalkBotMessage, Depends(talk_bot_app)] - returns the received message from Nextcloud upon successful authentication.

Additionally, if your bot can provide quick and fixed execution times, you may not need to create background tasks. However, in most cases, it’s recommended to segregate functionality and perform operations in the background, while promptly returning an empty response to Nextcloud.

An application can implement multiple bots concurrently, but each bot’s endpoints must be unique.

All authentication is facilitated by the Python SDK, ensuring you needn’t concern yourself with anything other than writing useful functionality.

Currently, bots have access only to three methods:

• send_message()

• react_to_message()

• delete_reaction()

Note

The usage of system application functionality for user impersonation in bot development is strongly discouraged. All bot messages should only be sent using the send_message method!

All other rules and algorithms remain consistent with regular external applications.

Full source of bot example can be found here: TalkBot

Wishing success with your Nextcloud bot integration! May the force be with you!