Fire zha_event from Custom Manufacturer cluster?

[mathoudebine]

Hello,
I'm writing a custom quirk for a Zigbee remote control for light bulbs for #889.

For the 4 "preset buttons" of the remote, it uses a manufacturer-specific cluster with ID 0xFE00: when one of the preset button is pressed, the remote sends a Zigbee client command at 0x000 with two uint8_t as parameters.

To manage these buttons, i wrote a custom cluster for my quirk that handles the request and sends a zha_event each time a button is pressed, to be able to use device_automation_triggers for these preset buttons.

class AdeoPresetCluster(EventableCluster):
    """Custom cluster for preset buttons 1-4"""

    cluster_id = MANUFACTURER_SPECIFIC_CLUSTER_ID_PRESET
    name = "AdeoPresetCluster"
    ep_attribute = "adeo_preset_cluster"
    manufacturer_client_commands = {0x000: ("preset", (t.uint8_t, t.uint8_t), False)}

    def handle_cluster_request(
        self,
        hdr: foundation.ZCLHeader,
        args: List[Any],
        *,
        dst_addressing: Optional[
            Union[t.Addressing.Group, t.Addressing.IEEE, t.Addressing.NWK]
        ] = None,
    ):
        """Send cluster client requests as events."""
        if (
            self.manufacturer_client_commands is not None
            and self.manufacturer_client_commands.get(hdr.command_id) is not None
        ):
            self.listener_event(
                ZHA_SEND_EVENT,
                self.manufacturer_client_commands.get(hdr.command_id)[0],
                args,
            )
            self.debug(
                "AdeoPresetCluster : fire event {action=%s, args=%s}",
                self.manufacturer_client_commands.get(hdr.command_id)[0],
                str(args),
            )
        else:
            self.debug("AdeoPresetCluster : NO event fired")
            super().handle_cluster_request(hdr, args, dst_addressing=dst_addressing)

Full quirk code is available here: https://github.com/mathoudebine/zha-device-handlers/blob/adeo-lxek-5/zhaquirks/adeo/lxek5.py

Quirk and cluster are loaded without error by ZHA after HA restart.
When i press one of the button, the HA logs indicate that the command is handled by the cluster:

2022-01-26 19:04:02 DEBUG (MainThread) [bellows.ezsp.protocol] Application frame 69 (incomingMessageHandler) received: b'00040100fe01014000000042ffc16665ffff070d771260000d01'
2022-01-26 19:04:02 DEBUG (MainThread) [bellows.zigbee.application] Received incomingMessageHandler frame with [<EmberIncomingMessageType.INCOMING_UNICAST: 0>, EmberApsFrame(profileId=260, clusterId=65024, sourceEndpoint=1, destinationEndpoint=1, options=<EmberApsOption.APS_OPTION_RETRY: 64>, groupId=0, sequence=66), 255, -63, 0x6566, 255, 255, b'\rw\x12`\x00\r\x01']
2022-01-26 19:04:02 DEBUG (MainThread) [zigpy.zcl] [0x6566:1:0xfe00] ZCL deserialize: <ZCLHeader frame_control=<FrameControl frame_type=CLUSTER_COMMAND manufacturer_specific=True is_reply=True disable_default_response=False> manufacturer=4727 tsn=96 command_id=0>
2022-01-26 19:04:02 DEBUG (MainThread) [zigpy.zcl] [0x6566:1:0xfe00] ZCL request 0x0000: [13, 1]
2022-01-26 19:04:02 DEBUG (MainThread) [zigpy.zcl] [0x6566:1:0xfe00] AdeoPresetCluster : fire event {action=preset, args=[13, 1]}

I can see my custom log AdeoPresetCluster : fire event but no zha_event is fired despite calling self.listener_event(ZHA_SEND_EVENT,[...]) from my custom cluster.

When I press on a button managed by generic clusters like LevelControl, the zha_event is sent as expected:

2022-01-26 19:03:57 DEBUG (MainThread) [bellows.ezsp.protocol] Application frame 69 (incomingMessageHandler) received: b'000401080001014000000040ffcb6665ffff09015e02001a05000000'
2022-01-26 19:03:57 DEBUG (MainThread) [bellows.zigbee.application] Received incomingMessageHandler frame with [<EmberIncomingMessageType.INCOMING_UNICAST: 0>, EmberApsFrame(profileId=260, clusterId=8, sourceEndpoint=1, destinationEndpoint=1, options=<EmberApsOption.APS_OPTION_RETRY: 64>, groupId=0, sequence=64), 255, -53, 0x6566, 255, 255, b'\x01^\x02\x00\x1a\x05\x00\x00\x00']
2022-01-26 19:03:57 DEBUG (MainThread) [zigpy.zcl] [0x6566:1:0x0008] ZCL deserialize: <ZCLHeader frame_control=<FrameControl frame_type=CLUSTER_COMMAND manufacturer_specific=False is_reply=False disable_default_response=False> manufacturer=None tsn=94 command_id=2>
2022-01-26 19:03:57 WARNING (MainThread) [zigpy.zcl] [0x6566:1:0x0008] Data remains after deserializing ZCL frame
2022-01-26 19:03:57 DEBUG (MainThread) [zigpy.zcl] [0x6566:1:0x0008] ZCL request 0x0002: [0, 26, 5]
2022-01-26 19:03:57 DEBUG (MainThread) [zigpy.zcl] [0x6566:1:0x0008] No handler for cluster command 2
2022-01-26 19:03:57 DEBUG (MainThread) [homeassistant.core] Bus:Handling <Event zha_event[L]: device_ieee=60:a4:23:ff:fe:70:39:a0, unique_id=60:a4:23:ff:fe:70:39:a0:1:0x0008, device_id=0501c9e8666648ccd073ffc9773f2962, endpoint_id=1, cluster_id=8, command=step, args=[0, 26, 5]>

What am I missing here? Is it not possible to fire zha_event from manufacturer clusters ?
Thanks for your help

[MattWestb]

Hi there i was having the same problem with one new IKEA controller but was getting it working OK with little trikes getting events from manufacture clusters.
Take one look in these issue #2223 (comment).

[mathoudebine]

Hi @MattWestb thanks for the info, I managed to do it differently by sending the event from my manufacturer cluster to a standard Scenes cluster that will fire it correctly:
https://github.com/zigpy/zha-device-handlers/blob/dev/zhaquirks/adeo/color_controller.py#L64

[MattWestb]

For my i think the class EventableCluster shall being refactored in ZHA so it sending zha_event in one normal way from one manufacture cluster and dont need trixing and making strange things.
Also you can having one cluster type one time on one endpoint and if you already is having one scene cluster on it you cant add one more without braking things.