The MQTT extension for Cognite extractor-utils provides a way to easily write your own extractors for systems exposing an MQTT interface.

The library is currently under development, and should not be used in production environments yet.

The MQTT extension for extractor utils subscribes to MQTT topics, automatically serializes the payload into user-defined DTO classes, and handles uploading of data to CDF.

The only part of the extractor necessary to for a user to implement are

• Describing the payload schema using Python dataclasses
• Implementing a mapping from the source data model to the CDF data model

As an example, consider this example payload:

{
    "elements": [
        {
            "pumpId": "bridge-pump-1453",
            "startTime": "2022-02-27T12:13:00",
            "duration": 16,
        },
        {
            "pumpId": "bridge-pump-254",
            "startTime": "2022-02-26T16:12:23",
            "duration": 124,
        },
    ]
}

We want to make an extractor that can turn these MQTT messages into CDF events. First, we need to create some data classes describing the expected schema of the payloads:

@dataclass
class PumpEvent:
    pumpId: str
    startTime: str
    duration: int

@dataclass
class PumpEventList:
    elements: List[PumpEvent]

Then, we can create an MqttExtractor instance, subscribe to the appropriate topic, and convert the payload into CDF events:

extractor = MqttExtractor(
    name="PumpMqttExtractor",
    description="Extracting pumping events from an MQTT source",
    version="1.0.0",
)

@extractor.topic(topic="mytopic", qos=1, response_type=PumpEventList)
def subscribe_pump_events(events: PumpEventList) -> Iterable[Event]:
    external_id_prefix = MqttExtractor.get_current_config_file()

    for pump_event in events.elements:
        start_time = arrow.get(pump_event.startTime)
        end_time = start_time.shift(seconds=pump_event.duration)

        yield Event(
            external_id=f"{external_id_prefix}{pump_event.pumpId}-{uuid.uuid4()}",
            start_time=start_time.int_timestamp*1000,
            end_time=end_time.int_timestamp*1000,
        )

with extractor:
    extractor.run()

A demo example is provided in the example.py file.

See the contribution guide for extractor-utils for details on contributing.