Including a Unicode character in a JSON string object causes end }} to be removed

[404invalid-user]

example code:

OUTDOOR_TEMP_TOPIC = '/sensor/outdoor'
device = {
    "identifiers": "tempsensor1",
    "name": "TempSensor1",
    "manufacturer": "Raspberry Pi",
    "model": "Pi Pico W",
    "sw_version": "1.0"
}
outdoor_temp_payload = {
    "unique_id":"DHT11_outdoor_front_temperature",
    "name": "Outdoor Temperature",
    "state_topic": OUTDOOR_TEMP_TOPIC+'/state', 
    "unit_of_measurement": "\u2103", #remove this and it will be fine again
    "value_template": "{{ value | float }}",
    "device_class": "temperature",
    "device": device
}

async def descovery_pub(client):
    await client.publish(OUTDOOR_TEMP_TOPIC + "/config", ujson.dumps(outdoor_temp_payload), qos=1, retain=True)
    # confirm its not ujson.dumps
    print(ujson.dumps(outdoor_temp_payload))
    
    # simply adding +"}}" fixes it
    # await client.publish(OUTDOOR_TEMP_TOPIC + "/config", ujson.dumps(outdoor_temp_payload)+"}}", qos=1, retain=True)

look with mqtt explorer shows {"unit_of_measurement": "℃", "name": "Outdoor Temperature", "state_topic": "homeassistant/sensor/tempsensor1/outdoor_temperature/state", "device_class": "temperature", "unique_id": "DHT11_outdoor_front_temperature", "value_template": "{{ value | float }}", "device": {"identifiers": "tempsensor1", "manufacturer": "Raspberry Pi", "name": "TempSensor1", "model": "Pi Pico W", "sw_version": "1.0"

[peterhinch]

MQTT publications and subscriptions should be bytes objcts. This is easily fixed using string.encode and bytes.decode:

>>> s = "\u2103"
>>> p = ujson.dumps(s).encode()  # encode the entire json string
>>> p
b'"\xe2\x84\x83"'
>>> ujson.loads(p.decode('utf8'))  # Decode the bytes object from MQTT before running through JSON
'\u2103'

[404invalid-user]

Ah i see this fixed my issue thank you