@Odianosen25 you would need to do a git checkout and merge which is a bit of a bear and would get overwritten next upgrade if it doesn’t get merged. An easy cheat if you want to test is to do a git clone and copy these 4 files
Into wherever your home assistant libs are, mine is /srv/homeassistant/lib/python3.6/site-packages/homeassistant/helpers with the standard virtualenv install. Docker or hassio you are on your own!
Not sure what you mean about json payload since that is one. Here is an example of literally just sending the entire intent to appdaemon
automations.yaml
INFO jarvis: jarvis_intent_not_parsed: {'payload': '{"sessionId":"6bbd31e1-8168-4ee0-bdcc-20073c313b10","customData":null,"siteId":"default","input":"turn christmas lights on","intent":{"intentName":"user_rZEKlr8M7M3__lightsTurnOnSet","probability":0.7173734},"slots":[{"rawValue":"christmas","value":{"kind":"Custom","value":"christmas"},"range":{"start":5,"end":14},"entity":"house_room","slotName":"house_room"}]}'}
With this, the intent was parsed properly in HASS (my christmas lights came on) and the full json intent was also passed to app daemon as you can see in the log.
But why go through the extra step of putting data into sensors to get it into AD? Parsing the data in yaml and mapping values is a pain and much easier to do in AD IMO. Plus then you have extra sensors cluttering things up.
And yeah, I could write an app to listen to mqtt events as well and had one running but why have another daemon when HA is already there? And there is also the benefit as you say of having the data split.
A use case for that is that I will do will be having HASS/AD/Snips be able to initiate conversations off of events, and to do so I need the full data that comes across MQTT. And again, yeah I could have a separate listener.
Yep that was my initial thought, until I figured out there was no way to listen for mqtt messages which @gpbenton did confirm was true. That was the only way I could get around the issue so it wasn’t intentional as per say. if you know otherwise, pls advise as I had tried this
But nothing worked even though the event trigger was triggering in yaml on the same topic.
The next option was to use custom event which I couldn’t until you came up with this template of event data. My use case is actually to communicate with snips over chatbot on telegram and create like a little “chatbot” for the home (as I refused to expose my system to the internet due to some limitations and other stuffs). For this I needed full access to the data on mqtt.
I also tried implementing a listener in appdaemon and it would start and connect but I couldn’t get the loop to run. But that felt like a kludge anyway since I have a perfectly good listener in hass.
Well at least you got the skill to develop the event_data_template. I am no programmer so was stuck . Will test your software and get back to you if any issues.
maybe my failing understanding from MQTT.
i just see it as entity communication. so for me the raw data is just something that needs to be translated to entities.
i see not much value from pulling raw data through my HA only to give it to something else.
then again, i dont also think of sensors cluthering things up, but as something to give me a view on things.
in AD i create sensors like last motion, last value changed, etc to view things in 1 view, without going through logs or history.
Oh sensor are perfect for something I want to keep a state of but a lot of what I will do is just pass one time info to AD in which case having a sensor isn’t a good fit since it is a static thing.
But everyone has their own ideas on how to implement things so having more options is always better.
Well I finally solved my issue by coping a cue from how the Alexa and Google API is handled. I created a python script that subscribes to the topic of interest, and made it a service so it starts at boot up. This then sends the data to AD using “requests.post”. In AD I just registered an endpoint and I get all the raw data without getting HA involved.
I think this helps me with extra flexibility, without using sensors and making it independent of HA is actually a plus as I don’t need to cluster it just because I need to pass data.
Kind regards everyone for all the help and suggestions.
Hi, @Odianosen25 can you give me more details or is there event the code of your python script available? Working on the same thing and this would save a lot of time
Thank you
Find below my script that sends the data from Snips to AppDaemon. I don’t send data from AppDaemon to Snips yet as I have no need of it for the mean time but have a script I could use for that also.
Basically I modified the handler I saw in this example project. I am parsing the data in the script, so a “clean” set of data is sent to AppDaemon instead of doing all that in AppDaemon. Snips runs on a different Pi from that of AppDaemon and HA. Also I have data being sent from Telegram to Snips to make use of its NLU without ASR, so I can use it to make the same queries over chat and that’s why its busier than it needs to. You can modify and remove the unneeded bits.
#!/usr/bin/python
from __future__ import unicode_literals
import datetime
import json
import paho.mqtt.client as mqtt
import requests
import re
fromtimestamp = datetime.datetime.fromtimestamp
# MQTT client to connect to the bus
mqtt_client = mqtt.Client()
HOST = "YOUR HOST"
PORT = 1883
TOPICS = ["hermes/nlu/intentParsed", "hermes/hotword/default/detected", "hermes/nlu/intentNotRecognized"]
URL = "http://localhost:5000/api/appdaemon/snips"
API_TOKEN = "AppDaemon API KEY"
header = {'content-type': 'application/json', 'x-ad-access': '{}'.format(API_TOKEN)}
MQTT_USERNAME = "YOUR MQTT BROKER USERNAME"
MQTT_PASSWORD = "YOUR MQTT BROKER PASSWORD"
USERS = ["USER1", "USER2", "USER3"]
global site
# Subscribe to the important messages
def on_connect(client, userdata, flags, rc):
print("AppDaemon-Snips Bridge Connected to Broker") # for debug reasons
for topic in TOPICS:
mqtt_client.subscribe(topic)
# Process a message as it arrives
def on_message(client, userdata, msg):
print(msg.topic) # for debug reasons
print(msg.payload) # for debug reasons
data = parse_data(msg)
if data == "default":
return
slots = parse_slots(msg)
for k,v in slots.items():
data[k] = v
print(json.dumps(data)) # for debug reasons
try:
requests.post(URL, data = json.dumps(data) , headers = header)
except:
raise Exception("Error When sending request")
print("Error When sending request")
def parse_data(msg):
global site
data = {}
if msg.topic == TOPICS[1]:
rawData = json.loads(msg.payload)
site = rawData["siteId"]
data = site
return data
rawData = json.loads(msg.payload)
intent = rawData['intent']['intentName']
value = re.findall("[a-zA-Z0-9]+", intent)
data['intent'] = value[len(data) - 1]
if rawData['id'] in USERS:
data['id'] = rawData['id']
else:
data['id'] = site
data['sessionId'] = rawData['sessionId']
return data
def parse_slots(msg):
'''
We extract the slots as a dict
'''
data = json.loads(msg.payload)
try:
slot_data = {slot['slotName']: slot['value']['value'] for slot in data['slots']}
except:
slot_data = {}
return slot_data
if __name__ == '__main__':
mqtt_client.on_connect = on_connect
mqtt_client.on_message = on_message
mqtt_client.username_pw_set(MQTT_USERNAME, MQTT_PASSWORD)
mqtt_client.connect(HOST, PORT)
mqtt_client.loop_forever()
In AD I have this to call my function when a message is received self.register_endpoint(self.api_interface, "snips") and it calls my function def api_interface(self, data).
. Will test your software and get back to you if any issues.
