104 lines
3.5 KiB
Markdown
104 lines
3.5 KiB
Markdown
---
|
|
layout: page
|
|
title: "Run local scripts"
|
|
description: "Instructions on how to run a local script for Home Assistant."
|
|
date: 2017-04-30 13:28
|
|
sidebar: true
|
|
comments: false
|
|
sharing: true
|
|
footer: true
|
|
---
|
|
|
|
Hass.io is a managed environment, which means that you can't install applications that can be embedded into Home Assistant using the `command_line` sensor/switch.
|
|
|
|
There are three options if you need to run a script to read data from a sensor or send commands to other devices on Hass.io.
|
|
|
|
The first option is to write a custom component for Home Assistant. Using Python, you can communicate with your device. For more information about developing a custom component, take a look at the [developer documentation][custom-component].
|
|
|
|
The second option is to use STDIN inside add-on and use the service `hassio.addon_stdin` to send data. More about this options look into [developer documentation][communication] for internal add-on communication. There is also describe how do you can easy access to Home-Assistant Rest API.
|
|
|
|
The third option is to make a local add-on for Hass.io that sends the data to Home Assistant via MQTT. Before we dive into this, read up on [Hass.io add-on development][addons-tutorial] first.
|
|
|
|
For security and speed, Hass.io does not provide a way for containers to communicate directly. So the first step is to set up a communication channel. We're going to use MQTT for this using the [MQTT broker add-on][mqtt-addon].
|
|
|
|
### {% linkable_title Sensors %}
|
|
|
|
We loop in our script to fetch data and push it to MQTT and wait until the next process is ready. Here is a basic example and structure for that process.
|
|
|
|
Our Dockerfile need to install:
|
|
|
|
```
|
|
RUN apk --no-cache add jq mosquitto-clients
|
|
```
|
|
|
|
Now we can process it with `run.sh`:
|
|
|
|
```bash
|
|
#!/bin/bash
|
|
set -e
|
|
|
|
CONFIG_PATH=/data/options.json
|
|
|
|
# possible options for processing
|
|
MQTT_SERVER=$(jq --raw-output '.server' $CONFIG_PATH)
|
|
MQTT_PORT=$(jq --raw-output '.port' $CONFIG_PATH)
|
|
TOPIC=$(jq --raw-output '.topic' $CONFIG_PATH)
|
|
USER=$(jq --raw-output '.user' $CONFIG_PATH)
|
|
PASSWORD=$(jq --raw-output '.password' $CONFIG_PATH)
|
|
WAIT_TIME=$(jq --raw-output '.seconds' $CONFIG_PATH)
|
|
|
|
# read data
|
|
while true
|
|
do
|
|
if OUTPUT="$(/read_my_sensor.sh)"
|
|
then
|
|
mosquitto_pub -h "$MQTT_SERVER" -p "$MQTT_PORT" -u "$USER" -P "$PASSWORD" -t "$TOPIC" -m "$OUTPUT" || true
|
|
else
|
|
echo "$(data) [ERROR] can't read sensor: $OUTPUT"
|
|
fi
|
|
|
|
sleep "$WAIT_TIME"
|
|
done
|
|
```
|
|
|
|
### {% linkable_title Commands %}
|
|
|
|
We wait for incoming data from the MQTT broker. We can also use an `input_boolean` that triggers an automation to publish a custom command to MQTT topic that can process multiple things in one add-on.
|
|
|
|
Our Dockerfile need to install:
|
|
|
|
```
|
|
RUN apk --no-cache add jq mosquitto-clients
|
|
```
|
|
|
|
Now we can process it with `run.sh`:
|
|
|
|
```bash
|
|
#!/bin/bash
|
|
set -e
|
|
|
|
CONFIG_PATH=/data/options.json
|
|
|
|
# possible options for processing
|
|
MQTT_SERVER=$(jq --raw-output '.server' $CONFIG_PATH)
|
|
MQTT_PORT=$(jq --raw-output '.port' $CONFIG_PATH)
|
|
TOPIC=$(jq --raw-output '.topic' $CONFIG_PATH)
|
|
USER=$(jq --raw-output '.user' $CONFIG_PATH)
|
|
PASSWORD=$(jq --raw-output '.password' $CONFIG_PATH)
|
|
|
|
# read data
|
|
while read -r message
|
|
do
|
|
if [ "$message" == "on" ]; then
|
|
/do_command_on.sh || true
|
|
else
|
|
/do_command_off.sh || true
|
|
fi
|
|
|
|
done < <(mosquitto_sub -h "$MQTT_SERVER" -p "$MQTT_PORT" -u "$USER" -P "$PASSWORD" -t "$TOPIC" -q 1)
|
|
```
|
|
|
|
[MQTT-addon]: /addons/mosquitto/
|
|
[custom-component]: /developers/component_loading/
|
|
[addons-tutorial]: /developers/hassio/addon_tutorial/
|
|
[communication]: /developers/hassio/addon_communication/
|