I’ve been using the BoM sensor for current weather data for a little while now, but was inspired by this thread to start fetching forecast data as well: Weather forecast for Australia using BOM website
But, while that thread uses the scrape sensor to scrape an HTML page, I wanted to use one of the data feeds. Doing this is probably pretty advanced, and definitely a lot more involved, because the files are in XML format and HA doesn’t seem to have the tools to easily parse these built in. But it also exposes a lot more useful information, and once you get a feel for it I think it’s a lot more flexible.
These sensors and automation are set up to send me notification of the weather forecast and some current conditions in the morning.
To start, I had to install the libxml2-utils package on my Raspberry Pi, just using the regular package manager.
After that I had to find the data file I was interested in. I went to the forecast page (for Sydney, http://www.bom.gov.au/nsw/forecasts/sydney.shtml ) and took a note of the “Product ID” at the bottom of the page (for the Sydney forecast, IDN10064). Data files are available from an FTP service at ftp://ftp.bom.gov.au/anon/gen/fwo/ , look for a file named after your product ID with a .xml extension. For Sydney, the full URL is ftp://ftp.bom.gov.au/anon/gen/fwo/IDN10064.xml .
I tried using the downloader component for this, but currently the downloader doesn’t seem to work for FTP (I’ve opened #10968 which will hopefully fix this). So I set up a shell command to download a file, then an automation to run the command once in the evening:
### configuration.yaml
# Substitute the URL for your forecast here
shell_command:
download_forecast: curl -s --retry 3 --output $HOME/.homeassistant/downloads/forecast/sydney-metro.xml ftp://ftp.bom.gov.au/anon/gen/fwo/IDN10064.xml
### automations.yaml
- id: Download BoM Sydney forecast
alias: Download BoM Sydney forecast
trigger:
platform: time
hours: 20
minutes: 12
seconds: 0
action:
service: shell_command.download_forecast
Then I set up some command_line sensors to extract the data I was interested in. These sensors use the xmllint command line tool, and this is where things get hairy. I’ve written queries that work for my use case in the Sydney forecast, but if you’re trying to adapt this you’ll probably need to tweak the queries. For that, you’d need to read the XML file to get an idea of the structure, have a look at how to construct basic XPath expressions, and do some trial and error.
Anyway, I wanted to get the full long forecast, the min and max forecast temperatures, chance and amount of rain, and finally the forecast icon code. These sensors run on match on the forecast-period element with index of 1. That is, they match on tomorrow’s forecast. I only download the forecast once, late in the evening, then these sensors are valid for the next day. In future I’d like to experiment with updating the forecast during the day, but this works for me now.
### configuration.yaml
sensor:
- platform: command_line
name: weather forecast
command: "xmllint --xpath 'string(//product/forecast[1]/area[@aac=\"NSW_ME00
1\"]/forecast-period[@index=\"1\"]/text)' $HOME/.homeassistant/downloads/forecas
t/sydney-metro.xml"
value_template: '{{ value | truncate(255) }}'
- platform: command_line
name: weather forecast icon
command: "xmllint --xpath 'string(//product/forecast[1]/area[@aac=\"NSW_PT13
1\"]/forecast-period[@index=\"1\"]/element[@type=\"forecast_icon_code\"])' $HOME
/.homeassistant/downloads/forecast/sydney-metro.xml"
- platform: command_line
name: weather forecast rain chance
command: "xmllint --xpath 'string(//product/forecast[1]/area[@aac=\"NSW_PT13
1\"]/forecast-period[@index=\"1\"]/text[@type=\"probability_of_precipitation\"])
' $HOME/.homeassistant/downloads/forecast/sydney-metro.xml"
- platform: command_line
name: weather forecast rain amount
command: "xmllint --xpath 'string(//product/forecast[1]/area[@aac=\"NSW_PT13
1\"]/forecast-period[@index=\"1\"]/element[@type=\"precipitation_range\"])' $HOM
E/.homeassistant/downloads/forecast/sydney-metro.xml"
- platform: command_line
name: weather forecast minimum temp
command: "xmllint --xpath 'string(//product/forecast[1]/area[@aac=\"NSW_PT13
1\"]/forecast-period[@index=\"1\"]/element[@type=\"air_temperature_minimum\"])'
$HOME/.homeassistant/downloads/forecast/sydney-metro.xml"
- platform: command_line
name: weather forecast maximum temp
command: "xmllint --xpath 'string(//product/forecast[1]/area[@aac=\"NSW_PT131\"]/forecast-period[@index=\"1\"]/element[@type=\"air_temperature_maximum\"])' $HOME/.homeassistant/downloads/forecast/sydney-metro.xml"
The final part I needed was the existing BoM sensor. All I’m using so far is current temperature, so:
### configuration.yaml
sensor:
- platform: bom
station: IDN60901.94768
name: Sydney
monitored_conditions:
- air_temp
Finally I have an automation to send a push notification with the forecast every morning. This notification is one of a few places I’m using the forecast icon code. The notification template defines an array of emoji representing the different weather icons (sunny, cloudy, rain, storm, etc). Then the title template uses the icon code as an index of this array to create a dynamic title.
### automations.yaml
- id: weather forecast notification
alias: weather forecast notification
trigger:
platform: time
at: 06:15
action:
service: notify.html5_notifier
data_template:
title: >
{%- set icons = ["", "\u2600\uFE0F", "\uD83C\uDF19", "\u26C5", "\u2601\uFE0F", "", "\uD83C\uDF01", "", "\uD83C\uDF27\uFE0F", "\uD83C\uDF2C\uFE0F", "\uD83C\uDF01", "\uD83C\uDF26\uFE0F", "\uD83C\uDF27\uFE0F", "\uD83D\uDE37", "‽", "\u2603\uFE0F", "\uD83C\uDF29\uFE0F", "\uD83C\uDF27\uFE0F", "\uD83C\uDF27\uFE0F", "\uD83C\uDF00"] -%}
{%- set index = states.sensor.weather_forecast_icon.state | int -%}
{{ icons[index] }} Weather forecast
message: |
Good morning!
{{ states('sensor.weather_forecast') }}
{{ states.sensor.weather_forecast_minimum_temp.state }}°-{{ states.sensor.weather_forecast_maximum_temp.state }}°, currently {{ states.sensor.bom_sydney_air_temp_c.state }}°. {{ states.sensor.weather_forecast_rain_chance.state }} chance of {{ states.sensor.weather_forecast_rain_amount.state }} rain.
