I have just updated to 2021.9.2 and am now getting ‘unknown’ for all the pollen entities.
It was working fine from the Canberra site just before the update and the data on the site is still available.
Any ideas ?
Gaz
Hey. I had a quick tinker and had the same behaviour (using Canberra site) but couldn’t quickly diagnose/resolve.
Inrerestingly you can still scrape the forecast value as an attribute (maybe you can work with that?)
- unique_id: melbourne_pollen_forecast_fc1
name: Pollen fc1
select: '#fc1'
attributes:
- name: Forecast Day
select: "#dy1"
value_template: "{{ value[:3] }}"
- name: Forecast Date
select: "#dt1"
#################NEW#################
- name: Forecast Value
select: "#fc1"
#####################################
value_template: "{{ value.strip() }}"
icon: >-
{% if is_state("sensor.melbourne_pollen_forecast_fc1", "Low") %}
mdi:emoticon-happy
{% elif is_state("sensor.melbourne_pollen_forecast_fc1", "Moderate") %}
mdi:emoticon-neutral
{% elif is_state("binary_sensor.melbourne_pollen_forecast_fc1", "High") %}
mdi:emoticon-sad
{% elif is_state("binary_sensor.melbourne_pollen_forecast_fc1", "Extreme") %}
mdi:emoticon-angry
{% else %}
mdi:help-circle
{% endif %}
It’s looking unlikely the Melbourne site will return, and I so I won’t be very active in developing/maintaining work to date.
You may get some assistance on the Multiscape thread, or have a go at tinkering with the scaping. It’s easy once you know how. See the Multiscrape wiki.
Hope that is of some help.
Thanks for taking a look.
The Melbourne site states that they will start again in October and I also noticed that there is an Android app but it does not support the data from Canberra.
Just noticed your edit at the top, no longer free is a bugger.
Gaz
Issue is as per here.
Solution is post after…
After upgrading to V5.5 of multiscrape it is working again.
Gaz
1 Like
Revised sensor for “today” based on what is currently available for Victoria.
This is where it scrapes from.
If you want another area, take a look at some of the multiscrape tutorials. It’s easy once you know how. If you try and fail, I can help.
In config…
############################################################################
##Melbourne Pollen Forecast
############################################################################
multiscrape:
- resource: https://www.melbournepollen.com.au/
scan_interval: 3600 #3600 sec = hr
name: Pollen Forecast
sensor:
- unique_id: melbourne_pollen_forecast_today
name: Pollen Forecast Today
select: '#district-pollen-div > div > div > div > div > div > div.uk-grid-match.uk-child-width-1-2\@s.uk-text-center.uk-grid-collapse.uk-grid > div:nth-child(2) > div:nth-child(1) > div:nth-child(2) > div'
value_template: '{{ value| trim}}'
attributes:
- name: Website Last Updated
select: '#district-pollen-div > div > div > div > div > div > div.uk-grid-match.uk-child-width-1-2\@s.uk-text-center.uk-grid-collapse.uk-grid > div:nth-child(1) > div.ta-notice'
value_template: "{{ value| replace('Last updated:', '')}}"
icon: >-
{% if value == 'Low' %}
mdi:emoticon-happy
{% elif value == 'Moderate' %}
mdi:emoticon-neutral
{% elif value == 'High' %}
mdi:emoticon-sad
{% elif value == 'Extreme' %}
mdi:emoticon-angry
{% else %}
mdi:help-circle
{% endif %}
in lovelace…
- type: custom:button-card
entity: sensor.melbourne_pollen_forecast_today
name: Pollen Forecast
label: >
[[[return
(states['sensor.melbourne_pollen_forecast_today'].attributes.website_last_updated);]]]
tap_action:
action: more-info
show_label: true
show_state: true
color: auto
size: 150%
aspect_ratio: 3/1
layout: icon_state
state:
- value: Low
color: rgb(16, 140, 10)
- value: Moderate
color: rgb(235, 235, 52)
- value: High
color: rgb(235, 147, 52)
- value: Extreme
color: rgb(235, 52, 52)
styles:
name:
- font-weight: bold
- font-size: 0.7vw
- color: white
state:
- font-weight: bold
- font-size: 3vw
- color: white
label:
- font-weight: bold
- font-size: 0.5vw
- color: grey
Thanks for that @Mahko_Mahko I have set it up. Works well. I did see that they didn’t answer you question on facebook about if the 5 day forecast would now be a paid for feature. I guess that non-answer probably means it will be.
Thanks again.
1 Like
Hi - great work maintaining this - I find the pollen card very helpful.
I would love to scrape the thunderstorm asthma warning from the same page as well - I’ve tried (and failed) using the css selector in chrome to replicate what you’ve done but I’m out of my depth.
I had a quick go but failed too. I think I’d failed previously too which is why I didn’t put it up.
I feel like this is right but it’s not working for me.
- unique_id: melbourne_thunderstorm_asthma_today
name: Thunderstorm Asthma Forecast Today
select: '#tae-div > div > div > div > div > div > div.uk-grid-match.uk-child-width-1-2\@s.uk-text-center.uk-grid-collapse.uk-grid > div.uk-first-column > div:nth-child(1) > div:nth-child(2) > div'
value_template: '{{ value| trim}}'
attributes:
- name: Website Last Updated
select: '#district-pollen-div > div > div > div > div > div > div.uk-grid-match.uk-child-width-1-2\@s.uk-text-center.uk-grid-collapse.uk-grid > div:nth-child(1) > div.ta-notice'
value_template: "{{ value| replace('Last updated:', '')}}"
icon: >-
{% if value == 'Low' %}
mdi:emoticon-happy
{% elif value == 'Moderate' %}
mdi:emoticon-neutral
{% elif value == 'High' %}
mdi:emoticon-sad
{% elif value == 'Extreme' %}
mdi:emoticon-angry
{% else %}
mdi:help-circle
{% endif %}
thanks for trying. that looks similar to what I was doing
1 Like
Oh wow. What’s the deal? 
No deal. Just an unofficial api that can easily be found if you know how to 
Legend. I actually had a dig around for it too but couldn’t figure it out.
thanks, this is super helpful!
I’m chipping away at it and making some progress. Don’t really know what I’m doing though.
I’ve been trying to hack something together this morning to use the asthma forecast.
The HTML nested within the JSON makes it hard to work with.
my solution at the moment is this:
I’m using a shell command to extract just the asthma forecast HTML from the API (this is div 5)
asthma_refresh: 'curl "https://api.pollenforecast.com.au/app/json/app_data.php?app=1&version=2" | jq .div5 >/config/www/asthma.html'
I’m running this as a service every 30 mins.
I’m then running multiscrape over the output:
- resource: http://[IP address]/local/asthma.html
scan_interval: 3600 #3600 sec = hr
name: Asthma forecast
sensor:
- unique_id: asthma_forecast_scrape
name: Asthma forecast
select: '#\\\"ta-forecast-table\\\" > tbody > tr:nth-child(1) > td.\\\"center\\\" > p'
attributes:
- name: Website Last Updated
select: 'body > div > div > div:nth-child(8) > p'
This is not a neat way to do it, but it does give me a sensor in homeassistant.
Nice progress. Better than my attempts. Yeah I had the same struggles with the html nested in the json.
I managed to get the forecast html in as text using the rest sensor, then I was tinkering with some very gross parsing of the HTML as text, despite finding some hilarious warnings.
Edit: Added detail
Like If you have this:
rest:
- authentication: basic
scan_interval: 600
resource: https://api.pollenforecast.com.au/app/json/app_data.php?app=1&version=2
headers:
Accept: application/json
Content-Type: application/xml
sensor:
- name: "Melbourne Pollen test"
value_template: "OK"
json_attributes:
- "forecastpage"
- "div5"
- "share"
Then you get a sensor with the requested attributes that have the massive html text as attributes (I was hoping HA might recognise it and parse it)
Then I was going to try to somehow parse the html from each of the above json_attributes it into some simplified json that could be stored nested and then queried. I dunno if that makes sense.
In template developer tools
{% set myxml = states.sensor.melbourne_pollen_test.attributes.div5 %}
{{
myxml
|replace('\n', '*+*+')
|striptags
|replace(' *+*+ *+*+ *+*+ ', '\n')
|replace('*+*+ *+*+ *+*+ ', '')
|replace('*+*+ *+*+ ', ': ')
|replace('*+*+', '')
}}
Output of above template starting to look a bit like json pairs.
Thunderstorm Asthma Forecast :
District Risk of thunderstorm asthma
: Central : low
East Gippsland : low
Mallee : low
North Central : low
North East : low
Northern Country : low
South West : low
West and South Gippsland : low
Wimmera : low
Note: The above epidemic thunderstorm asthma forecast information was provided by the Victorian Department of Health and Human Services and the Bureau of Meteorology. For more information pertaining to these forecasts visit this website. Last updated: 2021-11-22 14:00:00; Current date: 2021-11-22
Update #2: On my “steaming pile of poo solution”
With the same rest sensor above, you can then create a template sensor like this to do some dogy parsing of the html into json
sensor:
- platform: template
sensors:
melbourne_pollen_forecast_api_dev:
friendly_name: Melbourne Pollen Forecast API Dev
value_template: 'Ok'
attribute_templates:
pollen_forecast: >-
{% set myxml = states.sensor.melbourne_pollen_test.attributes.forecastpage %}
{{
'{\n' +
myxml
|replace('Melbourne Pollen Forecast', '"data": [')
|replace('<H2 ', '*NL*{*NL*"day_name":"<H1 ')
|replace('</h2>', '</H2>",*NL* ')
|replace('<P', '"date":"<P')
|replace('P>', '<\/P>",*NL* ')
|replace('<div class="pollen-forecast-level', '"pollen_level":"<<div class="pollen-forecast-level')
|replace('</span>', '"</span>*NL*},*NL* ')
|striptags
|replace('*NL*', '\n') + ']\n}'
}}
Which will look like this:
{
"data": [
{
"day_name": "Monday",
"date": "22 November 2021",
"pollen_level": "HIGH"
},
{
"day_name": "Tuesday",
"date": "23 November 2021",
"pollen_level": "HIGH"
},
{
"day_name": "Wednesday",
"date": "24 November 2021",
"pollen_level": "MODERATE"
},
{
"day_name": "Thursday",
"date": "25 November 2021",
"pollen_level": "LOW"
},
{
"day_name": "Friday",
"date": "26 November 2021",
"pollen_level": "MODERATE"
},
{
"day_name": "Saturday",
"date": "27 November 2021",
"pollen_level": "MODERATE"
}
]
}
Then you can access values like this:
{{states.sensor.melbourne_pollen_forecast_api_dev.attributes.pollen_forecast.data[1].pollen_level}}
Returns "HIGH" in template Developer tool
if it works, it works! you’ve made great progress
1 Like