I had scrape configured to scrape the temperature from a website, but suddenly it is not working anymore. The data on the page looks like this:
<data>
<sea>
<depth>
<dval>
<![CDATA[ 1.0 ]]>
</dval>
<date>
<![CDATA[ 2023-06-12 ]]>
</date>
<time>
<![CDATA[ 17:03:40 ]]>
</time>
<measure>
<type>
<![CDATA[ 0 ]]>
</type>
<value>
<![CDATA[ 17.65 ]]>
</value>
</measure>
<measure>
<type>
<![CDATA[ 1 ]]>
</type>
<value>
<![CDATA[ 22.47 ]]>
</value>
</measure>
</depth>
This used to work:
But after the latest update it stopped working.
The logfile says:
ValueError: Sensor sensor.temp75m has device class ‘temperature’, state class ‘measurement’ unit ‘°C’ and suggested precision ‘None’ thus indicating it has a numeric value; however, it has the non-numeric value: ‘’ (<class ‘str’>)
I’m assuming value contains <![CDATA[ 22.47 ]]> as a string.
This is a relatively new change to HA where sensors with “unit_of_measurement” need to be numeric. You’re stripping out the parts around the number, but leaving the leading and trailing spaces. You can also parse it as a float with a default value to cater for this:
{{ value.lstrip('<![CDATA[').rstrip("]]>") | float(0) }}
Note that techinically “lstrip” and “rstrip” strip any matching character from the left/right and stop when a non-matching character is found, so “<![CDATA[” and “<![CDAT” are the same, although I would probably also do the same.
Note that HA can also convert XML to JSON, so there may be a simpler way to achieve what you’re doing.
Tried to add space and float(0) to the template:
{{ value.lstrip('<![CDATA[ ').rstrip(" ]]>") | float(0) }}
The sensor is now only displaying 0°C.
