Template sensor check if string is NULL

Configuration
21

So now I am at the same situation as 24hours ago.
Works without unit_of_measurement. But not with.
But then why does it just stop working the next day.
The serial data is more or less the same.
And waterdepth I have kept disconnected.

22

That is what I thought yesterday.
But what I cannot explain is why does 24 hours later change its behavior.
I even restarted the hardware, Rpi 4. But no difference.
I will put the null condition check again and see if that improves it.

23
Script I used for testing 
alias: zTest Repeat For testing
sequence:
  - repeat:
      for_each:
        - PRESS,,f,61,M,,F*3F
        - WEGT,11,1,F,5,4F
        - PRESS,,f,62,M,,F*3F
        - WEGT,12,1,F,5,4F
        - WEGT,13,1,F,5,4F
        - PRESS,,f,,M,,F*3F
        - WEGT,14,1,F,5,4F
        - PRESS,,f,63,M,,F*3F
        - WEGT,15,1,F,5,4F
        - PRESS,,f,64,M,,F*3F
        - WEGT,16,1,F,5,4F
        - WEGT,17,1,F,5,4F
        - WEGT,18,1,F,5,4F
        - PRESS,,f,65,M,,F*3F
        - WEGT,19,1,F,5,4F
        - PRESS,,f,66,M,,F*3F
        - PRESS,,f,67,M,,F*3F
        - WEGT,20,1,F,5,4F
        - WEGT,21,1,F,5,4F
        - PRESS,,f,68,M,,F*3F
        - WEGT,22,1,F,5,4F
        - PRESS,,f,69,M,,F*3F
        - WEGT,23,1,F,5,4F
        - WEGT,24,1,F,5,4F
        - PRESS,,f,,M,,F*3F
        - WEGT,25,1,F,5,4F
        - PRESS,,f,70,M,,F*3F
        - WEGT,26,1,F,5,4F
        - PRESS,,f,71,M,,F*3F
        - WEGT,27,1,F,5,4F
        - WEGT,28,1,F,5,4F
        - WEGT,29,1,F,5,4F
        - PRESS,,f,72,M,,F*3F
        - WEGT,30,1,F,5,4F
        - PRESS,,f,73,M,,F*3F
        - PRESS,,f,74,M,,F*3F
        - WEGT,31,1,F,5,4F
        - WEGT,32,1,F,5,4F
      sequence:
        - service: input_text.set_value
          data:
            value: "{{ repeat.item }}"
          target:
            entity_id:
              - input_text.ztest_text_test
        - delay: 30
mode: single

Is that a typo (I think the forum software will auto-corrects double commas)? Previously you posted it as follows: "PRESS,f,,M,F*3F”

24

Yes must be some cut past error or typo I made. 65 is replaced with null.
PRESS,f,65,M,F3F
PRESS,f,M,F3F

The 3rd element on the 2nd line is null.
I added quotes just to show it is a string. The actual serial data does not include quotes.

25

I have figured out what the problem is.
If I define a unit_of_measurement: “m” then the template sensor MUST return a numerical value.
If it tries to return none or “” or ‘’ it will error out and not return anything.

So what can I set it If want to return a null value and I have a unit_of_measurement: “m” defined??
It works if I return a 0 but that is not correct in my case.

26

The short answer is you can’t…

What is the purpose of returning a null value?

If the goal is to determine that the sensor isn’t updating, you can and should set up an availability. You will need to figure out how you want to differentiate between when the template needs to report null/unavailable and when it should use the previous state. You will likely need to define attributes or other standalone sensors.

28

I figured it out! I used availability. And also got the timestamp working that we discussed a while back.
here is the code.

    - name: waterdepth
      unique_id: water_depth_id
      unit_of_measurement: "m"
      state: >-
        {% set new_data_condition = (d | count > 3 and d[0]|length >= 6 and d[0][0] == 'X' and d[0][3:6] == 'WBT') %}
        {{ d[3].split('*')[0] if new_data_condition else this.state }}
      availability: >-
        {% set new_data_condition = (d | count > 3 and d[0]|length >= 6 and d[0][0] == 'X' and d[0][3:6] == 'WBT') %}
        {% set time_difference_seconds = as_timestamp(now()) - as_timestamp(this.attributes.updated_timestamp | default(now())) %}
        {{ not (new_data_condition and d[3] == '') and (time_difference_seconds <= 60) }}
      attributes:
        updated_timestamp: >-
          {% set new_data_condition = (d | count > 3 and d[0]|length >= 6 and d[0][0] == 'X' and d[0][3:6] == 'WBT') %}
          {{ now() if new_data_condition else this.attributes.updated_timestamp | default(now()) }}

It would be nice not to have to repeat the new_data_condition logic 3 times. But I could not see how to set it to have scope for the whole sensor. Seems to only have scope per section.
Any ideas?
Since I now have 500 sensors I cant really declare a 500 different variables at the trigger level.

29

There has been a Feature Request thread for variables in template sensors since before there were this or trigger variables, 2 variants of trigger variables, or custom template macros…

You may be able to reduce the size by using more specific triggers instead of relying on a single, general trigger that you then have to parse over and over.