Autodetect relevant batteries instead of relying on hardcoded filenames.

[peterdk]

Battery on my Dell XPS 9345 Snapdragon X didn't show up in Resources. Turns out the naming of the battery in /sys/class/power_supply/ is not BATx but a whole different qcom-battmgr-bat. So it seems it's not very wise to rely on a hardcoded detection of ^BAT.

I did some research and figured out a quite simple way to detect which of the power_supply/* items are indeed system batteries and not your Mouse/Keyboard ones.

enumerate /sys/class/power_supply/*    -> conditions: type = "Battery" and scope != "Device"

Or to get a quick debug output:

for a in /sys/class/power_supply/*; do echo "$a: type: $(cat $a/type) scope: $(cat $a/scope 2>/dev/null)"; done

This was tested by a another user with a Dell laptop with a BAT0 entry. I tested it with my Snapdragon laptop. And will try to test it on another laptop as well. But it seems quite solid. Ofcourse, some extra testing would be really nice to see if we regress.

I wanted to add unittests for the logic, but there is no way to mock the fs part, and not sure if we want to use a temp filesystem structure. From what I see in the project the sysfs reading logic is not really tested, so I omitted them.

[peterdk]

Debug bash output on Dell XPS 9345:

/sys/class/power_supply/hidpp_battery_0: type: Battery scope: Device
/sys/class/power_supply/hidpp_battery_1: type: Battery scope: Device
/sys/class/power_supply/qcom-battmgr-ac: type: Mains scope: 
/sys/class/power_supply/qcom-battmgr-bat: type: Battery scope: 
/sys/class/power_supply/qcom-battmgr-usb: type: USB scope: 
/sys/class/power_supply/qcom-battmgr-wls: type: Wireless scope: 
/sys/class/power_supply/ucsi-source-psy-pmic_glink.ucsi.01: type: USB scope: System
/sys/class/power_supply/ucsi-source-psy-pmic_glink.ucsi.02: type: USB scope: System

On A x86 Dell laptop:

/sys/class/power_supply/AC: type: Mains scope: 
/sys/class/power_supply/BAT0: type: Battery scope: 
/sys/class/power_supply/hidpp_battery_4: type: Battery scope: Device
/sys/class/power_supply/ucsi-source-psy-USBC000:001: type: USB scope: System
/sys/class/power_supply/ucsi-source-psy-USBC000:002: type: USB scope: System
/sys/class/power_supply/ucsi-source-psy-USBC000:003: type: USB scope: System

HP Elitebook 840 G4 x86 laptop:

/sys/class/power_supply/AC: type: Mains scope: 
/sys/class/power_supply/BAT0: type: Battery scope: 

[peterdk]

Dell XPS 9345 with this change:

[peterdk]

Flatpak run fails, but seems a internal error not this PR. I can't rerun I guess

[nokyan]

Thank you for the effort! I'll check it out later today or tomorrow!

Flatpak run fails, but seems a internal error not this PR. I can't rerun I guess

Yeah, likely to be a Flatpak action thing (again). I'll look into getting it fixed soon, I'll tell you when I'm done so you can merge main again.

I wanted to add unittests for the logic, but there is no way to mock the fs part, and not sure if we want to use a temp filesystem structure. From what I see in the project the sysfs reading logic is not really tested, so I omitted them.

That's okay, no worries.

[nokyan]

Hi, I've merged a fix for the CI pipeline. :)

[peterdk]

Ah nice, I guess I merge in main and push again?

[nokyan]

Ah nice, I guess I merge in main and push again?

Correct

[peterdk]

Hmm, I rebased, maybe I should have normal merged? Sorry, figuring out a bit how this works with PRs and Github.
Not sure why the diff also shows the .github workflow file as change. Let me know if I need to change something here.

[nokyan]

Hmm, I rebased, maybe I should have normal merged? Sorry, figuring out a bit how this works with PRs and Github. Not sure why the diff also shows the .github workflow file as change. Let me know if I need to change something here.

The way you did it is correct, you've basically taken your changes and "put them on top" of the newest commit of main instead of the one that you started your branch with, hence my commits are now in there as well. :)