fix(ac): refactor to add new binary power format

[folfy]

merged AC energy / power functions
-> Rewrote these and corresponding calls and helper-functions, to make them more versatile / deduplicate common code

Fixed issue wuwentao/midea_ac_lan#678: AC energy values are x10 of real value (Midea PortaSplit)
-> Introduced format BINARY1 (power_analysis_method 12 = variant of method 2)
Lower digit still represents existing base-formats (BCD, binary, mixed/int), while higher digit(s) can be used to represent different precisions/scalars

Tested changes locally: power_analysis_method 2 and 12 (also got a Midea PortaSplit, actually needing mode 12)

Also, this fixes a bug with energy values >= 655.35 kWh (mode 12) / 6553.5 kWh (existing mode 2), line 1117:
return float((byte1 << 32) + (byte2 << 16) + (byte3 << 8) + byte4) / 10
-> byte1 should be shifted 24 instead of 32 bits

Change needs to be considered in documentation in https://github.com/wuwentao/midea_ac_lan/blob/master/doc/AC.md
Also document it's meaning of interpretation, since it's helpful when migrating from Midea Smart AC (mill1000/midea-ac-py):

• 1 = BCD
• 2 = binary (with default energy/consumption multiplier, so 0.1kWh resolution)
• 3 = mixed/integer
• 12 = binary (with additional x0.1 energy/consumption factor, so 0.01kWh resolution. e.g. Midea PortaSplit)

[wuwentao]

thanks for your PR. we may need to pass the github action.
in additon, could you have a try with devcontainer in vscode? as there is more builtin CI tool like pre-commit in it.

[folfy]

Well, TL;DR;, I fixed the typing and formatting issues, and also some old bugs, like a bad length check (same variable was assigned twice), and removed some redundant/unnecessary code (as highlighted by ruff / mypy / VSCode / etc and confirmed to be OK).

@wuwentao I've spent like at least an hour trying to setup VSCode on my Win 11, with all the plugins and everything, but it was a pain to get that all running with the devcontainer running here.

Checking your Dockerfile, it seems to be related to your pip mirror configuration, because of which it attempts connecting to -pypi.org (leading dash):

> [dev_container_auto_added_stage_label 6/7] RUN pip3 --disable-pip-version-check --no-cache-dir install -r /tmp/pip-tmp/requirements-all.txt    && rm -rf /tmp/pip-tmp:
0.497 Looking in indexes: https://-pypi.org/simple                              
0.501 WARNING: Retrying (Retry(total=4, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<pip._vendor.urllib3.connection.HTTPSConnection object at 0x72bb11003830>: Failed to establish a new connection: [Errno -2] Name or service not known')': /simple/aiofiles/

Dockerfile - looks OK:

RUN echo "[global]" > /etc/pip.conf && \
    echo "index-url = https://${PIP_MIRROR_DOMAIN}/simple" >> /etc/pip.conf && \
    echo "trusted-host = ${PIP_MIRROR_DOMAIN}" >> /etc/pip.conf

devcontainer.json - Idk how parameter expansion works when VSCode loads devcontainer.json, like I know bash would substitute ${<varname>:-<default-value>}, but here the dash in front of pypi.org clearly seems preserved:
"PIP_MIRROR_DOMAIN": "${localEnv:PIP_MIRROR_DOMAIN:-pypi.org}"
I'm also not sure what the first localEnv: is there for / intended to do, like in bash this wouldn't rly make sense, but this should be fixed somehow. Correcting this locally with a static configuration gets me just one step further:

Running the postCreateCommand from devcontainer.json...

[31195 ms] Start: Run in container: /bin/sh -c scripts/setup.sh
/usr/bin/env: ‘bash\r’: No such file or directory

So a Windows CRLF is causing trouble - Found the shebang in script/setup.sh and changed that file from CRLF to LF.

Running the postCreateCommand from devcontainer.json...

[8987 ms] Start: Run in container: /bin/sh -c scripts/setup.sh
pre-commit installed at .git/hooks/pre-commit
pre-commit installed at .git/hooks/commit-msg

added 11 packages in 2s
npm notice
npm notice New patch version of npm available! 11.6.1 -> 11.6.2
npm notice Changelog: https://github.com/npm/cli/releases/tag/v11.6.2
npm notice To update run: npm install -g [email protected]
npm notice
Done. Press any key to close the terminal.

Funny enough, all my files are marked as modified now suddenly, and ofc it's the Windows CRLF line-endings everywhere. Don't know what changed the core.autocrlf setting, would have needed that before already, but it's something to be mentioned for other devs (making sure they check this out with LF). Maybe you could consider writing up a brief CONTRIBUTING.md, with what all you expect to be setup/run and how, before creating a PR?

[wuwentao]

@folfy sorry, as most of us under GFW in china network, we can't directly access github/pip/docker/etc, so we need to config a local mirror about these sites: export APT_MIRROR_DOMAIN=xxxxx, or redirect these domain to proxy.

for windows user, WSL2 should be the expected solution with VScode + devcontainer. install a ubuntu under windows WS2, all these feature should works.
finally, I will try to add some info in CONTRIBUTING.md.

[wuwentao]

@folfy please check with this PR about docs:
#413
if there is any error exist, I will update it.
once got confirmed, i will merge it.
thanks

[wuwentao]

There is some testcases for 0xAC device.
please help to check the error and try to fix it.
thanks!

[folfy]

VSCode/devcontainer setup and proxy config

@folfy sorry, as most of us under GFW in china network, we can't directly access github/pip/docker/etc, so we need to config a local mirror about these sites: export APT_MIRROR_DOMAIN=xxxxx, or redirect these domain to proxy.

for windows user, WSL2 should be the expected solution with VScode + devcontainer. install a ubuntu under windows WS2, all these feature should works. finally, I will try to add some info in CONTRIBUTING.md.

Yeah, I already had WSL2 with Ubuntu installed, but still ran into all these issues mentioned above. At least the devcontainer.json should be fixed:
"PIP_MIRROR_DOMAIN": "${localEnv:PIP_MIRROR_DOMAIN:-pypi.org}"
I found out that "localEnv:PIP_MIRROR_DOMAIN" is a docker (compose?) directive to get this from the users environment. Not sure where / how the default value is expanded, but removing the dash after the colon fixes the issue (using the last value as default), so proposing that fix for devcontainer.json:7-8:

      "APT_MIRROR_DOMAIN": "${localEnv:APT_MIRROR_DOMAIN:deb.debian.org}",
      "PIP_MIRROR_DOMAIN": "${localEnv:PIP_MIRROR_DOMAIN:pypi.org}"

Documentation

@folfy please check with this PR about docs:
#413
if there is any error exist, I will update it.
once got confirmed, i will merge it.
thanks

Thanks, yeah, I'll check later on today or tomorrow 👍

Tests

There is some testcases for 0xAC device. please help to check the error and try to fix it. thanks!

Pylint config issue

Unfortunately tests don't seem to run here - I think cause I got some other Pylint error here still, that I didn't bother to investigate before:

Unrecognized option found: suggestion-mode Pylint(E0015:unrecognized-option)

Seems your setting pylintrc:25 suggestion-mode=yes has been deprecated in Pylint 4.0:
https://pylint.readthedocs.io/en/latest/whatsnew/4/4.0/index.html
So I removed that locally here now.

ruff issue

Also this (typical Windows) issue raised by ruff is probably still blocking my tests:

ruff.....................................................................Failed
- hook id: ruff
- exit code: 1

midealocal/devices/ac/message.py:1:1: EXE002 The file is executable but no shebang is present
Found 1 error.

Guess it's copying the checkout from Windows/NTFS to WSL, where NTFS doesn't have the executable flag, and correspondingly flags every file executable when copying or mount. Dunno how to fix that yet. Ignoring it though VSCode doesn't seem to work... @wuwentao Are you running VSCode natively on Windows? I usually develop on my headless server with vim and co., so not bothering with a graphical environment there. Thought if I need to use VSCode I best run it on Windows, but seems to cause a lot of issues.

Failed test

Yes, this test also is flawed, just like the function itself was:

Also, this fixes a bug with energy values >= 655.35 kWh (mode 12) / 6553.5 kWh (existing mode 2), line 1117:
return float((byte1 << 32) + (byte2 << 16) + (byte3 << 8) + byte4) / 10
-> byte1 should be shifted 24 instead of 32 bits

Since I fixed that, I'll had to correct the two tests in message_ac_test.py line 668 and 677 (shift the highest byte only by 3 bytes / 24 bits instead of 32). Also created a new test-case for method 12, re-using the test for method 2. Hope it's good now, due to the ruff issue above I still can't run tests... 😞

[folfy]

Third time's the charm, now it should be all good - Proposed fixes for devcontainer / pylint in #415

[wuwentao]

@folfy sorry, I'm always using Mac as my working PC or ssh to a Ubuntu Linux remote server, and just did some basic testing with windows PC and WSL2 last year.
these apt/pip mirror changes not test in windows PC, you have fixed it now.
thanks!

In addition, my daily work is too busy these months, so I don't have enough time to process new feature issue/add new feature in this project, anyway, I will continue to do some when time is enough.