Update the `seperator_munge_get` to provide flexibility for dependency lookup

[leeleavitt]

The fix in this PR adresses two example environment yamls,

In this example bullet-python a dependency of pytest-wdl breaks seperator_munge_get

name: python
platforms:
  - linux-64
  - osx-64
channels:
  - bioconda
  - conda-forge
dependencies:
  - python=3.9.12
  - pip
  - miniwdl=1.2.3
  - pip:
    - pytest-wdl

In this example python-tzdata breaks seperator_munge_get,

name: python
platforms:
  - linux-64
  - osx-64
channels:
  - bioconda
  - conda-forge
dependencies:
  - python=3.9.12
  - pybedtools
  - pip
  - pip:
    - pandas==1.3.5

[netlify[bot]]

✅ Deploy Preview for conda-lock ready!

Name	Link
🔨 Latest commit	8c79278b38468168c5bd852a989fac859c47890e
🔍 Latest deploy log	https://app.netlify.com/sites/conda-lock/deploys/643715c0a0c34f0008418f76
😎 Deploy Preview	https://deploy-preview-398--conda-lock.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site settings.

[maresb]

Thank you very much for this @leeleavitt!!!

I have been meaning to dig into this function to try to understand exactly what it's supposed to accomplish. There are some very helpful comments in #290 about what the surrounding code is doing.

I would like to merge #290 very soon. Would you be able to test your update with that branch?

Also, since you have spent some time on separator_munge_get, do you understand exactly what it is doing? If so, would you be able to explain it? In particular, I am wondering why for pip dependencies we aren't using canonicalize_name, and I am also wondering what exactly are the "inconsistencies" within Conda.

[Cakell]

Hi @maresb .

Regarding the logic behind separator_munge_get, see #139 and in particular this commit.

TL;DR:

This code is meant to "Add some slightly more robust fallback handling to find packages that have different separators when using a mixed poetry+conda solve".

Hopes this helps.

PS. I also came across this separator_munge_get bug, and would love to see a release which fixes this soon : )

[maresb]

Thanks @Cakell, does this PR solve the bug for your case?

[olivier-lacroix]

Thanks @Cakell, does this PR solve the bug for your case?

It does for me! encountered the issue with python-tzdata

[leeleavitt]

I will see if i can refactor the code to get around the flake8 complexity failure. Is there a specific version of flake8 or black to use? I couldn't see a version pinned in the repo. Should i just go with the latest versions?

Also @maresb in regards to testing #290 should I merge that branch into mine and then test it?

[maresb]

Thanks @leeleavitt!

To get around the complexity failure, I think you could simply do

try:
    ...
except KeyError:
   pass

and this should allow you to dedent the rest, and I expect this to decrease your complexity score.

should I merge that branch into mine and then test it?

Exactly. Thanks so much for your work on this, I really appreciate it!

[leeleavitt]

Thanks @leeleavitt!

To get around the complexity failure, I think you could simply do

try:
    ...
except KeyError:
   pass

and this should allow you to dedent the rest, and I expect this to decrease your complexity score.

should I merge that branch into mine and then test it?

Exactly. Thanks so much for your work on this, I really appreciate it!

Alright i spent some time testing this yesterday and #290 fixes the issue without the need for my fix. I'm going to write up my findings on that PR.

[lesteve]

Actually I did not see this one before opening #401 but it looks like both are related. Note #401 is NOT fixed by #290.

Removing python- does not seem like super robust. I think there should be a way somewhere to say something like:

• conda tzdata is NOT the same as pip tzdata
• conda python-tzdata is the same as pip tzdata, this is already done I am guessing by the conda pip mapping

[mariusvniekerk]

Thanks a lot for your PR.

As @lesteve pointed out this is not a generally robust solution and special cases should be dealt with by making pull requests to the upstream location wheree the mapping is defined over here https://github.com/regro/cf-scripts/blob/master/conda_forge_tick/pypi_name_mapping_static.yaml

As such I'm closing this PR.