fpgmaas / cookiecutter-poetry Goto Github PK

Describe the bug

the -mkdocstrings configuration is not producting docstrings for code

To Reproduce

Steps to reproduce the behavior:

1. just run the cookicutter for an empty project with no uploads set
2. add some docstring to basic code in a code file / test file
3. run make docs
4. look for the modules output

thanks using not sphinx here !
not sure if that is an issue with the cookie but if mkdocstringsworks for you what configuration is needed ?

Expected behavior

some mkdocs are produced andserved for the docstrings

System [please complete the following information]:

• OS: e.g. [Ubuntu 22.04]
• Language Version: [e.g. Python 3.10]
• Virtual environment: [mkdocs 1.5.3, poetry 1.8.0]

just started using the cookie and even first time cookicutter user

Is your feature request related to a problem? Please describe.

Describe the solution you would like

Additional context

Describe the solution you would like

Devcontainer specifications provide a reproducible environment with pre-installed software dependencies. They make it easier for new contributors / collaborators to jump into a project and make contributions instead of spending time setting up their dev environment.

Additional context

If this feature is of interest, I would be willing to contribute the necessary code.

Where it reads:

make lint

Should it be?

make check

I created a repo, did a few PRs which spawned GHA that worked fine. Then, I renamed the repo (typo!) and now the caching seems wrong. The Setup the environment gets a cache which contains nothing: no isort, black, or anything at all. This, of course, causes the next step to fail horribly:

🚀 Checking code formatting: Running isort
make: isort: Command not found
make: *** [Makefile:13: check] Error 127

Any idea what could be going wrong there?

Question
The project looks really interesting so I wonder why not more issues has been raised.

e.g. python 3.9 is pretty old now addays.

the renovate pr has been already been raised.

Is the project actively beeing maintained ?

When the workflow runs is returning error 1 and fails

make: *** [Makefile:28: lint] Error 1 Error: Process completed with exit code 2.

some discussions are here:
PyCQA/isort#1039

similar to fpgmaas/deptry#241

If two templates are installed in an environment at the same time (e.g. this one and cookiecutter-pdm), this causes an issue, since both try to add hooks, {{cookiecutter.project_name}} and cookiecutter.json in the site-packages directory.

This means that only one template is actually installed, and both the commands ccp and ccpdm will trigger the same template. e.g. when installing both packages:

pip install cookiecutter-poetry
pip install cookiecutter-pdm

and then running

ccp
ccpdm

both commands wil create a poetry project!

Solutions

Option 1

A possible solution would be to move the template to a subdirectory:

#91

However, this would mean that the template, if installed from the URL, would need to be installed as follows:

cookiecutter https://github.com/fpgmaas/cookiecutter-poetry.git \
  --directory templates/cookiecutter-poetry

which is less clean and less intuitive. This also breaks the existing functionality.

Option 2

Remove the possibility to install the package, and remove the publishing to pypi. I don't really like this option, because currently it is easy for someone to pin the specific version of the template they want to use by simply installing that version of the template.

Option 3

Find another way to bundle the template with the Python package?

Is your feature request related to a problem? Please describe.

I would like to see a coverage report generated by the unit tests. Clearly, this will have to happen to only the latest (?) python version.

Describe the solution you would like

We could use this GitHub action.

This takes an coverage.lcov report, which coverage can generate via coverage lcov. The latest (or supported) version of Python within tox could generate it.

Additional context

I am not attached to this GitHub action at all, so if there is something else, that would work too.

Question

I'm trying to setup a new project and noticed that a gh-pages branch wasn't created automatically at any point for making auto MkDocs possible. The documentation doesn't specify when/how the branch is created so I've become a bit confused on whether or not I was supposed to create it myself and then run a specific make command or similar to get the docs published on the remote repo. What are the proper steps here? Thanks!

Is your feature request related to a problem? Please describe.

Describe the solution you would like

Additional context

Question

During GitHub Actions tests after pushing to main branch, Tox for some reason can't import modules I've added with poetry add and imported into my source files.
I always get this error: ImportError: libEGL.so.1: cannot open shared object file: No such file or directory
I'm not sure if I need to add something to tox.ini or change how I'm importing in my python files, or both to fix this.
Also, when creating my project with ccp, I accepted all the options except for docker. I'm not sure if that's playing a role here.

For example...

/my-project
  /my_project
    /main.py
  /tests
    /test_main.py

If I do poetry add pyside6
then add something like from PySide6.QtWidgets import QApplication, QWidget to /my_project/main.py
then import said source file to /tests/test_main.py file with from my_project.main import main, I will get the following error from GitHub actions when pushing to the remote main branch...

 ==================================== ERRORS ====================================
  _____________________ ERROR collecting tests/test_main.py ______________________
  tests/test_main.py:3: in <module>
      from my_project.main import main
  my_project/main.py:4: in <module>
      from PySide6.QtWidgets import QApplication, QWidget
  E   ImportError: libEGL.so.1: cannot open shared object file: No such file or directory

What's the proper way to fix tox not able to import properly?
All config files are at their defaults when they were created from ccp and I haven't run any special commands to change anything.

Just wanted to open up a discussion. Have you thought about giving pyright a try instead of using mypy?
I'm curious if you find one more favorable or effective than the other, let's discuss! What are your thoughts?

Describe the bug

make docs fails on a new project with:

Error reading page 'modules.md': No module named 'mkdocstrings.handlers.python'

Is your feature request related to a problem? Please describe.

Requested by @Coinhexa and @lolrenx in #39

Describe the solution you would like

Add pre-commits, https://pre-commit.com/

Is your feature request related to a problem? Please describe.

Docs diffs become much easier to maintain when there is an opinionated code formatter for the Markdown files such as Prettier.
Prettier also allows opinionated formatting of other files relevant to python projects including YAML and JSON.

Describe the solution you would like

An ideal solution would consist of the following:

• Addition of the prettier pre-commit hook
• Reformatting of all existing files using Prettier

Additional context

If this feature is of interest, feel free to assign it to me for completion.

Is your feature request related to a problem? Please describe.

Describe the solution you would like

Additional context

https://github.com/charliermarsh/ruff

Promising new linter, that can also be configured through pyproject.toml

https://pre-commit.com/

As per title, when one initialises a project with

include_github_actions: y
cookiecutter.mkdocs: n
publish_to: none

Cookiecutter generates a on-release-main file with no steps, which fails the build.

Suggested solutions:

Either remove the file all together when publish_to: none or execute all steps until the actual publishing.

Hi there! 👋

I'm the maintainer of check-jsonschema, and a user recently reported the hooks failing on a project generated from this template: python-jsonschema/check-jsonschema#113

I see that there are actions defined in this repo in the workflows dir, which is what was confusing the pre-commit hook.
The hook behavior is to check any YAML files in .github/workflows/ against the workflow schema from schemastore. I'm not confident that the logic is correct, but it is consistent with the match rule in schemastore today.

By my reading of GitHub Workflows docs -- and the observed behavior of the service -- workflows are defined as .github/workflows/*.ya?ml, not .github/workflows/**/*.ya?ml. So I think there's a case to be made that check-jsonschema is not accurately describing github's behavior, but I'm continuing to study the situation a bit before making changes.

Schemastore defines a schema for github-actions, and check-jsonschema provides a hook. In the former case, it matches action.ya?ml in the repo root dir, and in the latter it matches action.ya?ml in the repo root or .github/actions/**/*.ya?ml. The rationale here is that GitHub's docs show use of that directory for local actions, so it seems somewhat common. Supporting validation of more local actions is therefore reasonable down this path.

It would therefore be nice, from my perspective, to move action definitions from subdirs of ./github/workflows to subdirs of ./github/actions/. The hooks would then run correctly without adjustments to check-jsonschema. Even if I make the workflow matching more strict (I probably will, but am taking time to think about it), actions won't be checked outside of the .github/actions/ dir.

Is your feature request related to a problem? Please describe.

Badges do not work in private GitHub repositories, due to the use of shields.io.

Describe the solution you would like

Badges should be functional in private repositories as well.

Using scm tags for versioning would be very useful imo. For example by implementing https://github.com/pypa/setuptools_scm

Is your feature request related to a problem? Please describe.

Describe the solution you would like

Additional context

Is your feature request related to a problem? Please describe.

I can't use my debugger (pudb) with poe api --dev because it is launched in the background. It triggers an error when the debugger is supposed to open:

  File "[...]/site-packages/pudb/debugger.py", line 2683, in event_loop
    keys = self.screen.get_input()
  File "[...]/site-packages/urwid/display/_raw_display_base.py", line 286, in get_input
    keys, raw = self.parse_input(None, None, self.get_available_raw_input())

  [...]

  File "[...]/site-packages/urwid/display/_posix_raw_display.py", line 298, in _getch
    return ord(os.read(fd, 1))
TypeError: ord() expected a character, but string of length 0 found

Describe the solution you would like

Launch the dev server with use_exec = true c.f. https://poethepoet.natn.io/tasks/options.html#defining-tasks-that-run-via-exec-instead-of-a-subprocess

For my project I changed the tasks declarations like that:

  [tool.poe.tasks.dev]
  help = "Serve a REST API for DEV purpose"
  use_exec = true
  cmd = """
    uvicorn \ 
      --host $host \
      --port $port \
      --reload \
      --log-level debug \
      network_manager.api:app
    """

    [[tool.poe.tasks.dev.args]]
    help = "Bind socket to this host (default: 0.0.0.0)"
    name = "host"
    options = ["--host"]
    default = "0.0.0.0"

    [[tool.poe.tasks.dev.args]]
    help = "Bind socket to this port (default: 8000)"
    name = "port"
    options = ["--port"]
    default = "8000"

  [tool.poe.tasks.api]
  help = "Serve a REST API"
  use_exec = true
  cmd = """
    gunicorn \
      --access-logfile - \
      --bind $host:$port \
      --graceful-timeout 10 \
      --keep-alive 10 \
      --log-file - \
      --timeout 30 \
      --worker-class uvicorn.workers.UvicornWorker \
      --worker-tmp-dir /dev/shm \
      --workers 2 \
      network_manager.api:app
    """

    [[tool.poe.tasks.api.args]]
    help = "Bind socket to this host (default: 0.0.0.0)"
    name = "host"
    options = ["--host"]
    default = "0.0.0.0"

    [[tool.poe.tasks.api.args]]
    help = "Bind socket to this port (default: 8000)"
    name = "port"
    options = ["--port"]
    default = "8000"

• remove the dev option and split the existing task in 2
• use cmd with use_exec instead of shell

Because when using use_exec, it's not possible to use shell. And when using cmd, it's not possible to use the shell if instruction... there might be a better way but this seems to work.

Additional context

There might be a better solution... but I can create a pull request if you are happy with this.

Currently mypy is not running automatically (neither locally nor as part of GitHub's pipeline).

A user has to manually run:

make check

If he wants to trigger type checking.

deptry runs as part of the pipeline, but, IMO, it should be also run as part a pre-commit hook as catching dependency problems after they have been introduced is not great .

Proposed solution

Add both tools to pre-commit

Additional context

I've just recently adopted cookiecutter-poetry (thanks a million for it) and butchered some type hints.

I was under the impression that mypy was running automatically but unfortunately this wasn't the case. Later I've realised that there's a dedicated make target for linting, but had to read the Makefile to find it out (see #79).

As both mypy and deptry are able to run very fast I would have no problem running both of them on every commit.

Is your feature request related to a problem? Please describe.

When enabling MyPy type checking, the install types are not included after the make install, unless we manually specify all needed type dependencies.

Describe the solution you would like

We should enable something to include all the install types automatically. A pragmatic solution would be to add @poetry run mypy --install-types. But there might be better solutions, such as poetry-types, as we would like an automatic way to include all used types as (dev) dependencies.

Additional context

Curious what you think about this?