DepHell -- project management for Python.
Why it is better than all other tools:
- Format agnostic. You can use DepHell with your favorite format: setup.py, requirements.txt, Pipfile, poetry. DepHell supports them all and much more.
- Use your favorite tool on any project. Want to install a poetry based project, but don't like poetry? Just tell DepHell to convert the project's meta information into a setup.py and install it with pip. Or directly work with the project from DepHell, because DepHell can do everything that you usually want to do with packages.
- DepHell doesn't try to replace your favorite tools. If you use poetry, you have to use poetry's file formats and commands. However, DepHell can be combined with any other tool or can even combine all these tools together by converting formats. You can use DepHell, poetry, and pip all at the same time.
- Easily extendable. DepHell has strong modularity and can be easily extended with new formats and commands.
- Developer friendly. We aren't going to place all of our modules into
_internal
. Also, DepHell has a large ecosystem with separated libraries so you can use only the parts of DepHell that you need. - All-in-one-solution. DepHell can manage dependencies, virtual environments, tests, CLI tools, packages, generate configs, show licenses for dependencies, assist with security audits, get download statistics from PyPI, search packages and much more.
- Smart dependency resolution. Sometimes pip and pipenv fail to lock your dependencies. Try to execute
pipenv install oslo.utils==1.4.0
. Pipenv can't handle it, but DepHell can:dephell deps add --from=Pipfile oslo.utils==1.4.0
to add new dependency anddephell deps convert --from=Pipfile --to=Pipfile.lock
to lock it. - Asyncio based. DepHell doesn't support Python 2.7, and that allows us to use modern features to make network and filesystem requests as fast as possible.
- Multiple environments. You can have as many environments for project as you want. Separate sphinx dependencies from your main and dev environment. Other tools like pipenv and poetry don't support this.
Features:
- Manage dependencies: convert between formats, instаll, lock, add new, resolve conflicts.
- Manage virtual environments: create, remove, inspect, run shell, run commands inside.
- Install CLI tools into isolated environments.
- Manage packages: install, list, search on PyPI.
- Build packages (to upload on PyPI), test, bump project version.
- Discover licenses of all project dependencies, show outdated packages, find security issues.
- Generate .editorconfig, LICENSE, AUTHORS, .travis.yml.
See documentation for more details.
Follow @PythonDepHell on Twitter to get updates about new features and releases.
Installation
curl -L dephell.org/install | python3
See installation documentation for alternatives.
Supported formats
- Archives:
- *.egg-info (
egginfo
) - *.tar.gz (
sdist
) - *.whl (
wheel
)
- *.egg-info (
- pip:
- requirements.txt (
pip
) - requirements.lock (
piplock
)
- requirements.txt (
- pipenv:
- Pipfile (
pipfile
) - Pipfile.lock (
pipfilelock
)
- Pipfile (
- pоetry:
- pyproject.toml (
poetry
) - poetry.lock (
poetrylock
)
- pyproject.toml (
- Environment:
- Imports in the package (
imports
). - Installed packages (
installed
).
- Imports in the package (
- Other:
- setup.py (
setuppy
) - flit (
flit
) - conda's environment.yml (
conda
) - pyproject.toml build-system requires (
pyproject
)
- setup.py (
Usage
First of all, install DepHell and activate autocomplete:
python3 -m pip install --user dephell[full]
dephell self autocomplete
Let's get sampleproject and make it better.
git clone https://github.com/pypa/sampleproject.git
cd sampleproject
This project uses setup.py for dependencies and metainfo. However, this format is over-complicated for most projects. Let's convert it into poetry:
dephell deps convert --from=setup.py --to=pyproject.toml
It will make next pyproject.toml
:
[tool.poetry]
name = "sampleproject"
version = "1.2.0"
description = "A sample Python project"
authors = ["The Python Packaging Authority <pypa-dev@googlegroups.com>"]
readme = "README.md"
[tool.poetry.scripts]
sample = "sample:main"
[tool.poetry.dependencies]
python = "!=3.0.*,!=3.1.*,!=3.2.*,!=3.3.*,<4,>=2.7"
coverage = {optional = true}
peppercorn = "*"
[tool.poetry.dev-dependencies]
check-manifest = "*"
[tool.poetry.extras]
test = ["coverage"]
Now, let's generate some useful files:
dephell generate authors
dephell generate license MIT
# https://editorconfig.org/
dephell generate editorconfig
Our users probably have not installed poetry, but they are likely to have pip and can install files from setup.py. Let's make it easier to generate setup.py
from our pyproject.toml
. Also, it points to DepHell as your default dependencies file. Add the following lines in the pyproject.toml
:
[tool.dephell.main]
from = {format = "poetry", path = "pyproject.toml"}
to = {format = "setuppy", path = "setup.py"}
You can see a full, real-world example of a config in DepHell's own pyproject.toml.
Now we can call DepHell commands without explicitly specifying from
and to
:
dephell deps convert
It will make setup.py and README.rst from pyproject.toml and README.md.
Now let's test our code in a virtual environment:
$ dephell venv run pytest
WARNING venv does not exist, creating... (project=/home/gram/Documents/sampleproject, env=main, path=/home/gram/.local/share/dephell/venvs/sampleproject-Whg0/main)
INFO venv created (path=/home/gram/.local/share/dephell/venvs/sampleproject-Whg0/main)
WARNING executable does not found in venv, trying to install... (executable=pytest)
INFO build dependencies graph...
INFO installation...
# ... pip output
# ... pytest output
We can now activate the virtual environment for our project and run any commands inside:
dephell venv shell
Ugh, we have tests, but don't have pytest
in our dependencies file. Let's add it:
dephell deps add --envs dev test -- pytest
Afer that our dev-dependencies looks like this:
[tool.poetry.dev-dependencies]
check-manifest = "*"
pytest = "*"
[tool.poetry.extras]
test = ["coverage", "pytest"]
Eventually we will have many more dependencies. Let's look at how many of them we have now:
$ dephell deps tree
- check-manifest [required: *, locked: 0.37, latest: 0.37]
- coverage [required: *, locked: 4.5.3, latest: 4.5.3]
- peppercorn [required: *, locked: 0.6, latest: 0.6]
- pytest [required: *, locked: 4.4.0, latest: 4.4.0]
- atomicwrites [required: >=1.0, locked: 1.3.0, latest: 1.3.0]
- attrs [required: >=17.4.0, locked: 19.1.0, latest: 19.1.0]
- colorama [required: *, locked: 0.4.1, latest: 0.4.1]
- funcsigs [required: >=1.0, locked: 1.0.2, latest: 1.0.2]
- more-itertools [required: <6.0.0,>=4.0.0, locked: 5.0.0, latest: 7.0.0]
- six [required: <2.0.0,>=1.0.0, locked: 1.12.0, latest: 1.12.0]
- more-itertools [required: >=4.0.0, locked: 7.0.0, latest: 7.0.0]
- pathlib2 [required: >=2.2.0, locked: 2.3.3, latest: 2.3.3]
- scandir [required: *, locked: 1.10.0, latest: 1.10.0]
- six [required: *, locked: 1.12.0, latest: 1.12.0]
- pluggy [required: >=0.9, locked: 0.9.0, latest: 0.9.0]
- py [required: >=1.5.0, locked: 1.8.0, latest: 1.8.0]
- setuptools [required: *, locked: 41.0.0, latest: 41.0.0]
- six [required: >=1.10.0, locked: 1.12.0, latest: 1.12.0]
Hm...Is it as many as it seems? Let's look at their size.
$ dephell inspect venv --filter=lib_size
11.96Mb
Ugh...Ok, it's Python. Are they actual?
$ dephell deps outdated
[
{
"description": "More routines for operating on iterables, beyond itertools",
"installed": [
"5.0.0"
],
"latest": "7.0.0",
"name": "more-itertools",
"updated": "2019-03-28"
},
]
Pytest
requires old version of more-itertools
. That happens.
If our tests and dependencies are OK, it's time to deploy. First of all, increment the project version:
$ dephell project bump minor
INFO generated new version (old=1.2.0, new=1.3.0)
And then build packages:
$ dephell project build
INFO dumping... (format=setuppy)
INFO dumping... (format=egginfo)
INFO dumping... (format=sdist)
INFO dumping... (format=wheel)
INFO builded
Now, we can upload these packages to PyPI:
dephell self auth upload.pypi.org my-username my-password
dephell project upload
These are some of the most useful commands. See documentation for more details.
Compatibility
DepHell is tested on Linux and Mac OS X with Python 3.5, 3.6, 3.7. And one of the coolest things is that DepHell is run by DepHell on Travis CI.
DepHell works on Windows but has no CI to keep in working yet.
How can I help
- Star project on Github. Developers believe in the stars.
- Tell your fellows that Gram has a made cool thing for you.
- Open an issue if you have thoughts on how to make DepHell better.
- Things that you can contribute in any project in DepHell ecosystem:
- Fix grammar and typos.
- Document new things.
- Tests, we always need more tests.
- Make READMEs more nice and friendly.
- View issues with the help wanted label to find things that you can fix.
- Anything what you want. If it is a new feature, please, open an issue before writing the code.
Thank you ❤️