µPipenv

https://travis-ci.com/thoth-station/micropipenv.svg?branch=master

A lightweight wrapper for pip to support requirements.txt, Pipenv and Poetry lock files or converting them to pip-tools compatible output. Designed for containerized Python applications but not limited to them.

For a brief video preview, check this demo (the micropipenv part starts at 9:00).

See also micropipenv: Installing Python dependencies in containerized applications for more info about this tool and a blog post published about it.

What's the difference in comparision to pip when using requirements.txt?

if requirements.txt state all the packages in a pinned version with hashes (e.g. pip-tools), micropipenv installs packages with a possible fallback if the installation order is relevant
- you don't need to care about the installation and maintain correct order or requirements in requirements.txt
- best effort installation - try until there is a possibility to succeed
if requirements.txt do not state all the packages in a pinned form
- pip's resolver algorithm is used and it's left on pip to resolve requirements
- the same behavior as micropipenv would not be used

What's the difference in comparision to Poetry?

a lightweight addition to Poetry, not a Poetry replacement
- micropipenv does not substitute Poetry it rather complements it for containerized deployments where the size of the container image and software shipped with it matters
no release management to Python package indexes
micropipenv does not implement resolver, it uses already resolved stack that application is shipped with based on poetry.lock and pyproject.toml
no virtual environment management
- virtual environment management is left on user, if needed

What's the difference in comparision to Pipenv?

a lightweight addition to Pipenv, not a Pipenv replacement
- micropipenv does not substitute Pipenv it rather complements it for containerized deployments where the size of the container image and software shipped with it matters
it does not vendor all the dependencies as Pipenv
micropipenv does not implement resolver, it uses already resolved stack that application is shipped with Pipfile.lock
no virtual environment management
- virtual environment management is left on user, if needed

micropipenv use cases

Why should I use micropipenv instead of Pipenv or Poetry?

I would like to have a tool that "rules them all" - one lightweight tool to support all Python dependency lock file managers (pip-tools, Poetry, Pipenv) and lets users decide what they want to use when deploying Python applications in containerized environments (e.g. Kubernetes, OpenShift, ...).
I would like to have a fast and minimalistic tool to install software packages in CI.
I would like to have containerized Python applications as small as possible with minimum software shipped and required to build and run the Python application in production.
I would like to convert files produced by Pipenv/Poetry to a pip-tools compatible output.
I don't want to install Pipenv/Poetry, but I would like to run a project that uses Pipenv/Poetry for dependency management (e.g. restricted environments).
My Pipenv installation is broken and Pipenv upstream did not issue any new Pipenv release.
I would like to deploy my application into a production environment and my application dependencies are managed by Pipenv/Poetry (dependencies are already resolved), but I don't want to run Pipenv/Poetry in production (e.g. OpenShift's s2i build process).

`micropipenv install`

The tool supports installing dependencies of the following formats:

Pipenv style lock format - files Pipfile and Pipfile.lock
Poetry style lock format - files pyproject.toml and poetry.lock
pip-tools style lock format - file requirements.txt
raw requirements.txt as used by pip (not a lock file)

In case of Pipenv, Poetry and pip-tools style format, the tool performs automatic recovery if the installation order of dependencies is relevant (one dependency fails to install as it depends on an another one).

To enforce the installation method used, specify --method option to the install subcommand. By default, micropipenv traverses the filesystem up from the current working directory and looks for the relevant files in the following order:

Pipfile.lock and optionally Pipfile (if --deploy set)
poetry.lock and pyproject.toml
requirements.txt for pip-tools and raw pip requirements

To install dependencies issue the following command:

micropipenv install --dev  # --dev is optional

You can supply additional positional arguments that will be passed to pip. Use double dashes to distinguish pip options from micropipenv options.

# issue `pip install --user'
micropipenv install -- --user

Virtual environment management

micropipenv does not create any virtual environment as in the case of Pipenv/Poetry. It simply executes pip, and constructs arguments out of the lock file used. pip's default behaviour is to install its arguments into the same Python environment where pip is installed (i.e., the system Python environment).

To rather install its arguments into a virtual environment, run micropipenv from a shell where that virtual environment has been activated. This causes micropipenv to run the pip command already installed in the virtual environment:

python3 -m venv venv/ && (. venv/bin/activate && micropipenv install)

Alternatively, micropipenv can run the pip command from the system Python environment, which can in turn be told to install its arguments into an explicitly specified virtual environment by using the PIP_PYTHON environment variable.

This allows for the installation of dependencies into a minimal virtual environment that does not itself have pip and its dependencies installed into it:

python3 -m venv venv/ --without-pip && PIP_PYTHON=venv/bin/python micropipenv install

Using another package index

To set the default Python Package Index to something other than https://pypi.org/simple, set the MICROPIPENV_DEFAULT_INDEX_URLS to one or more comma-separated URLs.

Note: if the package manager file contains a package index URL, it will be used over this value.

export MICROPIPENV_DEFAULT_INDEX_URLS=https://pypi.example.com/simple,https://pypi.org/simple
micropipenv install

`micropipenv install --deploy`

If you wish to mimic pipenv --deploy functionality, you can do so:

micropipenv install --deploy

Note however, there is a need to parse Pipfile and verify its content corresponds to Pipefile.lock used (digest computed on Pipfile content). micropipenv requires toml extras for this functionality, so you will need to install micropipenv[toml] (see installation instructions bellow).

The --deploy option takes no effect for Poetry and requirements installation methods.

`micropipenv install --dev`

Installation of "development" dependencies can be acomplished using the --dev flag. This flag has no effect when requirements.txt file is used.

`micropipenv requirements` / `micropipenv req`

To generate output compatible with pip-tools, you can issue the following command:

micropipenv requirements

This applies to conversion from Poetry and Pipenv specific lock files.

Additional configuration options can limit what is present in the output (e.g. --no-dev to remove development dependencies).

A special option --only-direct makes micropipenv work on Pipfile instead of Pipfile.lock. This requires toml extras, so install micropipenv[toml] for this functionality (see installation instructions bellow). To get direct dependencies of an application and store them in requirements.txt file:

micropipenv requirements --only-direct > requirements.txt

For a setup that follows pip-tools convention with requirements.in and requirements.txt

micropipenv requirements --no-dev > requirements.txt
micropipenv requirements --no-dev --only-direct > requirements.in
micropipenv requirements --no-default > dev-requirements.txt
micropipenv requirements --no-default --only-direct > dev-requirements.in

See micropipenv requirements --help for more info.

`micropipenv` as a library

micropipenv exposes some core functionality on top of Pipfile/Pipfile.lock. You can import its functions and use micropipenv as a lightweight library for Pipfile/Pipfile.lock and pyproject.toml/poetry.lock manipulation.

Adjusting options using environment variables

All options can be triggered using environment variables - the name of an environment variable is always prefixed with MICROPIPENV_ and consists of the name of the option converted to uppercase, dashes are replaced with underscores (example --no-dev is mapped to MICROPIPENV_NO_DEV). All environment variables corresponding to flags are parsed as integers and subsequently casted to a boolean. For example, to turn --no-dev flag on, set MICROPIPENV_NO_DEV=1 (0 disables the flag). Parameters supplied to CLI take precedence over environment variables.

A special environment variable MICROPIPENV_PIP_BIN can point to an alternate pip binary.

To run this tool in a verbose mode, you can set the MICROPIPENV_DEBUG=1 (the same behavior can be achieved with multiple --verbose supplied).

The tool prints software stack information to the standard error output. This was designed for Thoth to capture information about installed dependencies as a useful source of information for Thoth's build analyzers. This behaviour can be suppressed by setting MICROPIPENV_NO_LOCKFILE_PRINT=1 environment variable.

Besides printing, the tool also writes the content of Pipfile.lock (if a locked software stack is used) to the directory where lock files are present (for Pipenv files, the Pipfile.lock is kept untouched). This behaviour can be suppressed by providing MICROPIPENV_NO_LOCKFILE_WRITE=1 environment variable.

Example usage

Install dependencies managed by Poetry as pip install --user would do (option --method is optional, auto-discovery is performed if omitted):

$ ls
poetry.lock pyproject.toml project.py
$ micropipenv install --method poetry -- --user

Install dependencies (both main and develop) managed by Poetry into a virtual environment:

$ ls
poetry.lock pyproject.toml project.py
$ python3 -m venv venv/
$ . venv/bin/activate
(venv) $ micropipenv install --dev

Install dependencies managed by Pipenv (both main and develop) into a virtual environment (option --method is optional, auto-discovery is performed if omitted):

$ ls
Pipfile Pipfile.lock src/
$ python3 -m venv venv/
$ . venv/bin/activate
(venv) $ micropipenv install --dev

Perform deployment of an application as Pipenv would do with Python interpreter version check and Pipfile file hash check (you can create virtual environment only if necessary):

$ ls
Pipfile Pipfile.lock src/
$ python3 -m venv venv/
$ . venv/bin/activate
(venv) $ micropipenv install --deploy

Generate pip-tools compliant requirements.in, dev-requirements.in, requirements.txt and dev-requirements.txt out of Pipfile and Pipfile.lock - project dependencies managed by Pipenv:

$ ls
Pipfile Pipfile.lock src/
$ micropipenv requirements --no-dev > requirements.txt
$ micropipenv requirements --no-dev --only-direct > requirements.in
$ micropipenv requirements --no-default > dev-requirements.txt
$ micropipenv requirements --no-default --only-direct > dev-requirements.in

Generate pip-tools complaint requirements.in, dev-requirements.in, requirements.txt and dev-requirements.txt out of pyproject.toml and poetry.lock - project dependencies managed by Poetry:

$ ls
poetry.lock pyproject.toml src/
$ micropipenv requirements --no-dev > requirements.txt
$ micropipenv requirements --no-dev --only-direct > requirements.in
$ micropipenv requirements --no-default > dev-requirements.txt
$ micropipenv requirements --no-default --only-direct > dev-requirements.in

For OpenShift's s2i integration, check this repo with a demo.

Installation

The project is hosted on PyPI so installing it using pip works as expected:

pip install micropipenv

The default installation does not bring any dependencies so its just micropipenv that gets installed. However, the default installation supports only Pipfile.lock management. If you would like to manipulate also with Pipfile or Poetry specific lock files, you will need to install micropipenv with TOML support (TOML is not in the standard Python library):

pip install micropipenv[toml]

Once the project gets installed, you can browse the help message by invoking the micropipenv CLI:

micropipenv --help

If you wish to install micropipenv on your Fedora system:

dnf install -y micropipenv

See available RPM packages.

No installation

You can run micropipenv without actually installing it - simply download the file and execute it. If you do not wish to save micropipenv.py file to disk, you can issue:

curl https://raw.githubusercontent.com/thoth-station/micropipenv/master/micropipenv.py | python3 - --help

Anything after python3 - will be passed as an argument to micropipenv.py so installing packages can be simply performed using:

curl https://raw.githubusercontent.com/thoth-station/micropipenv/master/micropipenv.py | python3 - install -- --user

All arguments after -- will be passed to pip as options.

OpenShift s2i (Source-To-Image)

micropipenv is available in UBI, Fedora and RHEL based container images. To enable micropipenv and benefit from its features, you need to export ENABLE_MICROPIPENV=1 environment variable in more recent Python 3 container images. See sclorg/s2i-python-container repo for more information.

License and copying

This project is licensed under the terms of the GNU Lesser General Public License v3 or later. See LICENSE-LGPL and LICENSE-GPL files for the license terms.

Original author:

Fridolín 'fridex' Pokorný <fridolin@redhat.com>

Maintainers:

Lumír 'Frenzy' Balhar <lbalhar@redhat.com>
Max Gautier <max.gautier@redhat.com>

thoth-station/micropipenv