Wrapper around pip commands to auto-update requirements file


Keywords
easy_install, distutils, setuptools, egg, virtualenv, requirements, management-workflow, metadata, pip, pip-manager, python, python-package-management, python-requirements
License
MIT
Install
pip install pipm==18.2.0

Documentation

pipm

Build Status

Python package management using pip, requirements file & setup.cfg.

Installation

  • Adviced to install only inside virtualenv

Install from PyPI

pip install pipm

Note:

  • This tool manipulates all your requirements file. So be sure to use version control software or take backup of your files to keep track of changes.

Quickstart

  • Both pip and pipm command will work as the same.
  • Create a virtualenv for the project and install pipm with
pip install pipm
  • Create an alias as alias pip=pipm or use as it is - pipm

I. Install

  • install all your dependencies from the base requirements file (requirements.txt)
pipm install
pipm install --all  # *requirements*.txt - all environment -> test/prod/dev

2. Add new packages to project

pipm install pkg-name
pipm install pkg-name --dev # as development dependency
pipm install pkg-name --test # as testing dependency

3. Removal of packages

  • Remove one or more packages. Their dependencies will also get uninstalled. No orphaned packages.
pipm uninstall pkg-name

4. update all your dependencies in requirements.txt

pipm update
pipm update --dev

Features

  1. No new set of files. *-requirements.txt works like the lockfile with pinned versions
  2. Just a wrapper around the standard pip's install/uninstall command. So all the cli options will work
  3. Handles multiple requirements files and setup.cfg stores abstract dependencies.

Commands

1. install

  • a wrapper around standard pip install command and accepts all the standard options

Below are the things that pipm brings to the table

  1. Extra functionality
    • when package names are given it will be saved to the requirements.txt file in the current directory. If you have requirements directory structure with base.txt inside then that file will be used. Otherwise it will create one in the current directory.
    • when no package name is given then it is equivalent to -r requirements.txt and it will install all requirements from the current directory
  2. Additional options: It will search for the matching one in the following pattern <name>-requirements.txt or requirements/<name>.txt or requirements-<name>.txt the below saves to file when package name given otherwise equivalent to passing requirements file name.
    1. --dev - saves to development requirements
    2. --prod - saves to production requirements
    3. --test - saves to testing requirements
    4. --doc - saves to documentation requirements
    5. --env <name> - if you have any special set of requirements that belong to a separate file you could pass the name here.

2. uninstall

  • a wrapper around standard pip uninstall command
  • alias rm is available
  • when uninstalling a package, this command also checks packages that are no longer required by any of user installed packages and removes them
  • ofcourse it removes the packages from requirements files

3. update

  • new command
  • equivalent to calling pip install with --upgrade flag
  • update a single package or the whole environment when no argument given.
  • by default the packages are updated interactively
    • set --auto-update to disable this

4. save/freeze

  • extends the standard freeze command to save the currently installed packages

Development

  • clone the repository and create new virtualenv
git clone git@github.com:jnoortheen/pipm.git
cd pipm
pew new pipm -a .
pip install -r dev-requirements.txt
  • to test from local sources
pip install -e .
  • Commit message should follow this style-guide.

Testing

  • run invoke test from the root directory.

Alternatives and their problems (IMHO)

  1. pipenv
    • good for local development with only one virtual environment per project
    • Not good when we need to deploy over production server or keep multiple virtuals-envs
    • it is better to use pew alone instead of the shell command that comes with this
  2. pip-tools
    • another set of files to keep track of, additional commands to remember
  3. poetry
    • more robust than pipenv
    • ~~the problems I faced are related to installing dependencies in remote servers/docker environments. ~~
    • longer install/update times ~~As the project matures this problem might get resolved. ~~
    • since version 1.1 these are solved.
    • new projects are to advised to use poetry when possible.
  4. dephell
    • haven't used it. But seems to handle requirements files as well as other formats.

TODOs:

  • rm will check whether a package is present in setup.cfg