`pip index show` command which fetches the package information via the JSON API

[ashwinvis]

What's the problem this feature will solve?

Now that pip search is not working and would be removed in favour of pip index versions in #10431 , there is short description is not accessible from the command line. This is an opportunity to re-imagine the command.

Describe the solution you'd like

If numpy is installed in the environment, the following description is obtained

$ pip show numpy
Name: numpy
Version: 1.21.2
Summary: NumPy is the fundamental package for array computing with Python.
Home-page: https://www.numpy.org
Author: Travis E. Oliphant et al.
Author-email: 
License: BSD
Location: /home/avmo/.pyenv/versions/3.8.9/envs/snek/lib/python3.8/site-packages
Requires: 
Required-by: xarray, pandas, matplotlib, h5py

Say, numpy is not installed in the user environment.

$ pip show numpy
WARNING: Package(s) not found: numpy

The user would like to read the documentation, but only knows the name of the package. An output like this would be helpful

$ pip index show numpy
Name: numpy
Version: 1.21.2
Summary: NumPy is the fundamental package for array computing with Python.
Home-page: https://www.numpy.org
Author: Travis E. Oliphant et al.
Author-email: 
License: BSD
Location: < irrelevant >
Requires: 
Required-by:  < hard to deduce? >

Such an output can be populated from the JSON API

curl -s https://pypi.org/pypi/numpy/json | jq .info
{
  "author": "Travis E. Oliphant et al.",
  "author_email": "",
  "bugtrack_url": null,
  "classifiers": [
    "Development Status :: 5 - Production/Stable",
    "Intended Audience :: Developers",
    "Intended Audience :: Science/Research",
    "License :: OSI Approved :: BSD License",
    "Operating System :: MacOS",
    "Operating System :: Microsoft :: Windows",
    "Operating System :: POSIX",
    "Operating System :: Unix",
    "Programming Language :: C",
    "Programming Language :: Python",
    "Programming Language :: Python :: 3",
    "Programming Language :: Python :: 3 :: Only",
    "Programming Language :: Python :: 3.7",
    "Programming Language :: Python :: 3.8",
    "Programming Language :: Python :: 3.9",
    "Programming Language :: Python :: 3.10",
    "Programming Language :: Python :: Implementation :: CPython",
    "Topic :: Scientific/Engineering",
    "Topic :: Software Development",
    "Typing :: Typed"
  ],
  "description": "It provides:\n\n- a powerful N-dimensional array object\n- sophisticated (broadcasting) functions\n- tools for integrating C/C++ and Fortran code\n- useful linear algebra, Fourier transform, and random number capabilities\n- and much more\n\nBesides its obvious scientific uses, NumPy can also be used as an efficient\nmulti-dimensional container of generic data. Arbitrary data-types can be\ndefined. This allows NumPy to seamlessly and speedily integrate with a wide\nvariety of databases.\n\nAll NumPy wheels distributed on PyPI are BSD licensed.\n\n\n\n",
  "description_content_type": "",
  "docs_url": null,
  "download_url": "https://pypi.python.org/pypi/numpy",
  "downloads": {
    "last_day": -1,
    "last_month": -1,
    "last_week": -1
  },
  "home_page": "https://www.numpy.org",
  "keywords": "",
  "license": "BSD",
  "maintainer": "NumPy Developers",
  "maintainer_email": "[email protected]",
  "name": "numpy",
  "package_url": "https://pypi.org/project/numpy/",
  "platform": "Windows",
  "project_url": "https://pypi.org/project/numpy/",
  "project_urls": {
    "Bug Tracker": "https://github.com/numpy/numpy/issues",
    "Documentation": "https://numpy.org/doc/1.21",
    "Download": "https://pypi.python.org/pypi/numpy",
    "Homepage": "https://www.numpy.org",
    "Source Code": "https://github.com/numpy/numpy"
  },
  "release_url": "https://pypi.org/project/numpy/1.21.2/",
  "requires_dist": null,
  "requires_python": ">=3.7,<3.11",
  "summary": "NumPy is the fundamental package for array computing with Python.",
  "version": "1.21.2",
  "yanked": false,
  "yanked_reason": null
}

Alternative Solutions

Alternatively, simply pretty print the JSON after fetching it.

Right now I use https://github.com/jeffmm/pypi-simple-search/ to mimic the old pip search behaviour.

Additional context

The inspiration for this command is how pacman in ArchLinux works.

• pacman -Qi <package>: to query an installed package for information
• pacman -Si <package>: to query the package index for information

Code of Conduct

• [X] I agree to follow the PSF Code of Conduct.

[uranusjr]

The JSON API is not standardised, and our current principle is to only implements new features with a backing standard (old features are kept for compatibility). So you should push for the JSON API’s standardisation first.

See https://discuss.python.org/t/9205 for discussion on this.

[pfmoore]

Also, why is it necessary to implement this in pip when a 1-liner using curl and jq like you show gives the results so easily? (For people without curl or jq, a few lines of Python would give something similar).

[ashwinvis]

@uranusjr Then we wait for the standard.

@pfmoore Displaying package description is a common feature of all package managers (not just pacman, but apt search, cargo search, and so on). I would assume others would like to have that information before they install the package. I don't see the benefit of a third party tool just to query the JSON API.

[pfmoore]

I guess that begs the question of whether pip is a "package manager". I've always thought of it as an "installer" rather than a broader tool. But I can see why framing it as a package manager makes sense.

[mikedlr]

The reason for this command is that I would like to know what pip would do on an install before I do it. Right now I know of no way to predict the result of pip install unknown-and-maybe-dodgy-package except maybe doing pip download beforehand. What I'd like is to be able to run

pip info unknown-and-maybe-dodgy-package

and see at least

• name of package
• checksum of package to be installed
• url of documentation for package

so that I have enough information to go away and check that this package is the one that I meant to install.

BTW. I would propose "info" as an alias (or even primary name) for this which would match with RedHat's dnf command. For apt, show uses the network.

[ashwinvis]

see at least

* name of package

* checksum of package to be installed

* url of documentation for package

To do all of that the only way seems to be making a JSON API call (as I showed above) or visiting the PyPI project page (for example https://pypi.org/project/numpy/#files). If you only care about the dependencies and the checksums / hashes, you could use "pip-tools" by creating a requirements.in file and executing pip-compile --generate-hashes.

I know this is not the answer @mikedlr or me is looking for, but as a stop gap measure, hope it helps.

[wyattscarpenter]

I often want to do this, and therefore think something like this would be a good feature!

Ideally, pip show would show the information for both installed and uninstalled packages, and also have a line Installed: yes/no.

(When this functionality is added, there can be a big yellow warning at the top of pip show output for uninstalled packages: The package {package_name} is not installed, fetching details from the package index.; this will help users who are working on the heuristic "pip show prints something → the package is installed")

[uranusjr]

I think this is now somewhat doable with the distribution metadata api, at least for wheels. There’s still things to figure out (the metadata api is file-based; how do we handle cases where no wheels are available, etc.) but feel free to contribute if this is important to you.