How do I check whether a module is installed in Python, and install it if needed?

In terminal, after I start Python, how will I know what are the modules present in python?
Suppose I need to learn the modules NumPy and SciPy.

  • How will I install it if it is not installed?
  • How will I know if it is already installed?
Asked By: sigdelsanjog

||

How to know if a python module is installed or not in the system: You can do a very easy test in terminal,

$ python -c "import math"
$ echo $?
0                                # math module exists in system

$ python -c "import numpy"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ImportError: No module named numpy
$ echo $?
1                                # numpy module does not exist in system

How will I install it if it is not installed

You can install specific module by downloading respective packages from repository, for example you can install scipy as,

sudo apt-get install python-scipy ## for Python2
sudo apt-get install python3-scipy ## for Python3

Alternately You can also install a python module using python-pip as suggested by Zack Titan in the comment below, To install numpy you can use

pip install numpy

Warning: It is highly recommended to install python-modules using official Ubuntu repository only and not to use the pip method as superuser(i.e., as root or using sudo). In some cases it may leave your system unusable by breaking system python.

How to install packages using pip into local virtual environment.

Answered By: sourav c.

You could put the code inside try, except block.

$ python3 -c "
try:
    import cow  
    print('nModule was installed')
except ImportError:
    print('nThere was no such module installed')"

There was no such module installed

$ python3 -c "
try:
    import regex
    print('nModule was installed')
except ImportError:
    print('nThere was no such module installed')"

Module was installed
Answered By: Avinash Raj

In case we do not want to unwantedly import a module in question (which would happen in a try statement) we can make use of sys.modules to test modules that are installed and were imported before.

In the python shell issue:

>>> import sys

Then test for installed modules:

>>> 'numpy' in sys.modules
True
>>> 'scipy' in sys.modules
False

Note that only those modules that were imported before give True on this test, all other modules (even if installed) result in False.

Another alternative to try an import statement in the python console is calling the inbuilt help() function. This will not give a documentation for non-installed modules, e.g.

>>> help('scipy')
no Python documentation found for 'scipy'

The output of very long help documents of installed modules can be interrupted with Q.

Now to install missing modules it is recommended to use the Ubuntu package management (and not the Python pip way) because we need root access and also to prevent messing up our heavily Python-dependend system. For the module in question this would e.g. be:

sudo apt-get install python-scipy ## for Python2
sudo apt-get install python3-scipy ## for Python3

After installation we then can add them to the sys.modules dictionary by importing them once.

Answered By: Takkat

To provide another answer, for completion’s sake:

You can (ab)use the -m option. From Python’s manpage:

   -m module-name
          Searches sys.path for the named module and runs the  correspond‐
          ing .py file as a script.

Which will give us:

$ python2 -m numpy
/sbin/python2: No module named numpy.__main__; 'numpy' is a package and cannot be directly executed

$ python2 -m math
/sbin/python2: No code object available for math

But for non-existent modules, it will give us:

$ python2 -m doesnt_exist
/sbin/python2: No module named doesnt_exist

We could use grep to match for this:

$ python2 -m doesnt_exist |& grep -q 'No module named' && echo 'Nope' || echo 'Yup'
Nope

$ python2 -m math |& grep -q 'No module named' && echo 'Nope' || echo 'Yup'
Yup

This is slightly hack-ish, and not what -m was intended for; but it is the
method that requires the least typing if you want a quick test 🙂

Answered By: Martin Tournoij

Another way is the pkgutil module. Works with both Python 2 & 3:

python -c 'import pkgutil; print(1 if pkgutil.find_loader("module") else 0)'

You need to replace module with the name of your module, example:

$ python -c 'import pkgutil; print(1 if pkgutil.find_loader("math") else 0)'
1
Answered By: cuonglm

I know the OP originally asked for a solution after starting Python, but outside of python I use pip. On ubuntu: sudo apt-get install python-pip, if it’s not already installed.

Then to see what third party modules are available, just run:

pip freeze

Or even

pip list

And both will show you all modules installed and their versions.

If the module you’re looking for is not installed, most of the time you can easily install it with pip:

pip install <module-name>


If you’re not sure of whether a module exists or what its PyPI name is, use `pip search`:

pip search <keyword>

It appears that pip search no longer functions. (See @AJM’s comment below). I tend to use Pypi’s search directly

Answered By: OozeMeister

I wrote an example in Python:

import pip
import sys
from bigml.api import BigML


if not 'bigml' in sys.modules.keys():
    pip.main(['install', 'bigml'])
Answered By: Mona Jalal

One can also use pydoc modules, which can be filtered with grep to find a specific module. The output is displayed in columnated format. The only disadvantage of this approach is that it also will include python files in the current working directory. Nonetheless,I use it myself most of the time and it’s one of the highly cited approaches on this related question: https://stackoverflow.com/q/739993/3701431

Answered By: Sergiy Kolodyazhnyy

Shell oneliner check (useful for provisioning)

I found that in order to make my infrastructure provisioning** idempotent, I need to be able to check for a package from the shell in a oneliner. I built on @cuonglm’s answer. I had to reverse the 1 and 0 because I’m producing an exit status rather than printing a string.

python -c "import sys, pkgutil; sys.exit(0 if pkgutil.find_loader(sys.argv[1]) else 1)" pymongo

You could replace the sys.argv[1] with the single quoted name of your package, but for my provisioning scripts I like the readability of having it at the end.

python -c "import sys, pkgutil; sys.exit(0 if pkgutil.find_loader('pymongo') else 1)"

** I realize that chef, puppet, and ansible all have plugins for managing python packages, but you may find yourself in a situation like me where you are using an outdated version and don’t want to use deprecated plugins.

Answered By: Bruno Bronosky

From the Ubuntu Shell, by default bash, as simple as

pip list | grep <package-name-Case-Matters>

Examples

pip list | grep pywinrm

pip list | grep numpy

And, if you have doubts about the case (although I think all the package names are always lowercase):

pip list | grep [Nn]um[Pp]y  # it works with numpy, Numpy, numPy, and NumPy
Answered By: xCovelus

As of writing this, pip show ... seems to be the easiest way:
https://pip.pypa.io/en/stable/reference/pip_show/

But it is silent (ie. returns nothing) when the package is not installed.

Answered By: GuSuku

I would do something like this:

#!/bin/bash

pymodules=(
    requests
    termcolor
)

for module in "${pymodules[@]}"; do
    if python3 -c "import pkgutil; exit(1 if pkgutil.find_loader("$module") else 0)"; then
        pip3 install --user "$module"
    fi
done

It will install any module thats missing from the pymodules array.

Using --user makes pip install packages in your home directory instead of a system directory like /usr/local/lib/python3.7/, this is useful as it doesn’t require any special privileges and keeps your system installation clean.

Answered By: htaccess

this is what I come up with:

if [[ ! $(pip list|grep "numpy") ]]; then
    pip install numpy
fi

It will check if numpy has been installed on your system, if not, it will install numpy using pip.

Answered By: jdhao

This code will check if the modules in the list are installed or not. If a module is not yet installed, it will install it.

import sys
modules=['pandas','numpy'] # modules to install
for module in modules:
    if module in sys.modules:
        print('module: '+module+' already installed')
    else:
        print('install module: '+module)
        !pip install module
Answered By: B. Martin

to install or update pip packages (if they are not installed or up to date) from a terminal or shell script:

#!/usr/bin/env bash

# install or update pip and yq if they are not installed or up to date
for pkg in pip yq; do
  pip list --uptodate | grep "${pkg} " || pip install --upgrade ${pkg}
done
Answered By: grenade