Script to find out all third party imports from a code repository

findimports.py

# Script to find all third party imports from python projects
# How to Run
# sudo pip install find-import virtualenv pymod2pkg
# python findimports.py <path to python code repository>
# Author: Chandan Kumar <chkumar246@gmail.com>

import pymod2pkg
import os
import shutil
import subprocess
import shlex
import sys
import virtualenv
import tempfile

def list_all_pyfiles(path):
    """
    list down all .py files with complete path
    """
    files_path = []
    for root, subdir, files in os.walk(path):
        for file in files:
            if file.endswith('.py'):
                files_path.append(os.path.join(root, file))
    return files_path

def get_imports(filepath):
    """
    Returns a list of all imports
    """
    output = subprocess.check_output(['find-import', filepath])
    if output.split():
        return output.split()

def get_all_imports(file_list):
    """
    Return a dict of all imports with all imports
    {'file_path':[a, b, c, d]}
    """
    imports, import_list = {}, []
    for file in file_list:
        if get_imports(file) != None:
            imports.update({file: get_imports(file)})
    for data in imports.values():
        import_list.extend(data)
    return list(set(import_list))

def check_import(modules):
    """
    Find thirdparty modules
    """
    third_party_imports = []
    path = tempfile.mkdtemp()
    virtualenv.create_environment(path, clear=True)
    venv_path = os.path.join(path, 'bin', 'python')
    for module in modules:
        # FIX ME corner case for import import and import the statement
        if not module in ['import', 'the']:
            cmd = shlex.split("%s -c 'try:\n import %s\n print('True')\nexcept ImportError: print('False')'" % (venv_path, module))
            out = subprocess.check_output(cmd)
            if 'False'in out.split():
                third_party_imports.append(module)

    # Remove virtualenv
    shutil.rmtree(path)
    return third_party_imports

def list_third_party_pkgs(third_party_imports):
    """
    List all third party packages from imports
    """
    pkgs = [pymod2pkg.module2package(module, pymod2pkg.platform.linux_distribution()[0]) for module in third_party_imports]
    return pkgs

if __name__ == "__main__":
    if len(sys.argv) > 1:
        if os.path.exists(sys.argv[1]):
            files = list_all_pyfiles(sys.argv[1])
            imports = get_all_imports(files)
            third_party_imports = check_import(imports)
            pkgs = list_third_party_pkgs(third_party_imports)
            print('\n'.join(pkgs))
        else:
            print("Path Not found.")