Compare Pipfiles

compare_pipfiles.py

import json
import sys
import csv


def pipfile_to_pkg_version(path):
    with open(path, "r") as fp:
        new_pip = json.load(fp)

    by_env = {}

    for k, env in new_pip.items():
        by_env[k] = by_pkg = {}
        for pkg, data in env.items():
            if not isinstance(data, dict):
                continue
            by_pkg[pkg] = data.get("version", "NONE")
    return by_env


def find_common_keys(d1, d2):
    common = set(d1) & set(d2)
    d1_only = set(d1) - set(d2)
    d2_only = set(d2) - set(d1)
    return d1_only, d2_only, common


def compare_pipfiles(old_pf, new_pf, print_matches=False):
    """Compare old and new pipfile, printing 4 columns.

    Examples:
        env,pkg,old_version,new_version
        default,pandas,0.18.2,0.18.2
        default,boto3,1.0,1.1
        default,newpkg,n/a,1.0
    """
    null_value = 'n/a'
    old_env2pkg2version = pipfile_to_pkg_version(old_pf)
    new_env2pkg2version = pipfile_to_pkg_version(new_pf)
    old_env, new_env, common = find_common_keys(old_env2pkg2version, new_env2pkg2version)
    if old_env or new_env:
        print(f'Only in old pf {old_env}', file=sys.stderr)
        print(f'Only in new pf {new_env}', file=sys.stderr)
    writer = csv.writer(sys.stdout)
    writer.writerow(['env', 'pkg', 'old', 'new'])
    for env in set(common):
        old_pkgs, new_pkgs = old_env2pkg2version[env], new_env2pkg2version[env]
        for k in sorted(set(old_pkgs) | set(new_pkgs)):
            old_version = old_pkgs.get(k, null_value)
            new_version = new_pkgs.get(k, null_value)
            if old_version == new_version and not print_matches:
                continue
            writer.writerow([env, k, old_version, new_version])


if __name__ == '__main__':
    compare_pipfiles(sys.argv[1], sys.argv[2])