Quick and dirty script for git-extras

list_dependencies.py

#!/usr/bin/env python
# coding: utf-8
"""
Run this quick-and-dirty script to list dependencies used in git-extras/bin.
With -t option, print result with markdown table syntax.
With -v option, grep usage to filter out false positive result.
"""
from collections import defaultdict
import os
import re
import sys
import subprocess


def is_cmd(cmd):
    try:
        path = subprocess.check_output(['which', cmd])
        return os.path.isfile(path.strip())
    except subprocess.CalledProcessError:
        return False


print_as_table = False
print_verbose = False
if len(sys.argv) > 1:
    print_as_table = sys.argv[1] == '-t'
    print_verbose = sys.argv[1] == '-v'

word_pattern = re.compile('[^\-_$\w]([a-z]+)[^=_\-(\w]')
file_token = {}
for fn in os.listdir('.'):
    if not fn.startswith('git-'):
        continue
    with open(fn) as f:
        context = f.read()
        file_token[fn] = set(filter(lambda x: len(x) > 1, re.findall(word_pattern, context)))

token_file = defaultdict(set)
blacklist = set()
whitelist = set()

for fn in file_token:
    for token in file_token[fn]:
        if token in blacklist:
            continue
        if token not in whitelist and not is_cmd(token):
            blacklist.add(token)
            continue
        else:
            whitelist.add(token)
        token_file[token].add(fn)

if print_as_table:
    print('| dependency | script |')
    print('| --- | --- |')

sorted_token = sorted(token_file.keys())
for token in sorted_token:
    if print_as_table:
        files = ' '.join(map(lambda x: '`%s`'%x,  token_file[token]))
        print('| %-20s | %s |' % (token, files))
    elif print_verbose:
        args = ['grep', '--color', token]
        args.extend(token_file[token])
        print(' '.join(args))
        subprocess.call(args)
        print('')
    else:
        files = ' '.join(token_file[token])
        print('%-20s: %s' % (token, files))