diff_stat.py

#!/usr/bin/env python3
"""
transform the diff program output format to "git diff --stat" style.
./diff_stat.py [diff_file]
"""

import sys

def get_fname(diff_data):
    """
    format:
    diff -r qemu/arch_init.c qemu-arm_fastvm/arch_init.c

    diff_data_split_by_space[2]: qemu/arch_init.c
    """
    diff_data_split_by_space = diff_data.split(' ')
    name = diff_data_split_by_space[2]
    return name

def raw_diff_to_stat(diff_filename):
    f = open(diff_filename, 'r')
    content_line = f.readlines()

    stat_files = []
    fname_max_len = 0
    for line in content_line:
        if line.startswith('diff'):
            cur_file = get_fname(line)
            fname_max_len = max(fname_max_len, len(cur_file))
            cur_diff_descr = {'decre': 0, 'incre': 0}

            stat_files.append((cur_file, cur_diff_descr))

        elif line.startswith('<'):
            stat_files[-1][1]['decre'] += 1
        elif line.startswith('>'):
            stat_files[-1][1]['incre'] += 1

    return stat_files, fname_max_len

def print_diff_stat(stat_files, fname_max_len):
    for file_ in stat_files:
        name = file_[0] + (fname_max_len + 1 - len(file_[0]))*' '
        incre = '+' + str(file_[1]['incre'])
        decre = '-' + str(file_[1]['decre'])
        print('{name:20}| {incre:>5}, {decre:>5}'.format(name=name, incre=incre, decre=decre))

if __name__ == '__main__':
    if len(sys.argv) != 2:
        print("./diff_stat.py [diff_file]")

    diff_filename = sys.argv[1]
    stat_files, fname_max_len = raw_diff_to_stat(diff_filename)
    print_diff_stat(stat_files, fname_max_len)