diff

diff.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import difflib

FILE1 = 'file1'
FILE2 = 'file2'

def main():
    fh1 = open(FILE1, 'r')
    fh2 = open(FILE2, 'r')

    c1 = fh1.read()
    c2 = fh2.read()

    differ = difflib.Differ()
    diff = differ.compare(c1.split('\n'), c2.split('\n'))
    orig_line_num = 1
    diffs = []

    left_start = right_start = left_n_line = right_n_line = 0
    initial_skip = 2
    for line in difflib.unified_diff(c1.split('\n'), c2.split('\n'), n=0):
        print line
        if initial_skip > 0:
            initial_skip -= 1
            continue

        if line.startswith('@'):
            left_range, right_range = line.split()[1:3]
            left = left_range.split(',')
            right = right_range.split(',')

            left_start = abs(int(left[0]))
            right_start = int(right[0])

            left_n_line = 1 if len(left) == 1 else int(left[1])
            right_n_line = 1 if len(right) == 1 else int(right[1])
        else:
            delta_mark = line[0]
            code = line[1:]
            if delta_mark == '+':
                orig_line_num = left_start
                if left_n_line == 0:
                    orig_line_num = left_start + 1
                diffs.append({
                    'action': 'insert',
                    'orig_line_num': orig_line_num,
                    'code': code,
                    })
            elif delta_mark == '-':
                diffs.append({
                    'action': 'delete',
                    'orig_line_num': left_start,
                    'code': code,
                    })
                left_start += 1

    print
    print diffs


if __name__ == '__main__':
    main()