Output file size and max length of filepath per division to csv from robocopy log.

analyze_robocopy_log.py

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

# Usage:
# script.py <robocopy.log> [robocopy.log]...

import re
from sys import argv 

# key = division name
# value = list of total file size and max file path length
info_per_division = {}

for file in argv:
    if file == argv[0]:
        continue
    f = open(file)
    for line in f:
        # get file size, path and path length.
        l = line.strip().split()
        size = long(l[0])
        file_path = ''
        for path in l[1:]:
            file_path += ' ' + path
        file_path = file_path.lstrip()
        uni_str = file_path.decode('cp932')
        file_path_length = len(uni_str)

        # get division.
        division = ''
        r = re.compile(r"\\(A[0-9]{1,2}_.*?)\\")
        result = r.search(file_path)
        if result is not None:
            division = result.group(1)

        # update information.
        if division not in info_per_division:
            info_per_division[division] = [size, file_path_length]
        else:
            total_size = info_per_division[division][0]
            max_filepath_length = info_per_division[division][1]
            info_per_division[division] = [
                    total_size + size,
                    max_filepath_length if max_filepath_length > file_path_length else file_path_length]
    f.close()

print '部署名, total_size, max_filepath_length'
for division in info_per_division:
    print division + ', ' + str(info_per_division[division][0]) + ', ' + str(info_per_division[division][1])