""" checking how many duplicate files are in source folder"""

import os
import os.path
from pprint import pprint

import ImporterConst as const


exceptions = ['.ann', '.db']

if __name__ == '__main__':
    files = {}

    for dirpath, dirnames, filenames in os.walk(const.ImageSourceDir):
        for filename in filenames:
            ext = os.path.splitext(filename)[1]
            if str.lower(ext) in exceptions:
                continue
            tmp = files.get(filename, [])
            tmp.append(dirpath)
            files[filename] = tmp
    
    duplicates = {}
    for key in sorted(files.keys()):
        if len(files[key]) > 1:
            duplicates[key] = files[key]

    with open('test01.txt', 'w') as f:
        print >>f, "Total number of duplicates: {}".format(len(duplicates))
        pprint(duplicates, f)