""" dump duplicate files from source folder into SQL query"""

import os
import os.path

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, v in files.iteritems():
    #    if len(v) > 1:
    #        duplicates[key] = v
    #
    queryTmpl = "insert filemapping (fname, fpath, duplicate) values ('{filename}', '{filepath}', '{duplicate}')"
#    for filename, paths in duplicates.iteritems():
    for filename, paths in files.iteritems():
        duplicate = 1 if len(paths) > 1 else 0
        for path in paths:
            filepath = '\\'.join(path.split('\\')[-2:])
            query = queryTmpl.format(filename=filename, filepath=filepath, duplicate=duplicate)
            print query