#! /usr/bin/python

import os
import sys
import stat
import re
import itertools

dirname = sys.argv[1]
d = os.getcwd()
os.chdir(dirname)

all_files = []
fnames = os.listdir(dirname)

dupes = []
all_prefix = []
all_suffix = []

for line in fnames:
    if "Data_" in line:
        matches = re.findall(r"^Data_[a-z]{1,2}_FlatNTuple_.[0-9]{0,3}_[0-9]{1}_[0-9a-zA-Z]{3}\.root" ,line)
        #matches = re.findall(r"^Data_[a-z]{1,2}_FlatNTuple_.[0-9]{0,3}_" ,line)
        matches_prefix = re.findall(r"^Data_[a-z]{1,2}_FlatNTuple_.[0-9]{0,3}_" ,line)
        matches_prefix = str(matches_prefix[0])
        matches = str(matches[0])
        matches_suffix = matches[-10:]
        all_prefix.append(matches_prefix)
        all_suffix.append(matches_suffix)
cleaned_files = dict(itertools.izip(all_prefix,all_suffix))

os.chdir(d)
all_files = dict(itertools.izip(all_suffix,all_prefix))

cleaned_values = []
filesToRemove = []
for key in cleaned_files.keys():
    cleaned_values.append(cleaned_files[key])
for key in all_files.keys():
    #print key
    if key not in cleaned_values:
        filesToRemove.append(str(all_files[key])+str(key))


for i in filesToRemove:
    dupes.append(os.path.join(dirname, i))

for i in dupes:
    print "rm "+str(i)