#!/usr/bin/env python

import sys
import re

#set a boolean flag to control the while loop
fileentered = True
while fileentered == True:
    #ask the user to input a filename to check
    filename = raw_input('Please enter a file to check: ')
    #check the input size to see if anything was entered
    if len(filename) >= 1:
        #try to open the file, if it fails go to except
        try:
            seqlist = open(filename, 'r').readlines()
            sequence = ''.join(seqlist)
            sequence = sequence.replace('\n', '')
            totalA = sequence.count('A')
            totalC = sequence.count('C')
            totalG = sequence.count('G')
            totalT = sequence.count('T')
            #create a RegEx to check for extraneous nucleotides
            otherletter = re.compile('[BDEFHIJKLMNOPQRSUVXZ]+')
            #and search for them
            extra = re.findall(otherletter, sequence)
            output = open(filename+'.count', 'w')
            output.write('Count report for file ' + filename + '\n')
            output.write('A = ' + str(totalA) + '\n')
            output.write('C = ' + str(totalC) + '\n')
            output.write('G = ' + str(totalG) + '\n')
            output.write('T = ' + str(totalT) + '\n')
            #check if any 'wrong' nucleotide was found
            if len(extra) > 0:
                output.write('Also were found ' + str(len(extra)) + ' errors\n')
                for i in extra:
                    output.write(i + ' ')
            else:
                output.write('No error found')
            print 'Result file saved on ' + filename + '.count'
        except:
            #this deals with a nonexistent file
            print 'File not found. Please try again.'
    else:
        sys.exit()