#!/usr/bin/python
# programmer : Bo
# usage:

import sys
import re
import random
import string
import time

def main(X):
    try:
        
        infile = open(X,"r+").readlines()
        print X, 'has opened, total ',len(infile),' Lines'
        return infile
    except IOError,message:
        print >> sys.stderr, "cannot open file",message
        sys.exit(1)

if __name__=="__main__":
    tS = time.time()
    G = main('/home/zhangbo/Genome/mm10/gencode.vM8.annotation.gtf.gene')
    g = {}
    for ea in G:
        te = ea[:-1].split('\t')
        tte = te[-1].split(',')
        g[tte[0]] = [tte[1],tte[2]]
    F = main(sys.argv[1])
    o = file('Mod_'+sys.argv[1],'w')
    o.write('\tEN_ID\tGene_type'+F[0])#.replace(',','\t'))
    for i in range(1,len(F)):
        te = F[i].replace('"','').split('\t')
        o.write(te[0])
        for ea in g[te[0]]:
            o.write('\t'+ea)
        for j in range(1,len(te)):
            o.write('\t'+te[j])
        

    tE = time.time()
    print 'Cost ',(tE-tS),' sec'

