"""This tool merges given sets of paired reads into one pseudo fastq file. Several input file can be given - the list of files containing the pairs MUST have the same order for first and second in pair! (-p PAIRED_1_1,PAIRED_2_1,PAIRED_3_1,... -P PAIRED_1_2,PAIRED_2_2,PAIRED_3_2,...)"""

import sys

def parse_options(argv):

    """Parses options from the command line """

    from optparse import OptionParser, OptionGroup

    parser = OptionParser()
    required = OptionGroup(parser, 'REQUIRED')
    required.add_option('-p', '--pair_first', dest='pair_1', metavar='FILE', help='comma separated list of files containing the first reads of the pair', default='-')
    required.add_option('-P', '--paor_second', dest='pair_2', metavar='FILE', help='comma separated list of files containing the second reads of the pair', default='-')
    required.add_option('-o', '--outfile', dest='outfile', metavar='FILE', help='file to store the merged file in pseudo fastq', default='-')
    parser.add_option_group(required)

    (options, args) = parser.parse_args()
    
    if len(argv) < 4 :
        parser.print_help()
        sys.exit(2)

    return options



def main():
    """Main unction parsing and merging the paired files. """

    options = parse_options(sys.argv)

    ### open outstream
    outfile = open(options.outfile, 'w')

    pair_1_list = options.pair_1.strip().split(',')
    pair_2_list = options.pair_2.strip().split(',')

    assert(len(pair_1_list) == len(pair_2_list))

    for idx in range(len(pair_1_list)):
        pair_1_file = pair_1_list[idx]
        pair_2_file = pair_2_list[idx]

        ### open first of pair
        pair_1 = open(pair_1_file, 'r')
        ### open second of pair
        pair_2 = open(pair_2_file, 'r')

        counter = 1
        line1 = pair_1.readline()
        line2 = pair_2.readline()

        while line1 and line2:
            if counter % 100000 == 0:
                print 'lines [ %s ] in %s and %s' % (counter, pair_1_file, pair_2_file)

            if counter % 4 == 1:    # read_id
                print >> outfile, line1.strip()[1:-2] + '\t',
            elif counter % 4 == 2:  # sequence
                print >> outfile, line1.strip() + '\t' + line2.strip() + '\t',
            elif counter % 4 == 0:  # quality
                print >> outfile, line1.strip() + '\t' + line2.strip()

            line1 = pair_1.readline()
            line2 = pair_2.readline()
            counter += 1

        pair_1.close()
        pair_2.close()
    outfile.close()

if __name__ == "__main__":
    main()