#!/usr/bin/env python3

import sys

# CLU2IG - Convert interleaved Clustal V .aln output to sequential
# .ig format suitable for use with MASE
# Brian Fristensky        August 3, 1992

# Usage: cul2ig clustalfile > masefile

""" ensure that there are enough command line arguments to parse """
if len(sys.argv) < 2:
    print("Usage: cul2ig.py clustalfile > masefile");
    exit();

name_list = ()
sequences = {}

# Clean up the file by removing title lines, '*' and blank lines
h_file = open(sys.argv[1])
for line in h_file:
    if line != '' and "CLUSTAL V" not in line and '*' not in line:
        TEMP.CLU2MASE
        
        # Create a namefile from the first 16 columns of the cleaned-up file.
        name = line[:16]
        if not sequences[name]:
            seuqneces[name] = line[16:80]
            name_list.append(name)
h_file.close()

# Find out how many unique names there are, and store in the environment
# variable numnames. Write the first $NUMNAMES lines of RAWNAMES.CLU2MASE
# to UNIQUE.CLU2MASE

name_list = set(name_list.sort())

# For each unique name in the namefile, grep out the corresponding lines
# and write to the output file
for name in name_list:
    #create a dummy comment
    print ";"

    #write the name
    print name
    print sequences[name]