#! /usr/bin/env python
import os, sys, string, cp, SeqIO

def main():
	"""
%(filename)s

Converts fasta format sequence alignment to phylip2 (sequential)
format. Mercilessly truncates sequence names to 10 characters.

{{out	Name of output file
	out.phylip2}}
{{re	Renames sequences s1, s2 .... sN. Useful for long sequence names}}
{{h 	Prints documentation.}}
{{v 	verbosity
	1}}
{{range	Restricts the set of sequences written to the output file to a range.\
		For example, use -range=5,27 to use the fifth to 27th sequences \
		in the input file.}}
{{s		superclean. Changes all characters found in either names or \
		sequence not in 'ACGTN\\n -1234567890S' to N.}}
{{version   print version info and exit}}

%(version)s
"""
	debug = 0
	
	docstringdict = {'filename':os.path.split(sys.argv[0])[-1],
	'version':'$Id: fasta2phylip.py,v 1.1 2004/10/10 02:53:10 nghoffma Exp $'}
	
	optlist, formattedDocString, formattedSummary = cp.optStringParser(main.__doc__ % docstringdict,
		optWidth=15, lineWidth=60, valOffset = 4, putVals=1)	
	dict = cp.commandparser(options=optlist, usage='Options:\n' + formattedSummary, 
		debug = debug, exitWithUsage=1)
		
	if dict.status('version'):
		sys.exit( 'Version: %(version)s\n' % docstringdict )
		
	if dict.status('h'):
		print formattedDocString
		sys.exit( 0 )
	v = dict.value('v')
	superclean = dict.status('s')
		
	outfilename = dict.value('out')
		
	rename = dict.status('re')
	if dict.status('range'):
		range = dict.value('range').split(',')
		if len(range) != 2:
			sys.exit('The range was not specified properly')
		# convert to 0-index
		first = eval(range[0]) - 1
		last = eval(range[1])
		
	inlist = dict.value('infile_list','in','file')
	
	if v > 1:
		print 'Reading sequences from', inlist
	
	fasta = SeqIO.readFastaList(inlist)
	
	if v> 1:
		print 'A total of %s sequences were read.' % len(fasta)
		
	if v>0: print 'Writing %s ...' % outfilename
	
	outfile = open(outfilename,'w')
	
	if dict.status('range'):
		print 'Using sequences %s to %s.' % (range[0],range[1])
		outStr = SeqIO.writePhylip( fasta[first:last], rename=rename, superclean=superclean, v=0 )
	else:
		outStr = SeqIO.writePhylip( fasta, rename=rename, superclean=superclean, v=0 )
	
	outfile.write(outStr)
	outfile.close()
	
	
	
	
	
	
		
main()