#
# This file demonstrates using a basic class
#
# Conrad Huang
# November 2, 2004
#

# A FASTA file contains a series of sequences, each of which
# consists of a one-line description followed by any number of
# lines of nucleotide data.  The description line begins with
# the '>' characters.  The data lines consist of single-character
# nucleotide or amino acid symbols, but many programs (including
# this one) treat any lines that do not begin with '>' as data lines.

# FastaReader is an implementation of a FASTA file reader using
# classic constructs available in nearly all versions of Python.

# This module may be tested by executing it as a Python script.
# Command line arguments are treated as the names of FASTA files.
# If no FASTA file is specified, the default test case of "ncbi.fa" is used.

"Read sequences in FASTA format."

class FastaSequence:
	"A FASTA sequence and its description."
	
	def __init__(self, desc, seq):
		self.description = desc[1:].strip()
		self.sequence = seq

class FastaReader:
	"Read generic FASTA files.  (Classic version)"

	count = 0

	def __init__(self, f):
		import types
		if type(f) is types.StringType:
			# Argument is a string, treat as filename
			self.f = open(f)
			self.autoClose = True
			self.name = f
		elif type(f) is types.FileType:
			# Argument is a file object
			self.f = f
			self.autoClose = False
			self.name = "<file handle>"
		else:
			raise ValueError, "expecting string or file object"
		try:
			#self.description = self.f.next()
			self.description = self.f.readline()
		except StopIteration:
			self.close()
		else:
			self.finished = False
		FastaReader.count += 1

	def __del__(self):
		self.close()

	def getName(self):
		return self.name

	def nextSequence(self):
		if self.finished:
			return None
		# We have the sequence description from a previous
		# call or from __init__.  Now we read sequence data
		# until we hit the next description or the end of file.
		desc = self.description
		seq = []
		while 1:
			line = self.f.readline()
			if not line:
				self.close()
				break
			if line[0] == '>':
				self.description = line
				break
			else:
				seq.append(line.strip())
		return self.makeSequence(desc, ''.join(seq))

	def makeSequence(self, desc, seq):
		return FastaSequence(desc, seq)

	def close(self):
		if self.f and self.autoClose:
			self.f.close()
			self.f = None
		self.finished = True

if __name__ == "__main__":
	def readFasta(filename):
		fa = FastaReader(filename)
		while 1:
			fs = fa.nextSequence()
			if fs is None:
				break
			print fs.description, len(fs.sequence), fs.sequence
		print
	def main():
		import sys
		args = sys.argv[1:]
		if not args:
			readFasta("ncbi.fa")
		else:
			for filename in args:
				readFasta(filename)
	main()