#!/bin/python

################################################################################
#
#	MarkovChar 2.21.09 - Joe Mariglio <www.joemariglio.com>
#
#===============================================================================
#
#	analyzes a text file for transitions and produces a dictionary
#	units are measured in chars, not words
#	variable depth of memory units (in chars)
#	uses the dictionary to generate a markov chain of a length (in units)
#
#===============================================================================
#
#	usage: markovChar.py <inFile> <depth> <length>
#
################################################################################

import re, sys, string, os.path, getopt, random
dic = {}
def parse(inpath):
	infile = open(inpath, "r")
	lines = infile.read()
	return lines
def analyze(lines, depth):
	line = lines.replace('\n', '')
	i = 0
	for char in line:
		if char in dic.keys():
			end = i + depth +1
			if len(line) > end:
				addContext(char, line[i+1:end])
		else:
			end = i + depth+1
			if len(line) > end:
				dic[char] = [line[i+1:end]]
		i = i + 1
def addContext(token, context):
	dic[token].append(context)
def generate(depth, length):
	start = random.choice(dic.keys())
	output = ""
	for i in range(0,length):
		value = random.choice(dic[start])
		start = value[depth-1]
		output = output + value
	return output
def main(argv):
	if len(sys.argv) == 4:
		inpath = sys.argv[1]
		depth = int(sys.argv[2])
		length = int(sys.argv[3])
		lines = parse(inpath)
		analyze(lines, depth)
		output = generate(depth, length)
		#output = random.choice(dic["a"])
		print output
	else:
		print "usage: markovChar.py <inFile> <depth> <length>"
		print "where <depth> is number of chars per unit"
		print "and <length> is the output length in units."
if __name__ == '__main__':
	import re, sys, string, os.path, getopt
	main(sys.argv[1:])