here is repository for python generative text workshop:

Here's the markov.py module:

 
# Copyright 2005 Brendan Howell
# All rights reserved under the GPL: see http://www.fsf.org
 
# markov.py
# implements a markov class to generate psuedo text based on
#  training from input.	 based on shaney.py by Greg McFarlane
#  and Joe Strout.
# search on "Mark V. Shaney" and "Markov Chains" for more info.
 
import sys
import random
import cPickle
 
class Markov:
    def __init__(self):
        self.dict = {}
        self.freq = {}
        self.ends = []
 
    def learn_text(self, learntext):
        words = learntext.split()
        prev1 = ''
        prev2 = ''
        for word in words:
            if prev1 != '' and prev2 != '':
                key = (prev2, prev1)
                if key in self.dict:
                    self.dict[key].append(word)
                else:
                    self.dict[key] = [word]
                    if prev1[-1:] == '#':
                        self.ends.append(key)
                if word in self.freq:
                    self.freq[word] += 1
                else:
                    self.freq[word] = 1
            prev2 = prev1
            prev1 = word
 
    def learn_file(self, infile):
        file = open(infile)
        text = file.read()
        file.close()
        self.learn_text(text)
 
    def load_bin(self, binfile):
        return cPickle.load(open(binfile))
 
    def save_bin(self, outfile):
        cPickle.dump(self,open(outfile,'w'))
 
    def print_sen(self):
        key = ()
 
        while True:
            if key in self.dict:
                word = random.choice(self.dict[key])
                print word,
                key = (key[1], word)
                if key in self.ends:
                    print
                    break
            else:
                key = random.choice(self.ends)
 
    def get_joke(self):
        key = ()
        rtn = ""
 
        while True:
            if key in self.dict:
                word = random.choice(self.dict[key])
                if word != "#":
                    rtn += word + " "
                key = (key[1], word)
                if key in self.ends:
                    rtn += "\n"
                    break
            else:
                key = random.choice(self.ends)
 
        return rtn
 
 
    def print_sens(self, n):
        for count in range(n):
            self.print_sen()
 
    def respond(self, intext):
        n = 3 * random.betavariate(1,3)
        print self.sens_context(n,intext)
 
    def responds(self, intext):
        n = 3 * random.betavariate(1,3)
        return self.sens_context(n,intext)
 
    def sen_context(self, intext):
        key = ()
        inwords = intext.split()
        words = ''
 
        while True:
            if key in self.dict:
                context = []
                for inword in inwords:
                    if inword in self.dict[key]:
                        context.append(inword)
                if context:
                    word = random.choice(context)
                else:
                    word = random.choice(self.dict[key])
                words += word + ' '
                key = (key[1], word)
                if key in self.ends:
                    break
            else:
                key = random.choice(self.ends)
 
        return words
 
    def sens_context(self, n, intext):
        words = ''
        for count in range(n):
            words += self.sen_context(intext) + ' '
        return words
 
 
    def print_sen_context(self, intext):
        key = ()
        inwords = intext.split()
 
        while True:
            if key in self.dict:
                context = []
                for inword in inwords:
                    if inword in self.dict[key]:
                        context.append(inword)
                if context:
                    word = random.choice(context)
                else:
                    word = random.choice(self.dict[key])
                print word,
                key = (key[1], word)
                if key in self.ends:
                    print
                    break
            else:
                key = random.choice(self.ends)
 
 
    def print_sens_context(self, n, intext):
        for count in range(n):
            self.print_sen_context(intext)