import nltk
import random

f = open("/root/experiment/crying/crying")
cry = f.read()
tokens = nltk.word_tokenize(cry)
text = nltk.Text(tokens)
words = [w.lower() for w in text]

fdist1 = nltk.FreqDist(words)
vocab = fdist1.keys()
bigrams=nltk.bigrams(words)
cfd = nltk.ConditionalFreqDist(bigrams)

#TODO: save the frequency distribution

def markov_encode(cfdist, word, encodedword):
    wd = bitstring = ''
    addedbits=wert=tootal=xx=jj=rr=0
    encodedword+="x"
    loppy=(len(encodedword)*8)
    lippy=len(encodedword)
    while wert < lippy: # number of bits
        wd += " " + word
        # how many bits can we encode in next decision
        lenny = len(cfdist[word].items())
        while lenny<=1:
            # choose random next word
            word = random.choice(vocab)
            lenny = len(cfdist[word].items())
            wd = wd + " " + word
        bits = 1
        while (2 << bits) < lenny:
            bits+=1
        addedbits=0
        
        for j in range(bits):
            addedbits+=(ord(encodedword[wert])>>(rr%8)&1)<<j 
            bitstring += str(ord(encodedword[wert])>>(rr%8)&1)
            rr+=1
            xx+=1
            if (xx%8)==0:
                bitstring=''
                wert+=1
            if xx == loppy:
                break

        word =''.join(cfdist[word].items()[addedbits][:1])
        tootal+=bits
    return wd

def markov_decode(cfdist, text):
    newtext=text.split()
    decodedchar=j=xx=result=jj=rr=0
    bitstring = wd = ""
    for i in range(len(newtext)-1):
        lenny = len(cfdist[newtext[i]].items()) # bits are encoded for next word
        if lenny<=1:
            continue # goto next word
        bits = 1
        while (2 << bits) < lenny:
            bits+=1
        xx=0
        for x in cfdist[newtext[i]].items():
            yy = ''.join(x[:1])
            if yy == newtext[i+1]:
                result=xx
                xx = 0
                break
            xx+=1
        for j in range(bits):
            if (j+jj)%8 == 0:
                if (j+jj)>0:
                    wd+= chr(decodedchar)
                    bitstring=''
                    decodedchar=0
            decodedbit=(result>>j)&1
            bitstring += str(decodedbit)
            rr+=1
            decodedchar+=decodedbit<<((j+jj)%8)
        jj+=bits
    return wd