textGenerator.txt

def sample_words(prob_list,word_list):
    u = random()
    #print (sum(prob_list))
    p = 0
      
    for i in range(0,len(prob_list)):
        if p > u:
            return word_list[i-1]
        p += prob_list[i]
    return word_list[-1]

def bigram_sent(lm,sent):
    last_word = sent[-1]
    
    next_word_prob = []
    next_words = []
    
    for i in lm:
        if i[0] == last_word:
            next_word_prob.append(lm[i])
            next_words.append(i[1:])
    
    next_word = sample_words(next_word_prob,next_words)
    
    if next_word == None:
        raise NotImplementedError ('Next word is none!')
    
    sent.append(next_word[0])
    return sent

def trigram_sent(lm,sent):
    last_two_words = sent[-2:]
    
    next_word_prob = []
    next_words = []
    
    for i in lm:
        if last_two_words == list(i)[:2]:
            next_word_prob.append(lm[i])
            next_words.append(i[-1])
    next_word = sample_words(next_word_prob,next_words)
    
    if next_word == None:
        raise NotImplementedError ('Next word is none!')
        
    sent.append(next_word)
    
    return sent

def generate_sent(lm, N):
    ## YOUR CODE HERE ##
    
    if N < 2 or N > 4:
        raise NotImplementedError ('N should be 2, 3 or 4!')
        
    start = '<s>'
    end = '</s>'
    
    sentence = [start]
    
    if N == 4:
        sentence = bigram_sent(bigram,sentence)
        sentence = trigram_sent(trigram,sentence)
    elif N == 3:
        sentence = bigram_sent(bigram,sentence)
    
    looking_back = N-1
    previous_word = sentence[-looking_back:]
    
    while previous_word != end: 
        last_part = sentence[-looking_back:]
        next_words = []
        next_word_prob = []
                
        for i in lm:
            if last_part == list(i)[:looking_back]:
                next_word_prob.append(lm[i])
                next_words.append(i[-1])
        print (next_words)
        next_word = sample_words(next_word_prob,next_words)
        
        if next_word == None:
            raise NotImplementedError ('Next word is none!')
        
        sentence.append(next_word)
        print (sentence)
        
        if next_word == '.':
            sentence.append(end)
            return sentence
        
        if next_word == end:
            return sentence
        
N = 4
k = 0
bigram, unigram = train_ngram(data, 2, k)
print ('Done training bigram')
trigram, unigram = train_ngram(data, 3, k)
print ('Done training trigram')
quadgram,unigram = train_ngram(data , 4 , k)
print ('Done training')
sentence = generate_sent(quadgram,N)
sentence_str = str()

for i in sentence[1:-1]:
    sentence_str += i + ' '

print (sentence_str)