# Import necessary libraries
import random
import numpy as np
from helper import softmax, get_weights

# Function to predict the next set of characters which forms the dinosaur name
def rnn_model(parameters, char_to_ix):

    # Get the weights and biases from the parameters dictionary
    U, V, W = parameters['U'], parameters['V'], parameters['W']
    beta1, beta2 = parameters['beta1'], parameters['beta2']

    # Get the size of the vocabulary i.e. 27
    # One for each alphabet plus the new line character
    vocab_size = beta2.shape[0]

    # Get the size of the weights
    n_h = U.shape[1]

    # Initialize the input as an array of zeroes with size as (vocab_size,1)
    # This one-hot encodes the input
    x = ___

    # Initialize the inital hidden state as an array of zeroes with size as (n_h,1)    
    h_prev = ___

    # Initialize a list to store the indices of the predicted characters
    indices = []
    
    # Initialize an idx variable to hold the index values of the characters 
    idx = -1 
    
    # Initialize a counter to fix the maximum length of the predicted word
    counter = 0

    # Get the value of the new line from the char_to_ix dictionary
    newline_character = char_to_ix['\n']
    
    # Loop until the newline_character is predicted or until the max length of the word is 50
    while (idx != newline_character and counter != 50):

        # Compute the new state h of the RNN unit using the equation given in the instructions
        h = ___

        # Compute the output of the RNN unit using the equation 
        # given in the instructions using the softmax function
        y = softmax(___)

        # Get the index value of the predicted/generated character
        # Instead of taking the argmax, we perform sampling on the probabilities 
        # got from the softmax function
        idx = np.random.choice(list(range(vocab_size)), p=y.ravel())

        # Append the index value to the indices list
        indices.append(idx)
        
        # Initialize an array of with zeroes with size (vocab_size,1)
        x = np.zeros((vocab_size, 1))

        # Set the idx position of x as 1.
        # This will act as the output y and the next input. 
        x[idx] = 1
        
        # Update the previous state value with the current state
        h_prev = ___
        
        # Increment the counter
        counter+=1
    
    # If the counter value reaches 50 append a newline character to the indices list
    if (counter == 50):
        indices.append(char_to_ix['\n'])
    
    # Return the list of indices
    return indices

# Read the dinos.txt file
data = open('dinos.txt', 'r').read()

# Convert the data to lower case
data= data.lower()

# Convert the file data into list
chars = list(set(data))

# Get length of the file and length of the vocabulary
data_size, vocab_size = len(data), len(chars)

# Define a dictionary with the sorted vocabulary as key and 
# value as a unique integer using enumerate
char_to_ix = ___

# Define a dictionary with the unique integers assigned to the sorted vocabulary as key
# and value as a unique integer using enumerate
ix_to_char = ___

# Call the get_weights function to get the model weights
# To get random weights set random=1
# To get the trained weights specify the number of iterations and set random=0
parameters = get_weights(num_iterations=1000, random=0)

# Call the predict function defined above passing 
# the parameters dictionary, char_to_ix dictionary
sampled_indices = rnn_model(parameters, char_to_ix)

# Convert the list of indices returned by the predict function to 
# their respective characters and then join to form a word
txt = ___

# Captializing the first character
txt = txt[0].upper() + txt[1:]

# Print the generated dinosaur name 
print('%s' % (txt, ), end='')

### edTest(test_chow1) ###
# Submit an answer choice as a string below (eg. if you choose option A, put 'A')
answer1 = '___'

### edTest(test_chow2) ###
# Type your answer within in the quotes given
answer2 = '___'