'''
    DS2000    
    Spring 2023
    Sample code from class - March 3rd
    
    We have the transript from the totally unhinged
    Bing AI chatbot conversation with a NY Times reporter.
    https://mashable.com/article/microsoft-bing-ai-chatbot-weird-scary-responses
    
    What was the word it used most frequently?
'''

FILENAME = "chatbot.txt"

def read_txt(filename):
    ''' Function: read_txt
        Parameter: filename, a string
        Returns: list of strings
        Does: iterate over each line of the file,
                split on spaces and append each word to a list
    '''
    words = []
    with open(filename, "r") as infile:
        for line in infile:
            words += line.split()
    return words
    
def wordcount(lst):
    ''' FUnction: wordcount
        Parameter: list of strings, including repeats
        returns: a ditionary, where key = word and value - count
        Does: iterate over the list of words, updating
              the value in the dictionary as we go
    '''
    wc = {}
    for word in lst:
        if word in wc:
            wc[word] += 1
        else:
            wc[word] = 1
    return wc

def find_max(dct):
    ''' Function: find_max
        Parameters: dictionary where key is anythig,
                    but value must be a number
        Return: key, value pair where value is higheset
        Does: iterate over given dictoinary, find highest
              value and return both the value and its key
    '''
    max_val = -1
    max_key = ""
    for key, value in dct.items():
        if value > max_val:
            max_val = value
            max_key = key
    return max_key, max_val
    

def main():
    # Gather data - read in the text file to a 1d list
    words = read_txt(FILENAME)
    
    # Computation - turn the list of words into a dictionary
    dct = wordcount(words)
    
    # Computation - find the top 10 most common words
    # Keep a list of common words, and stop when we hit 10
    most_common = []
    while len(most_common) < 10:
        max_word, max_count = find_max(dct)
        most_common.append(max_word)
        del(dct[max_word])
        
    # Communiation - print our list of top-10 common words
    print(most_common)
 
    
main()