#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Felix Muzny
10/25/2022
DS 2000
Lecture 14 - movie review program

A program to read and analyze some movie reviews.

Using the movie review data (blackadam.txt and tickettoparadise.txt), write
a larger program that will read in each review for the given movie, line-by-line
and does three things:
    1) Lets the user ask how many times a word exists in the reviews
    2) Reports the combined word counts for all reviews for that movie
"""

# we'll probably do some plotting
import matplotlib.pyplot as plt

# can also switch to "tickettoparadise.txt"
REVIEW_FILE = "blackadam.txt"

def read_data(filename):
    """
    Reads text data assuming that you have one review per line.
    Splits on whitespace.

    Parameters
    ----------
    filename : string
        file path to be read in.

    Returns
    -------
    List of reviews/lines
    in the input file.
    """
    file = open(filename, "r")
    reviews = []
    for line in file:
        # split the review on whitespace
        reviews.append(line.strip())
    # close the file in the same function that we opened it in
    file.close()
    return reviews

def count_words(text):
    """
    Count each word in a given string.

    Parameters
    ----------
    text : str
        words separated by whitespace.

    Returns
    -------
    dict of word counts.
    """
    words = text.split()
    counts = {}
    for word in words:
        if word in counts:
            counts[word] += 1
        else:
            counts[word] = 1
    return counts

def add_dicts(d1, d2):
    """
    Create a new dictionary that adds values for keys that exist in
    both dictionaries together.

    Parameters
    ----------
    d1 : dict
        First dictionary to add. Values must be updateable using +.
    d2 : dict
        Second dictionary to add. Types should match first dictionary.

    Returns
    -------
    final_dict : dict
        Final dictionary that is the result of adding our first two 
        dictionaries together.
    """
    final_dict = d1.copy()
    for k in d2:
        if k in final_dict:
            final_dict[k] += d2[k]
        else:
            final_dict[k] = d2[k]
    return final_dict
    

def main():
    print("We'll implement this!")
    # First, let's write down a roadmap
    
    # load in the data
    data = read_data(REVIEW_FILE)
    # make sure we've loaded it in properly
    print(len(data))
    print(data[:2])
    print()
    
    # count up the words
    total_counts = {}
    for review in data:
        counts = count_words(review)
        
        # combine the counts into a new dictionary of overall counts
        total_counts = add_dicts(total_counts, counts)
    
    # 1) ask the user about a word
    target_word = input("What word do you want to search for? ")
    if target_word in total_counts:
        print(target_word, "occurred:", total_counts[target_word])
    else:
        print(target_word, "did not occur!")
    print()
    
    # 2) Report all the combined word counts (formatted nicely, of course!)
    for key in total_counts:
        print(key, ":", total_counts[key])    

    
    
# from now on, we'll always use this "guard" for our call to
# our main function!
if __name__ == "__main__":
    main()