'''
    DS2000
    Fall 2024

    Starter code for class on 11/8/24

    The starter code reads in chatGPT responses from a csv file (responses only, ignoring questions/timestamps)

    In class we'll fill in together the two function stubs below:
    - clean_response - given a chatGPT response (string), remove punctuation and turn all letters lowercase
    - sentiment_score - given a list of words (strings) and a dictionary of word : score,
                        compute and return a sentiment score from -1 to +1

'''

import matplotlib.pyplot as plt
import csv

TXTFILE = "chatgpt.csv"
RESPONSE_COL = 0

SENTS = {"good": .6, "great" : .8, "celebrated": 1, "happier": 1,
         "love": .5, "like": .5, "happy" : .5,
         "dislike": -.5, "fight": -.5, "sad": -1, "divisive": -.25,
         "conflict" : -.25, "angry" : -.3, "argument" : -.2, "hate" : -1}


def read_text(filename, resp_col):
    '''
        parameters: string for filename, int for col where reponse it
        returns: list of strings (sentences/comments)
        does: opens the given csv file and turns one column into a list of strings
    '''
    responses = []
    with open(filename, "r") as infile:
        csvfile = csv.reader(infile, delimiter = ",")
        next(csvfile)
        for row in csvfile:
            responses.append(row[resp_col])
    return responses

def clean_response(response):
    '''
        Parameters: a single response from chatGPT (a string)
        Returns: a string (copy/cleaned up version of input string)
        Does: creates a new string that matches the original, with punctuation removed and all lowercase
    '''
    clean_str = ""
    for letter in response:
        if letter.isalpha() or letter == " ":
            clean_str = clean_str + letter.lower()
    return clean_str


def sentiment_score(words, sentiments):
    '''
        parameters: a single response from chatGPT (a list of strings),
                    and a dictionary of word (str) : sentiment score (fl)
        Returns: float
        Does: Computes a sentiment score for the given response, based on values from the given dictionary
    '''
    score = 0
    for word in words:
        if word in sentiments:
            score += sentiments[word]
    score = score / len(words)
    return score

def main():
    # Step one: gather data by reading in the chatGPT text file and creating a list of responses
    sentences = read_text(TXTFILE, RESPONSE_COL)

    # Step two: clean up the strings to remove punctuation and turn lowercase
    for i in range(len(sentences)):
        sentences[i] = clean_response(sentences[i])

    # Step two: computation! For each response, compute its sentiment score (-1 to +1)
    scores = []
    for sentence in sentences:
        score = sentiment_score(sentence.split(), SENTS)
        # just for funsies, what was that one super-negative score we saw?
        if score < -0.01:
            print("That one negative response:", sentence)
        scores.append(score)

    # Step three: communication. Print or plot the scores.
    plt.hist(scores, color = "magenta")
    plt.title("Sentiment Scores for ChatGPT Responses")
    plt.xlabel("Score Range")
    plt.ylabel("Number of Responses")
    plt.show()

main()