Analyze discord usage

DiscordAnalysis.py

import json
import csv
from datetime import date
import numpy as np
from glob import glob
from matplotlib import pyplot as plt

join_date = date(2017, 8, 3)
path_to_package_directory = "/Users/kskrueger/PycharmProjects/DiscordLearning/Personal/package-2/"


class Conversation:
    def __init__(self, folderpath):
        self.messages = []
        with open(folderpath + '/messages.csv', encoding="utf8") as csv_file:
            csv_reader = csv.reader(csv_file, delimiter=',')
            for row in csv_reader:
                # Format: ID,Timestamp,Contents,Attachments
                self.messages.append(row)
        if len(self.messages) < 1:
            raise Exception("Empty file provided!")
        self.messages.pop(0)
        self.length = len(self.messages)
        self.words = 0
        for message in self.messages:
            self.words = self.words + len(str(message).split())

    def get_date_distribution(self):
        messages_dates = np.zeros((date.today() - join_date).days, np.int)
        for _, time, message, _ in self.messages:
            message_date = date(int(time[:4]), int(time[5:7]), int(time[8:10]))
            days = (message_date - join_date).days
            messages_dates[days] = messages_dates[days] + 1
        return messages_dates


with open(path_to_package_directory + 'messages/index.json', encoding="utf8") as json_data:
    name_map = json.load(json_data)
    json_data.close()

conversations = {}
all_running_total = np.zeros((date.today() - join_date).days, np.int)

total_words = 0
for x in glob(path_to_package_directory + 'messages/*/'):
    a = conversations[str(x[-19:-1])] = Conversation(x).get_date_distribution()
    total_words += Conversation(x).words
    all_running_total = all_running_total + a


# Plot function is used by others to plot all conversations in a plot_list variable
# total=False will plot the rate per day istead of running_total of messages
def plot(plot_list, total=True):
    if len(plot_list) < 1:
        print("Invalid list to plot")
        return
    plt.figure()
    legend_names = []
    for id, value in plot_list.items():
        if id in name_map.keys():
            name = str(name_map[id])
        else:
            name = id
        if 'with' in name:
            name = name[name.index('with') + 5:]
        legend_names.append(name)
        messages_day = [0]
        for x in value:
            messages_day.append(x + messages_day[-1] if total else x)
        plt.plot(range(0, len(messages_day) - 1), messages_day[:-1])
    plt.title("Activity by total")
    plt.xlabel("Days since joining discord")
    plt.ylabel("Total Messages")
    plt.legend(list(legend_names))
    plt.show()


# Plot only conversations that are over a certain threshold of messages (ex: plot_thresh(1000))
def plot_thresh(thresh, total=True):
    plot_list = {}
    for k, v in conversations.items():
        if sum(v) > thresh:
            plot_list[k] = v
    plot(plot_list, total)


# Plot only conversations of target usernames/channels
# (this is a rough matching for username/channel name and could break for duplicates?)
# ex: plot_names(['Karter', 'general-robotics', 'Ilan 9421#0431'])
def plot_names(names_in, total=True):
    plot_list = {}
    id_list = None
    for name_in in names_in:
        for id, name in name_map.items():
            if str(name_in) in str(name):
                id_list = id
        for k, v in conversations.items():
            if str(id_list) == str(k):
                plot_list[k] = v
    plot(plot_list, total)


# Plot a running total of all messages sent
plot({'ALL HISTORY': all_running_total})

# Plot daily rate graph (by adding total=False)
plot({'ALL HISTORY': all_running_total}, total=False)

# Examples
plot_thresh(3000)  # plot all conversations over 3000 message threshold
plot_names(['general-robotics', 'Karter'])  # plot these conversations

# Scary statistics of days without discord and words sent
days = 0
for day_total in all_running_total:
    if day_total < 1:
        days = days + 1
print("Days without going on discord: " + str(days))
print("Total words sent on discord: " + str(total_words))