# Install packages
import config
import requests
import json
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import datetime

sns.set(style="darkgrid")

# Set keys and links
OAUTH = config.OAUTH
CLIENT_ID = config.CLIENT_ID
CLIENT_SECRET = config.CLIENT_SECRET

BASE_URL = {"v5":"https://api.twitch.tv/kraken/",
            "new":"https://api.twitch.tv/helix/"}
HEADERS = {"Authorization":"Bearer {}".format(OAUTH),
           "client-id":CLIENT_ID}

# Get response from API call
def get_response(query, params = {}, version = "new"):
    response = requests.get(BASE_URL[version] + query,
                            headers = HEADERS,
                            params = params)
    return response
    
# Get user info
# https://dev.twitch.tv/docs/api/reference#get-users
def get_query_user_info(login):
    return "users?login={}".format(login)

# Get video info
# https://dev.twitch.tv/docs/api/reference#get-videos
def get_query_video_info(user_id):
    return "videos?user_id={}".format(user_id)

# List OTV members
otv_members = ["scarra","pokimane","lilypichu","disguisedtoast","yvonnie","michaelreeves"]

# Get otv member info
otv_member_info = pd.DataFrame()

for member in otv_members:
    member_json = get_response(get_query_user_info(member)).json()['data']
    member_df = pd.json_normalize(member_json)
    otv_member_info = otv_member_info.append(member_df)
    
otv_member_info

# Get otv video data
otv_videos = pd.DataFrame()

for id in otv_member_info.id:
    videos_json = get_response(get_query_video_info(id), params = {"first":"100"}).json()['data']
    videos_df = pd.json_normalize(videos_json)
    otv_videos = otv_videos.append(videos_df)

# Clean created_at date
otv_videos["created_at"] = pd.to_datetime(otv_videos["created_at"]).dt.normalize().dt.date

# Calculate total minutes streamed per video
otv_videos["hours"] = otv_videos["duration"].str.extract("(\d*)(?=[h])").fillna(0).astype(float)
otv_videos["minutes"] = otv_videos["duration"].str.extract("(\d*)(?=[m])").fillna(0).astype(float)
otv_videos["seconds"] = otv_videos["duration"].str.extract("(\d*)(?=[s])").fillna(0).astype(float)
otv_videos["total_minutes"] = otv_videos["hours"]*60 + otv_videos["minutes"] + otv_videos["seconds"]/60

# Calculate total minutes streamed per day
otv_videos_sum = otv_videos.groupby(["user_name","created_at"]).sum().reset_index()

# Sort index by mean
sorted_index = otv_videos_sum.groupby("user_name")["total_minutes"].median().sort_values(ascending=False).index

# Plot boxplot
plt.subplots(figsize=(12, 5))
ax = sns.boxplot(data = otv_videos_sum,
            x = "user_name", 
            y = "total_minutes",
            order = sorted_index,
            palette = sns.color_palette("Paired"))
ax.tick_params(rotation=-45)
plt.title("Distribution of Time Spent Streaming per Session")
plt.xlabel("Streamer")
plt.ylabel("Minutes spent streaming")

# Filter date range
min_date = otv_videos_sum[otv_videos_sum["user_name"] == "pokimane"]["created_at"].min()
max_date = otv_videos_sum[otv_videos_sum["user_name"] == "pokimane"]["created_at"].max()
date_range = pd.date_range(min_date, max_date, freq = "D")
otv_videos_filtered = otv_videos_sum[otv_videos_sum["created_at"] >= min_date]

# Fill missing dates
otv_videos_filled = (otv_videos_filtered.set_index("created_at")
                                          .groupby("user_name")
                                          .apply(lambda x: x.reindex(date_range))
                                          .drop("user_name", axis = 1)
                                          .reset_index()
                                          .rename(columns = {"level_1":"created_at"})
                                          .fillna(0))

# Calculate cumulative sum of minutes streamed for each otv member
otv_videos_filled["total_minutes_cumsum"] = otv_videos_filled.groupby("user_name")["total_minutes"].cumsum()

# Pivot data for plotting
otv_videos_pivot = otv_videos_filled[["user_name","total_minutes_cumsum","created_at"]].pivot(index = "created_at", 
                                                                     columns = "user_name", 
                                                                     values = "total_minutes_cumsum")

# Plot cummulative minutes 
fig = plt.figure()
otv_videos_pivot.plot(figsize=(12, 5))

plt.title("Cumulative Stream Time of Saved Videos")
plt.xlabel("Date")
plt.ylabel("Cumulative minutes")

plt.show()