import tweepy
import csv
import json
from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer
import string
from matplotlib import pyplot as plt
import numpy as np
import math
import statistics as stats

with open('twitter_credentials.json') as cred_data:
    info = json.load(cred_data)
    consumer_key = info['CONSUMER_KEY']
    consumer_secret = info['CONSUMER_SECRET']
    access_key = info['ACCESS_KEY']
    access_secret = info['ACCESS_SECRET']

auth = tweepy.OAuthHandler(consumer_key, consumer_secret)
api = tweepy.API(auth)

maximum_number_of_tweets_to_be_extracted = \
int(input('Enter the number of tweets that you want to extract - '))

hashtag = input('Enter the keyword(s) you want to scrape - ').replace(", ", " OR ")
lines = [["Tweet", "Score", "Location", "User"]]
analyzer = SentimentIntensityAnalyzer()
scores = []

for tweet in tweepy.Cursor(api.search, q= hashtag, geocode="37.752721,-122.327076,30mi", rpp=100).items(maximum_number_of_tweets_to_be_extracted):
    loc = tweet.user.location
    user = tweet.user.name
    tweet = str(tweet.text.encode("utf-8")).strip("'b")
    tweet = tweet.translate(string.punctuation)
    line = [tweet]
    score = analyzer.polarity_scores(tweet)["compound"]
    scores.append(score)
    line.append(score)
    line.append(loc)
    line.append(user)
    lines.append(line)


with open(hashtag+'.csv', 'w') as csvFile:
    writer = csv.writer(csvFile)
    writer.writerows(lines)

csvFile.close()

plt.xlim([-1, 1])
plt.hist(scores, bins=20, alpha=0.5)
plt.title('Sentiment Analysis Score Distribution for keyword(s) ' + hashtag.replace(" OR ", ", "))
plt.xlabel('Score Interval')
plt.ylabel('Count')

print("Average Score: " + str(stats.mean(scores)))
print("Median Score: " + str(stats.median(scores)))
plt.show()