Saving tweets into MongoDB using Tweepy and PyMongo - TBI004 - FATEC SJC

saving_tweets_mongodb.py

# Saving tweets into MongoDB using Tweepy and PyMongo
# Author: Felipe Koblinger
import tweepy
from pymongo import MongoClient
from datetime import datetime

# Application Settings
consumer_key = "replace with your consumer key"
consumer_secret = "replace with your consumer secret"
# Your Access Token
access_token = "replace with your access token"
access_token_secret = "replace with your access token secret"

# API from OAuth
auth = tweepy.OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_token, access_token_secret)
api = tweepy.API(auth)

class StreamListener(tweepy.StreamListener):
    def on_status(self, status):
        connection = MongoClient("mongodb://localhost")
        database = connection.tweets

        retweet = status._json.get("retweeted_status")
        
        # If is a Retweet or a commom tweet
        if retweet:
            if retweet.get('full_text'):
                tweet_type = 'retweet_extended'
                tweet_text = retweet.get('full_text')
            else:
                tweet_type = 'retweet'
                tweet_text = retweet.get('text')
        else:
            extended_tweet = status._json.get('extended_tweet')
            if extended_tweet and 'full_text' in extended_tweet:
                tweet_type = 'tweet_extended'
                tweet_text = status.extended_tweet['full_text']
            else:
                tweet_type = 'tweet'
                tweet_text = status.text
        
        tweet = { "_id": status.id_str,
                  "tweet_@": status.user.screen_name,
                  "tweet_type": tweet_type,
                  "tweet_text": tweet_text,
                  "tweet_created_at": status.created_at,
                  "created_at": datetime.utcnow() }
        database.ruby.insert(tweet)
        

# Stream
stream = tweepy.Stream(auth = api.auth,
                       listener=StreamListener(), tweet_mode='extended')
stream.filter(track=['ruby'])