Streaming twitter data to an S3 bucket using firehose - python

I am trying to stream data from twitter to an aws bucket. The good news is I can get the data to stream to my bucket but the data comes in approx 20 kb chunks (I think this may be due to some firehose settings) and its not saving as JSON even after I specified it to in my python code using JSON.LOAD. Rather than saving as JSON, the data in my S3 bucket looks like it does not have a file extension and has long string of alphanumeric characters. I think it may be something to do with the parameters being used in client.put_record()
Any help is greatly appreciated!
Please find my code below, which I got from github here.
from tweepy.streaming import StreamListener
from tweepy import OAuthHandler
from tweepy import Stream
import json
import boto3
import time
#Variables that contains the user credentials to access Twitter API
consumer_key = "MY_CONSUMER_KEY"
consumer_secret = "MY_CONSUMER_SECRET"
access_token = "MY_ACCESS_TOKEN"
access_token_secret = "MY_SECRET_ACCESS_TOKEN"
#This is a basic listener that just prints received tweets to stdout.
class StdOutListener(StreamListener):
def on_data(self, data):
tweet = json.loads(data)
try:
if 'extended_tweet' in tweet.keys():
#print (tweet['text'])
message_lst = [str(tweet['id']),
str(tweet['user']['name']),
str(tweet['user']['screen_name']),
tweet['extended_tweet']['full_text'],
str(tweet['user']['followers_count']),
str(tweet['user']['location']),
str(tweet['geo']),
str(tweet['created_at']),
'\n'
]
message = '\t'.join(message_lst)
print(message)
client.put_record(
DeliveryStreamName=delivery_stream,
Record={
'Data': message
}
)
elif 'text' in tweet.keys():
#print (tweet['text'])
message_lst = [str(tweet['id']),
str(tweet['user']['name']),
str(tweet['user']['screen_name']),
tweet['text'].replace('\n',' ').replace('\r',' '),
str(tweet['user']['followers_count']),
str(tweet['user']['location']),
str(tweet['geo']),
str(tweet['created_at']),
'\n'
]
message = '\t'.join(message_lst)
print(message)
client.put_record(
DeliveryStreamName=delivery_stream,
Record={
'Data': message
}
)
except (AttributeError, Exception) as e:
print (e)
return True
def on_error(self, status):
print (status)
if __name__ == '__main__':
#This handles Twitter authetification and the connection to Twitter Streaming API
listener = StdOutListener()
auth = OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_token, access_token_secret)
#tweets = Table('tweets_ft',connection=conn)
client = boto3.client('firehose',
region_name='us-east-1',
aws_access_key_id='MY ACCESS KEY',
aws_secret_access_key='MY SECRET KEY'
)
delivery_stream = 'my_firehose'
#This line filter Twitter Streams to capture data by the keywords: 'python', 'javascript', 'ruby'
#stream.filter(track=['trump'], stall_warnings=True)
while True:
try:
print('Twitter streaming...')
stream = Stream(auth, listener)
stream.filter(track=['brexit'], languages=['en'], stall_warnings=True)
except Exception as e:
print(e)
print('Disconnected...')
time.sleep(5)
continue

Its possible that you have enabled S3 compression for your firehose. Please ensure that the compression is disabled if you want to store raw json data in your bucket:
You could also have some transformation applied to your firehose which encode or otherwise transform your json messages into some other format.

So it looks like the files were coming on with JSON formatting, i just had to open the files in S3 with firefox and i could see the contents of files. The issue with the file sizes is due to the firehose buffer settings, i have them set to the lowest which is why files were being sent in such small chunks

Related

how do i transfer live twitter steam to mongo db in python

I need to make an app that will live stream tweets from twitter api to one of two things. I can either have them save to a file then from the file save them to my mongo db database, or I can have it go from the stream directly to the database.
Right now the program will save it to a file and print it in the terminal but I can not figure out where I would do the transfer. if i put the insert one in the on_status it just jumps out as soon as it hits that line and if i put it in the main area, it never does it.
from tweepy import Stream
from tweepy import OAuthHandler
import tweepy
import json
import pymongo
#tokens and access
access_token = ''
access_token_secret = ''
API_KEY = ''
API_KEY_secret = ''
#make connection to database
connection = pymongo.MongoClient('localhost', 27017)
#create the database and collection
database = connection.twitter_db
collection = database.tweets.create_index([("id", pymongo.ASCENDING)], unique=True,)
class MyListener (tweepy.Stream):
def on_status(self, status):
json_str=json.dumps(status._json)
print(json_str)
try:
with open("UkraineTweets.json","a") as file:
file.write(json_str + "\n")
print("written to the file")
#collection.insert_one(data)
#print("entered into mongo_db")
return True
except:
pass
return False
if __name__ == "__main__":
authentication = tweepy.OAuthHandler(API_KEY,API_KEY_secret)
authentication.set_access_token(access_token, access_token_secret)
api = tweepy.API(authentication)
stream = MyListener(API_KEY, API_KEY_secret, access_token, access_token_secret)
stream.filter(track=['Russia', 'Ukraine', 'war', 'Putin', '#StandWithUkraine' ,'#StopPutinNOW'], languages=['en'])''''

Handle 420 response code returned by Tweepy api

Whenever a user logs in to my application and searches I have to start a streaming API for fetching data required by him.
Here is my stream API class
import tweepy
import json
import sys
class TweetListener(tweepy.StreamListener):
def on_connect(self):
# Called initially to connect to the Streaming API
print("You are now connected to the streaming API.")
def on_error(self, status_code):
# On error - if an error occurs, display the error / status code
print('An Error has occured: ' + repr(status_code))
return False
def on_data(self, data):
json_data = json.loads(data)
print(json_data)
Here is my python code file which calls class above to start Twitter Streaming
import tweepy
from APIs.StreamKafkaApi1 import TweetListener
consumer_key = "***********"
consumer_secret = "*********"
access_token = "***********"
access_secret = "********"
hashtags = ["#ipl"]
def callStream():
auth = tweepy.OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_token, access_secret)
api = tweepy.API(auth,wait_on_rate_limit=True)
tweetListener = TweetListener(userid,projectid)
streamer = tweepy.Stream(api.auth, tweetListener)
streamer.filter(track=hashtags, async=True)
if __name__ == "__main__":
callStream()
But if I hit more than twice my application return error code 420.
I thought to change API(using multiple keys) used to fetch data whenever Error 420 occurs.
How to get error raised by the on_error method of TweetListener class in def callStream()
I would like to add onto #Andy Piper's answer. Response 420 means your script is making too many requests and has been Rate Limited. To resolve this, here is what I do(in class TweetListener):
def on_limit(self,status):
print ("Rate Limit Exceeded, Sleep for 15 Mins")
time.sleep(15 * 60)
return True
Do this and the error will be handled.
If you persist on using multiple keys. I am not sure but try exception handling on TweetListener and streamer, for tweepy.error.RateLimitError and use recursive call of the function using next API key?
def callStream(key):
#authenticate the API keys here
try:
tweetListener = TweetListener(userid,projectid)
streamer = tweepy.Stream(api.auth, tweetListener)
streamer.filter(track=hashtags, async=True)
except tweepy.TweepError as e:
if e.reason[0]['code'] == "420":
callStream(nextKey)
return True
Per the Twitter error response code documentation
Returned when an application is being rate limited for making too many
requests.
The Twitter streaming API does not support more than a couple of connections per user and IP address. It is against the Twitter Developer Policy to use multiple application keys to attempt to circumvent this and your apps could be suspended if you do.

Python API Streaming, write new file after certain size

I have a python script that maintains an open connection to the Twitter Streaming API, and writes the data into a json file. Is it possible to write to a new file, without breaking the connection, after the current file being written reaches a certain size? For example, I just streamed data for over 1 week, but all the data is contained in a single file (~2gb) making it slow to parse. If I could write to a new file after, say 500mb, then I would have 4 smaller files (e.g. dump1.json, dump2.json etc) to parse instead of one large one.
import tweepy
from tweepy import OAuthHandler
from tweepy import Stream
from tweepy.streaming import StreamListener
# Add consumer/access tokens for Twitter API
consumer_key = '-----'
consumer_secret = '-----'
access_token = '-----'
access_secret = '-----'
auth = OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_token, access_secret)
api = tweepy.API(auth)
# Define streamlistener class to open a connection to Twitter and begin consuming data
class MyListener(StreamListener):
def on_data(self, data):
try:
with open('G:\xxxx\Raw_tweets.json', 'a') as f:
f.write(data)
return True
except BaseException as e:
print("Error on_data: %s" % str(e))
return True
def on_error(self, status):
print(status)
return True
bounding_box = [-77.2157,38.2036,-76.5215,39.3365]#filtering by location
keyword_list = ['']#filtering by keyword
twitter_stream = Stream(auth, MyListener())
twitter_stream.filter(locations=bounding_box) # Filter Tweets in stream by location bounding box
#twitter_stream.filter(track=keyword_list) # Filter Tweets in stream by keyword
Since you re-open your file every time, it is rather simple - use an index in file name and advance it if your file size reaches threshold
class MyListener(StreamListener):
def __init(self):
self._file_index = 0
def on_data(self, data):
tweets_file = 'G:\xxxx\Raw_tweets{}.json'.format(self._file_index)
while os.path.exists(tweets_file) and os.stat(tweet_file).st_size > 2**10:
self._file_index += 1
tweets_file = 'G:\xxxx\Raw_tweets{}.json'.format(self._file_index)
....
The cycle will take care of your app being restarted

Mine Tweets between two dates in Python

I would like to mine tweets for two keywords for a specific period of time. I currently have the code below, but how do I add so it only mine tweets between two dates? (10/03/2016 - 10/07/2016) Thank you!
#Import the necessary methods from tweepy library
from tweepy.streaming import StreamListener
from tweepy import OAuthHandler
from tweepy import Stream
#Variables that contains the user credentials to access Twitter API
access_token = "ENTER YOUR ACCESS TOKEN"
access_token_secret = "ENTER YOUR ACCESS TOKEN SECRET"
consumer_key = "ENTER YOUR API KEY"
consumer_secret = "ENTER YOUR API SECRET"
#This is a basic listener that just prints received tweets to stdout.
class StdOutListener(StreamListener):
def on_data(self, data):
print data
return True
def on_error(self, status):
print status
if __name__ == '__main__':
#This handles Twitter authetification and the connection to Twitter Streaming API
l = StdOutListener()
auth = OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_token, access_token_secret)
stream = Stream(auth, l)
#This line filter Twitter Streams to capture data by the keywords: 'python', 'javascript', 'ruby'
stream.filter(track=['python', 'javascript', 'ruby'])
You can't. Have a look at this question, that is the closest you can get.
The Twitter API does not allow to search by time. Trivially, what you can do is fetching tweets and looking at their timestamps afterwards in Python, but that is highly inefficient.

Twitter API - get tweets with specific id

I have a list of tweet ids for which I would like to download their text content. Is there any easy solution to do this, preferably through a Python script? I had a look at other libraries like Tweepy and things don't appear to work so simple, and downloading them manually is out of the question since my list is very long.
You can access specific tweets by their id with the statuses/show/:id API route. Most Python Twitter libraries follow the exact same patterns, or offer 'friendly' names for the methods.
For example, Twython offers several show_* methods, including Twython.show_status() that lets you load specific tweets:
CONSUMER_KEY = "<consumer key>"
CONSUMER_SECRET = "<consumer secret>"
OAUTH_TOKEN = "<application key>"
OAUTH_TOKEN_SECRET = "<application secret"
twitter = Twython(
CONSUMER_KEY, CONSUMER_SECRET,
OAUTH_TOKEN, OAUTH_TOKEN_SECRET)
tweet = twitter.show_status(id=id_of_tweet)
print(tweet['text'])
and the returned dictionary follows the Tweet object definition given by the API.
The tweepy library uses tweepy.get_status():
auth = tweepy.OAuthHandler(CONSUMER_KEY, CONSUMER_SECRET)
auth.set_access_token(OAUTH_TOKEN, OAUTH_TOKEN_SECRET)
api = tweepy.API(auth)
tweet = api.get_status(id_of_tweet)
print(tweet.text)
where it returns a slightly richer object, but the attributes on it again reflect the published API.
Sharing my work that was vastly accelerated by the previous answers (thank you). This Python 2.7 script fetches the text for tweet IDs stored in a file. Adjust get_tweet_id() for your input data format;
original configured for data at https://github.com/mdredze/twitter_sandy
Update April 2018: responding late to #someone bug report (thank you). This script no longer discards every 100th tweet ID (that was my bug). Please note that if a tweet is unavailable for whatever reason, the bulk fetch silently skips it. The script now warns if the response size is different from the request size.
'''
Gets text content for tweet IDs
'''
# standard
from __future__ import print_function
import getopt
import logging
import os
import sys
# import traceback
# third-party: `pip install tweepy`
import tweepy
# global logger level is configured in main()
Logger = None
# Generate your own at https://apps.twitter.com/app
CONSUMER_KEY = 'Consumer Key (API key)'
CONSUMER_SECRET = 'Consumer Secret (API Secret)'
OAUTH_TOKEN = 'Access Token'
OAUTH_TOKEN_SECRET = 'Access Token Secret'
# batch size depends on Twitter limit, 100 at this time
batch_size=100
def get_tweet_id(line):
'''
Extracts and returns tweet ID from a line in the input.
'''
(tagid,_timestamp,_sandyflag) = line.split('\t')
(_tag, _search, tweet_id) = tagid.split(':')
return tweet_id
def get_tweets_single(twapi, idfilepath):
'''
Fetches content for tweet IDs in a file one at a time,
which means a ton of HTTPS requests, so NOT recommended.
`twapi`: Initialized, authorized API object from Tweepy
`idfilepath`: Path to file containing IDs
'''
# process IDs from the file
with open(idfilepath, 'rb') as idfile:
for line in idfile:
tweet_id = get_tweet_id(line)
Logger.debug('get_tweets_single: fetching tweet for ID %s', tweet_id)
try:
tweet = twapi.get_status(tweet_id)
print('%s,%s' % (tweet_id, tweet.text.encode('UTF-8')))
except tweepy.TweepError as te:
Logger.warn('get_tweets_single: failed to get tweet ID %s: %s', tweet_id, te.message)
# traceback.print_exc(file=sys.stderr)
# for
# with
def get_tweet_list(twapi, idlist):
'''
Invokes bulk lookup method.
Raises an exception if rate limit is exceeded.
'''
# fetch as little metadata as possible
tweets = twapi.statuses_lookup(id_=idlist, include_entities=False, trim_user=True)
if len(idlist) != len(tweets):
Logger.warn('get_tweet_list: unexpected response size %d, expected %d', len(tweets), len(idlist))
for tweet in tweets:
print('%s,%s' % (tweet.id, tweet.text.encode('UTF-8')))
def get_tweets_bulk(twapi, idfilepath):
'''
Fetches content for tweet IDs in a file using bulk request method,
which vastly reduces number of HTTPS requests compared to above;
however, it does not warn about IDs that yield no tweet.
`twapi`: Initialized, authorized API object from Tweepy
`idfilepath`: Path to file containing IDs
'''
# process IDs from the file
tweet_ids = list()
with open(idfilepath, 'rb') as idfile:
for line in idfile:
tweet_id = get_tweet_id(line)
Logger.debug('Enqueing tweet ID %s', tweet_id)
tweet_ids.append(tweet_id)
# API limits batch size
if len(tweet_ids) == batch_size:
Logger.debug('get_tweets_bulk: fetching batch of size %d', batch_size)
get_tweet_list(twapi, tweet_ids)
tweet_ids = list()
# process remainder
if len(tweet_ids) > 0:
Logger.debug('get_tweets_bulk: fetching last batch of size %d', len(tweet_ids))
get_tweet_list(twapi, tweet_ids)
def usage():
print('Usage: get_tweets_by_id.py [options] file')
print(' -s (single) makes one HTTPS request per tweet ID')
print(' -v (verbose) enables detailed logging')
sys.exit()
def main(args):
logging.basicConfig(level=logging.WARN)
global Logger
Logger = logging.getLogger('get_tweets_by_id')
bulk = True
try:
opts, args = getopt.getopt(args, 'sv')
except getopt.GetoptError:
usage()
for opt, _optarg in opts:
if opt in ('-s'):
bulk = False
elif opt in ('-v'):
Logger.setLevel(logging.DEBUG)
Logger.debug("main: verbose mode on")
else:
usage()
if len(args) != 1:
usage()
idfile = args[0]
if not os.path.isfile(idfile):
print('Not found or not a file: %s' % idfile, file=sys.stderr)
usage()
# connect to twitter
auth = tweepy.OAuthHandler(CONSUMER_KEY, CONSUMER_SECRET)
auth.set_access_token(OAUTH_TOKEN, OAUTH_TOKEN_SECRET)
api = tweepy.API(auth)
# hydrate tweet IDs
if bulk:
get_tweets_bulk(api, idfile)
else:
get_tweets_single(api, idfile)
if __name__ == '__main__':
main(sys.argv[1:])
You can access tweets in bulk (up to 100 at a time) with the status/lookup endpoint: https://dev.twitter.com/rest/reference/get/statuses/lookup
I don't have enough reputation to add an actual comment so sadly this is the way to go:
I found a bug and a strange thing in chrisinmtown answer:
Every 100th tweet will be skipped due to the bug. Here is a simple solution:
if len(tweet_ids) < 100:
tweet_ids.append(tweet_id)
else:
tweet_ids.append(tweet_id)
get_tweet_list(twapi, tweet_ids)
tweet_ids = list()
Using is better since it works even past the rate limit.
api = tweepy.API(auth_handler=auth, wait_on_rate_limit=True, wait_on_rate_limit_notify=True)

Categories