Tweepy with multiple access tokens

This fork merges the current version of Tweepy (3.5.0) with the multiple authentication functionality from the fork by Alexandru Stancui, available here: https://github.com/svven/tweepy

At the time of writing, there exists an issue with the Twitter API whereby the reset time for a resource, returned either in the response header as x-rate-limit-reset, or by the function api.rate_limit_status(), is incorrect. This is currently handled in the master and this fork by continuing a loop (binder.py, line 204).

Changes

• Multiple access tokens with RateLimitHandler (https://github.com/svven/tweepy/blob/master/tweepy/limit.py)

RateLimitHandler class inherits from OAuthHandler, and introduces add_access_token that can be used as follows:

from tweepy import RateLimitHandler
from config import CONSUMER_KEY, CONSUMER_SECRET, ACCESS_TOKENS

def get_api():
	auth = RateLimitHandler(CONSUMER_KEY, CONSUMER_SECRET)
	for key, secret in ACCESS_TOKENS:
		try:
			auth.add_access_token(key, secret)
		except Exception, e:
			print key, e
	print 'Token pool size: %d' % len(auth.tokens)
	api = API(auth,
		wait_on_rate_limit=True, wait_on_rate_limit_notify=True)
	return api

api = get_api()

Provided access tokens are used selectively based on requested resource (https://dev.twitter.com/docs/rate-limiting/1.1/limits) and current rate limits. The access token with most remaining requests per window for the specified resource is being selected and used when applying the authentication, before the actual request is performed. This pattern ensures the usage of available access tokens in a round robin fashion, exploiting to maximum the rate limits.

Installation

pip install -e git+https://github.com/rosscg/tweepy.git#egg=tweepy