Inconsistent rate limit behavior

[jeremylow]

If the Api isn't instantiated with sleep_on_rate_limit=True, then the Api.rate_limit object doesn't actually decrement the remaining value on any of the end points, even after calling api.InitializeRateLimit(). Since that might be valuable information, it seems like we should retain it. First impression is that the offending code is L4821:

if url and self.sleep_on_rate_limit:
    limit = self.CheckRateLimit(url)

    if limit.remaining == 0:
        try:
            time.sleep(int(limit.reset - time.time()))
        except ValueError:
            pass

Seems to me that the if url and self.sleep_on_rate_limit check should be if url and self.rate_limit and the if limit.remaining should be if limit.remaining and self.sleep_on_rate_limit so that if the user wants to have access to the rate_limit object, they can do so by calling api.InitializeRateLimit() and then they'd have access to the rate_limit object and it would properly keep track of its resource limits.

[jeremylow]

One more place (L4851):

# could just be if url and self.rate_limit:
if url and self.sleep_on_rate_limit and self.rate_limit:
    limit = resp.headers.get('x-rate-limit-limit', 0)
    remaining = resp.headers.get('x-rate-limit-remaining', 0)
    reset = resp.headers.get('x-rate-limit-reset', 0)

    self.rate_limit.set_limit(url, limit, remaining, reset)

[bear]

Your analysis and solution both make sense - good job!