uburuntu / throttler

🔀⏳ Easy throttling with asyncio support

Throttler

Zero-dependency Python package for easy throttling with asyncio support.

📝 Table of Contents

• 🎒 Install
• 🛠 Usage Examples
  • Throttler and ThrottlerSimultaneous
    • Simple Example
    • API Example
  • ExecutionTimer
  • Timer
• 👨🏻‍💻 Author
• 💬 Contributing
• 📝 License

🎒 Install

Just

pip install throttler

🛠 Usage Examples

All run-ready examples are here.

Throttler and ThrottlerSimultaneous

Throttler:

Context manager for limiting rate of accessing to context block.

from throttler import Throttler

# Limit to three calls per second
t = Throttler(rate_limit=3, period=1.0)
async with t:
    pass

Or

import asyncio

from throttler import throttle

# Limit to three calls per second
@throttle(rate_limit=3, period=1.0)
async def task():
    return await asyncio.sleep(0.1)

ThrottlerSimultaneous:

Context manager for limiting simultaneous count of accessing to context block.

from throttler import ThrottlerSimultaneous

# Limit to five simultaneous calls
t = ThrottlerSimultaneous(count=5)
async with t:
    pass

Or

import asyncio

from throttler import throttle_simultaneous

# Limit to five simultaneous calls
@throttle_simultaneous(count=5)
async def task():
    return await asyncio.sleep(0.1)

Simple Example

import asyncio
import time

from throttler import throttle


# Limit to two calls per second
@throttle(rate_limit=2, period=1.0)
async def task():
    return await asyncio.sleep(0.1)


async def many_tasks(count: int):
    coros = [task() for _ in range(count)]
    for coro in asyncio.as_completed(coros):
        _ = await coro
        print(f'Timestamp: {time.time()}')

asyncio.run(many_tasks(10))

Result output:

Timestamp: 1585183394.8141203
Timestamp: 1585183394.8141203
Timestamp: 1585183395.830335
Timestamp: 1585183395.830335
Timestamp: 1585183396.8460555
Timestamp: 1585183396.8460555
...

API Example

import asyncio
import time

import aiohttp

from throttler import Throttler, ThrottlerSimultaneous


class SomeAPI:
    api_url = 'https://example.com'

    def __init__(self, throttler):
        self.throttler = throttler

    async def request(self, session: aiohttp.ClientSession):
        async with self.throttler:
            async with session.get(self.api_url) as resp:
                return resp

    async def many_requests(self, count: int):
        async with aiohttp.ClientSession() as session:
            coros = [self.request(session) for _ in range(count)]
            for coro in asyncio.as_completed(coros):
                response = await coro
                print(f'{int(time.time())} | Result: {response.status}')


async def run():
    # Throttler can be of any type
    t = ThrottlerSimultaneous(count=5)        # Five simultaneous requests
    t = Throttler(rate_limit=10, period=3.0)  # Ten requests in three seconds

    api = SomeAPI(t)
    await api.many_requests(100)

asyncio.run(run())

Result output:

1585182908 | Result: 200
1585182908 | Result: 200
1585182908 | Result: 200
1585182909 | Result: 200
1585182909 | Result: 200
1585182909 | Result: 200
1585182910 | Result: 200
1585182910 | Result: 200
1585182910 | Result: 200
...

ExecutionTimer

Context manager for time limiting of accessing to context block. Simply sleep period secs before next accessing, not analog of Throttler. Also it can align to start of minutes.

import time

from throttler import ExecutionTimer

et = ExecutionTimer(60, align_sleep=True)

while True:
    with et:
        print(time.asctime(), '|', time.time())

Or

import time

from throttler import execution_timer

@execution_timer(60, align_sleep=True)
def f():
    print(time.asctime(), '|', time.time())

while True:
    f()

Result output:

Thu Mar 26 00:56:17 2020 | 1585173377.1203406
Thu Mar 26 00:57:00 2020 | 1585173420.0006166
Thu Mar 26 00:58:00 2020 | 1585173480.002517
Thu Mar 26 00:59:00 2020 | 1585173540.001494

Timer

Context manager for pretty printing start, end, elapsed and average times.

import random
import time

from throttler import Timer

timer = Timer('My Timer', verbose=True)

for _ in range(3):
    with timer:
        time.sleep(random.random())

Or

import random
import time

from throttler import timer

@timer('My Timer', verbose=True)
def f():
    time.sleep(random.random())

for _ in range(3):
    f()

Result output:

#1 | My Timer | begin: 2020-03-26 01:46:07.648661
#1 | My Timer |   end: 2020-03-26 01:46:08.382135, elapsed: 0.73 sec, average: 0.73 sec
#2 | My Timer | begin: 2020-03-26 01:46:08.382135
#2 | My Timer |   end: 2020-03-26 01:46:08.599919, elapsed: 0.22 sec, average: 0.48 sec
#3 | My Timer | begin: 2020-03-26 01:46:08.599919
#3 | My Timer |   end: 2020-03-26 01:46:09.083370, elapsed: 0.48 sec, average: 0.48 sec

👨🏻‍💻 Author

Ramzan Bekbulatov:

• Telegram: @rm_bk
• Github: @uburuntu

💬 Contributing

Contributions, issues and feature requests are welcome!

📝 License

This project is MIT licensed.

throttler's Issues

Updating ratelimit and period during runtime

Would the implementation support an updating of the ratelimit during runtime depending on the responses from the API?

[Demo request] Is that possible to run with multiprocessing?

Thanks for this awesome lib. It makes throttling easily.

I have many heavy jobs which not only needs to be throttled through HTTP requests, but also needs to calculate the results heavily (CPU-bound).

The README shows the concurrent way with throttler. Is that possible running jobs on multi-processes with throttler?

For example, using joblib:

import asyncio
import time

from joblib import Parallel, delayed
from throttler import throttle

# Limit to two calls per second
@throttle(rate_limit=1, period=0.5)
async def task(i):
    print(f"{i}: {time.time()} start")
    await asyncio.sleep(5)
    print(f"{i}: {time.time()} end")
    return i


async def many_tasks(count: int):
    print("=== START ===")
    results = Parallel(n_jobs=-1)(delayed(task)(i) for i in range(count))
    print(results)
    print("=== END ===")

asyncio.run(many_tasks(14))

The above code shows error:

...
TypeError: cannot pickle 'coroutine' object

from source installation fails because `readme.md` is missing

I'm running into the following when using pip install using the source tarball for throttle 0.2.2 obtained from PyPI:

    Running command python setup.py egg_info
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/tmp/eb-pc17jo6j/pip-req-build-o1s1r0pd/setup.py", line 43, in <module>
        long_description=read('readme.md'),
      File "/tmp/eb-pc17jo6j/pip-req-build-o1s1r0pd/setup.py", line 10, in read
        with open(filename, encoding='utf-8') as file:
    FileNotFoundError: [Errno 2] No such file or directory: 'readme.md'
WARNING: Discarding file:///tmp/vsc40023/easybuild_build/snakemake/7.18.2/foss-2021b/throttler/throttler-1.2.1. Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output.
ERROR: Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output.

The problem is that readme.md is not included in the source tarball.