将asyncio.get_event_loop().run_until_complete替换为ayncio.run

Question

我正在尝试运行这些异步测试，我在这个库的根目录node run-tests gateio --python-async中通过命令行运行这些测试来运行它们

下面的代码导致了一个弃用警告，因为我使用的是依赖于python3.10的东西(tox(，并且看起来get_event_loop与python3.10一起被弃用

if __name__ == '__main__':
asyncio.get_event_loop().run_until_complete(main())

折旧警告

[100%] Testing gateio WARN (testExchange @ run-tests.js:172)
WARN gateio (Python 3 Async): (explain @ run-tests.js:191)
/path/ccxt/python/ccxt/test/test_async.py:558: DeprecationWarning: There is no current event loop
asyncio.get_event_loop().run_until_complete(main())
gateio using proxy ``
EXCHANGE: gateio
SYMBOL: BTC/USD
CODE: ZRX
gateio BTC/USD ticker None high: 65609.0 low: 64073.0 bid: 64958.24 ask: 64991.5 volume: 40525928.149905
gateio fetched all 2176 tickers
gateio fetched 10 OHLCVs
gateio BTC/USD order book 2021-11-12T03:25:31.759Z bid: 64958.24 bidVolume: 0.3225 ask: 64991.5 askVolume: 0.3225
gateio BTC/USD fetched 100 trades
gateio BTC/USD fetch_orders() not supported
gateio BTC/USD fetched 0 open orders
gateio BTC/USD fetched 0 closed orders
gateio ZRX fetch_transactions() not supported
gateio fetched balance (explain @ run-tests.js:193)
WARN ["gateio"] (run-tests.js:272)
All done, 1 warnings (run-tests.js:276)

---

我试着更新这一行，改为使用asyncio.run，但后来我得到了一个无限循环，最终测试超时，失败

asyncio.run(main(), debug=True)

超时错误

[100%] Testing gateio FAIL (testExchange @ run-tests.js:172)
FAILED gateio (Python 3 Async): (explain @ run-tests.js:190)
timed out (explain @ run-tests.js:193)
FAIL ["gateio"] (run-tests.js:271)
All done, 1 failed (run-tests.js:276)

---

我想知道如何正确更新此方法，以避免出现弃用警告和超时

Answer 1

这里的问题在于如何管理事件循环。asyncio.run总是在内部创建一个全新的事件循环。您正在使用的API似乎在Exchange类中使用了asyncio.get_event_loop()，该类创建了与asyncio.run创建并传递该事件循环不同的事件循环。这会导致问题，因为Exchange类创建的循环只是闲置在那里，所以尝试使用该循环运行任何东西都不会真正运行。要解决此问题，您需要确保使用asyncio.run创建的循环，这意味着稍微重构代码并使用asyncio.get_running_loop()。asyncio.get_running_loop()将返回当前正在运行的循环或抛出异常。如果一个循环没有运行，它将永远不会创建新的循环。

要特别修复您的问题(请记住，此更改可能会破坏代码的其他区域(

将exchange.py中Exchange类的构造函数更改为使用asyncio.get_running_loop()而不是asyncio.get_event_loop()，如下所示：

class Exchange(BaseExchange):
def __init__(self, config={}):
if 'asyncio_loop' in config:
self.asyncio_loop = config['asyncio_loop']
self.asyncio_loop = asyncio.get_running_loop() #this is the change here, note that this overwrites the loop from the config, so this could cause problems.
self.aiohttp_trust_env = config.get('aiohttp_trust_env', self.aiohttp_trust_env)
self.verify = config.get('verify', self.verify)
self.own_session = 'session' not in config
self.cafile = config.get('cafile', certifi.where())
super(Exchange, self).__init__(config)
self.throttle = None
self.init_rest_rate_limiter()
self.markets_loading = None
self.reloading_markets = False

接下来，您需要将您的exchangeinit代码移动到main协程中。这样，当构造函数调用get_running_loop时，它将返回asyncio.run创建的循环：

async def main():
# instantiate all exchanges
for id in ccxt.exchanges:
if id == 'theocean':
continue
exchange = getattr(ccxt, id)
exchange_config = {'verbose': argv.verbose}
if sys.version_info[0] < 3:
exchange_config.update({'enableRateLimit': True})
if id in config:
exchange_config = ccxt.Exchange.deep_extend(exchange_config, config[id])
exchanges[id] = exchange(exchange_config)
#other code as before...

此修复有效，但请记住，API本身可能需要以支持其处理事件循环的方式进行重构，以支持asyncio.run创建和管理事件循环的方法。特别是，Exchange类中可以从配置中提取事件循环的代码可能需要重构。