Python非同步需要使用到asyncio,简单测试如下
import asyncioimport requestsimport jsonimport timefrom datetime import datetimeasync def syncHello(delay): print("Hello:",delay) await asyncio.sleep(delay) print("Bye:",delay)if __name__ == "__main__": ts = [5,3,1 ] now = lambda: time.time() start = now() tasks = [syncHello(i) for i in ts] asyncio.run(asyncio.wait(tasks)) print('Total TIME:', now() - start)
代码执行后,在await时,就会转去执行其他task,然而实务上你直接用改成API呼叫是不会work的,因为呼叫API并没有使用到await,以至于没有效果(会卡住等待Response),所以这时候就要再多使用 loop.run_in_executor,让呼叫API时在等待Response时,转去继续执行其他task
话不多说,上代码!
方法如下:
import asyncioimport requestsimport jsonimport timeimport uuidfrom datetime import datetimeloop = asyncio.get_event_loop()async def doBuyBook(bookName):logkey = uuid.uuid4().hex[:6]print(f"[{logkey}] doBuyBook.start ==> {bookName}")req = {"BookName": bookName}#非同步执行r = await loop.run_in_executor(None, lambda: requests.post("http://127.0.0.1:8321/bookstore/buybook", json = req))response_data = json.loads(r.text)print(f"[{logkey}] response_data:{response_data}")print(f"[{logkey}] doBuyBook.end")if __name__ == "__main__": bookNames = ["PYTHON_DEV","JAVA_DEV","GOLANG_DEV","C#_DEV" ] now = lambda: time.time() start = now() tasks = [loop.create_task(doBuyBook(i)) for i in bookNames] loop.run_until_complete(asyncio.wait(tasks)) print('Total TIME: ', now() - start)
使用 get_event_loop 建立迴圈事件 loop,使用 run_in_executor 把呼叫API POST丢入线程池中执行,在await时,就会转去执行下一个task,当API回应时间较慢时,会明显看到打印出来的LOG,会先全部打印出,然后才会依序打印出API的Response
大概就是这样,Cheer!! 测试用的API如何建立写在下一篇