Python线程间如何通信
2023-04-18 12:50:26 时间
我对于线程这部分知识比较薄弱,并发是一个复杂的问题,在测试过程中很少用到这个知识点。 由于全局解释器锁 (GIL),CPU 绑定任务不适合 Python 线程。Python 中的并行计算应在多个进程(而不是线程)中完成。
实例: 工作线程从队列中获取目录名称, 然后递归查找其中的所有文件并返回结果
import os, time
import threading, Queue
class WorkerThread(threading.Thread):
""" A worker thread that takes directory names from a queue, finds all
files in them recursively and reports the result.
Input is done by placing directory names (as strings) into the
Queue passed in dir_q.
Output is done by placing tuples into the Queue passed in result_q.
Each tuple is (thread name, dirname, [list of files]).
Ask the thread to stop by calling its join() method.
"""
def __init__(self, dir_q, result_q):
super(WorkerThread, self).__init__()
self.dir_q = dir_q
self.result_q = result_q
self.stoprequest = threading.Event()
def run(self):
# As long as we weren't asked to stop, try to take new tasks from the
# queue. The tasks are taken with a blocking 'get', so no CPU
# cycles are wasted while waiting.
# Also, 'get' is given a timeout, so stoprequest is always checked,
# even if there's nothing in the queue.
while not self.stoprequest.isSet():
try:
dirname = self.dir_q.get(True, 0.05)
filenames = list(self._files_in_dir(dirname))
self.result_q.put((self.name, dirname, filenames))
except Queue.Empty:
continue
def join(self, timeout=None):
self.stoprequest.set()
super(WorkerThread, self).join(timeout)
def _files_in_dir(self, dirname):
""" Given a directory name, yields the names of all files (not dirs)
contained in this directory and its sub-directories.
"""
for path, dirs, files in os.walk(dirname):
for file in files:
yield os.path.join(path, file)
def main(args):
# Create a single input and a single output queue for all threads.
dir_q = Queue.Queue()
result_q = Queue.Queue()
# Create the "thread pool"
pool = [WorkerThread(dir_q=dir_q, result_q=result_q) for i in range(4)]
# Start all threads
for thread in pool:
thread.start()
# Give the workers some work to do
work_count = 0
for dir in args:
if os.path.exists(dir):
work_count += 1
dir_q.put(dir)
print 'Assigned %s dirs to workers' % work_count
# Now get all the results
while work_count > 0:
# Blocking 'get' from a Queue.
result = result_q.get()
print 'From thread %s: %s files found in dir %s' % (
result[0], len(result[2]), result[1])
work_count -= 1
# Ask threads to die and wait for them to do it
for thread in pool:
thread.join()
if __name__ == '__main__':
import sys
main(sys.argv[1:])
池中的所有工作线程共享相同的输入队列和输出队列。这绝对没有问题。相反,正如您所看到的,它使线程池的简单实现具有相当的功能。
相关文章
- Hadoop0.20更新笔记
- 我们一起分析下BL(B)/LDR指令
- Hadoop入门经典
- 手把手教你用Pycharm连接远程Python环境
- 鸿蒙移植树莓派(上)搭建环境下载源码
- 微软:Microsoft Defender 将默认全自动执行补救措施
- 终于来了,OpenAI 正式开放 ChatGPT API
- [云原生] Kubernetes(k8s)健康检查详解与实战演示(就绪性探针 和 存活性探针)
- OpenAI 宣布开放 API,开发人员可将 ChatGPT 集成到自己的产品中
- 想要学好 Python,这七个 Github 仓库你值得拥有!
- F5 NGINX:API蔓延问题出现的六大迹象
- 如何使用NLP库解析Python中的文本
- 瑞数信息入选Gartner《Hype Cycle for Security in China, 2022》云安全资源池代表厂商
- 如何用 Python 玩转 ChatGPT
- 云上 OLAP 引擎查询性能评估框架:设计与实现
- 答网友问:Await 一个 Promise 对象到底发生了什么
- 为什么适用于Python的TensorFlow正在缓慢消亡
- 如何扩展Kubernetes API?
- 开源和云原生技术如何使API策略现代化
- 前后端分离开发模式下后端质量的保证 —— 单元测试