线程中的 subprocess.Popen

Question

我有许多文件（超过 4000 个）要同时加载到 PostgreSQL 中。我已将它们分成 4 个不同的文件列表，我希望有一个线程遍历每个列表来加载数据。

我遇到的问题是我使用 os.system 来调用加载程序，但这会阻止其他线程同时运行。如果我使用 subprocess.Popen，那么它们会同时运行，但线程认为它们已完成执行，因此请转到我脚本的下一部分。

我这样做是否正确？或者有没有更好的方法从线程中调用子进程。

def thread1Load(self, thread1fileList):
    connectionstring = settings.connectionstring
    postgreshost = settings.postgreshost
    postgresdatabase = settings.postgresdatabase
    postgresport = settings.postgresport
    postgresusername = settings.postgresusername
    postgrespassword = settings.postgrespassword

    tablename = None
    encoding = None
    connection = psycopg2.connect(connectionstring)

    for filename in thread1fileList:
        load_cmd = #load command
        run = subprocess.Popen(load_cmd, shell=True)
    print "finished loading thread 1"


def thread2Load(self, thread2fileList):
    connectionstring = settings.connectionstring
    postgreshost = settings.postgreshost
    postgresdatabase = settings.postgresdatabase
    postgresport = settings.postgresport
    postgresusername = settings.postgresusername
    postgrespassword = settings.postgrespassword

    tablename = None

    connection = psycopg2.connect(connectionstring)
    for filename in thread2fileList:
        load_cmd = #load command            
        run = subprocess.Popen(load_cmd, shell=True)
    print "finished loading thread 2"


def thread3Load(self, thread3fileList):
    connectionstring = settings.connectionstring
    postgreshost = settings.postgreshost
    postgresdatabase = settings.postgresdatabase
    postgresport = settings.postgresport
    postgresusername = settings.postgresusername
    postgrespassword = settings.postgrespassword

    tablename = None
    connection = psycopg2.connect(connectionstring)

    for shapefilename in thread3fileList:
        load_cmd = #load command
        run = subprocess.Popen(load_cmd, shell=True)
    print "finished loading thread 3"

def thread4Load(self, thread4fileList):
    connectionstring = settings.connectionstring
    postgreshost = settings.postgreshost
    postgresdatabase = settings.postgresdatabase
    postgresport = settings.postgresport
    postgresusername = settings.postgresusername
    postgrespassword = settings.postgrespassword

    tablename = None

    connection = psycopg2.connect(connectionstring)

    for filename in thread4fileList:
        load_cmd = #load command
        run = subprocess.Popen(load_cmd, shell=True)

    print "finished loading thread 4"


def finishUp(self):
    print 'finishing up'


def main():
load = Loader()

thread1 = threading.Thread(target=(load.thread1Load), args=(thread1fileList, ))
thread2 = threading.Thread(target=(load.thread2Load), args=(thread2fileList, ))
thread3 = threading.Thread(target=(load.thread3Load), args=(thread3fileList, ))
thread4 = threading.Thread(target=(load.thread4Load), args=(thread4fileList, ))
threads = [thread1, thread2, thread3, thread4]
for thread in threads:
    thread.start()
    thread.join()


load.finishUp(connectionstring)

if __name__ == '__main__':
main()

Answer 1

• Don't repeat yourself。一种threadLoad 方法就足够了。这样，如果您需要修改方法中的某些内容，则无需在 4 个不同的地方进行相同的修改。
• 使用run.communicate() 阻塞直到子进程完成。

• 这会启动一个线程，然后阻塞直到该线程完成，然后 启动另一个线程等：

  for thread in threads:
      thread.start()
      thread.join()
  

  相反，先启动所有线程，然后加入所有线程：

  for thread in threads:
      thread.start()
  for thread in threads:
      thread.join()
  

---

import subprocess
import threading


class Loader(object):
    def threadLoad(self, threadfileList):
        connectionstring = settings.connectionstring
        ...
        connection = psycopg2.connect(connectionstring)

        for filename in threadfileList:
            load_cmd =  # load command
            run = subprocess.Popen(load_cmd, shell=True)
            # block until subprocess is done
            run.communicate()
        name = threading.current_thread().name
        print "finished loading {n}".format(n=name)

    def finishUp(self):
        print 'finishing up'


def main():
    load = Loader()
    threads = [threading.Thread(target=load.threadLoad, args=(fileList, ))
               for fileList in (thread1fileList, thread2fileList,
                                thread3fileList, thread4fileList)]
    for thread in threads:
        thread.start()
    for thread in threads:
        thread.join()

    load.finishUp(connectionstring)

if __name__ == '__main__':
    main()