多进程、多线程处理文件对比

分别通过多进程、多线程方式处理文件，将结果保存到一个list中：

1.多进程：

import multiprocessing,cjson,os,collections
from multiprocessing import Process,freeze_support,Manager,Pool,Queue

def handlefile(lock,rst,fp):
    lst_tmp=[]
    #print type(rst)
    with open(fp,‘rb‘) as fo:
        for line in fo:
            line = cjson.decode(line)
            lst_tmp.append(line[‘s-ip‘])
    #print collections.Counter(lst_tmp)
    lock.acquire()
    rst.extend(lst_tmp)
    lock.release()

if __name__ == ‘__main__‘:
    lock = Manager().Lock()
    rst = Manager().list()

    starttime = datetime.datetime.now()
    f1 = ‘e:\\logtest\\iis__20160519105745.json‘
    f2 = ‘e:\\logtest\\iis__20160519105816.json‘
    f3 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134616.json‘
    f4 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134955.json‘
    f5 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134616.json‘
    f6 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134955.json‘
    files = [f1,f2,f3,f4,f5,f6]
    p=Pool()
    for file in files:
        p.apply_async(handlefile,args=(lock,rst,file))
    p.close()
    p.join()

    print collections.Counter(rst)

    print (datetime.datetime.now() - starttime).total_seconds() #耗时16.631s

2.多线程：

import threading
global rst
rst = []
def query(mutex,fp):
    lst_tmp=[]
    #print type(rst)
    with open(fp,‘rb‘) as fo:
        for line in fo:
            line = cjson.decode(line)
            lst_tmp.append(line[‘s-ip‘])
    #print collections.Counter(lst_tmp)
    mutex.acquire()
    rst.extend(lst_tmp)
    mutex.release()

if __name__ == ‘__main__‘:
    threads=[]
    mutex=threading.Lock()
    starttime = datetime.datetime.now()
    f1 = ‘e:\\logtest\\iis__20160519105745.json‘
    f2 = ‘e:\\logtest\\iis__20160519105816.json‘
    f3 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134616.json‘
    f4 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134955.json‘
    f5 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134616.json‘
    f6 = ‘e:\\logtest\\iis_IDC-ExFE01_20160524134955.json‘
    files = [f1,f2,f3,f4,f5,f6]

    for filepath in files:
        t = threading.Thread(target=query,args=(mutex,filepath))
        t.setDaemon(True)
        t.start()
        threads.append(t)
    for t in threads:
        t.join()

    print collections.Counter(rst)

    print (datetime.datetime.now() - starttime).total_seconds() #耗时4.425s

结论：多进程和多线程在分别处理每个文件，将结果写入各自tmp list中，多线程耗时2.468s，多线程耗时4.24s，多进程优于多线程（进程数量未控制，默认CPU核心数量）。

但当多线程各结果写入到共享变量list()时，多线程严重耗时较久，多线程共计耗时4.425s，多进程耗时16.631s。多进程中的共享变量效率低下。