Python 爬虫（2）多线程

前面说过由于GIL的存在，Python的多线程效率没有希望的那么高，python的多线程适合IO密集型的情况，而爬虫恰好就是一个IO密集的情况，因为爬虫中很大一部分时间，是在等待socket返回数据。

下面写一个例子：

import requests
import time

if __name__ == '__main__':
    codes = ['sh600993', 'sh000006', 'sh600658', 'sh600153', 'sh600005']
    start = time.time()
    for code in codes:
        url = 'http://hq.sinajs.cn/list=' + code
        response = requests.get(url).text
        print response
    print time.time() - start

　　

var hq_str_sh600993="马应龙,20.020,20.090,20.060,20.060,19.950,20.040,20.060,486809,9740634.000,2100,20.040,8300,20.030,1300,20.020,2300,20.010,4100,20.000,101,20.060,10000,20.070,14400,20.080,19000,20.090,25700,20.100,2017-01-24,11:30:00,00";

var hq_str_sh000006="地产指数,6567.8364,6574.1060,6568.6375,6577.7249,6542.6599,0,0,1486830,1392918131,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,2017-01-24,11:35:51,00";

var hq_str_sh600658="电子城,13.320,13.200,13.270,13.320,13.040,13.270,13.280,559733,7389992.000,30800,13.270,300,13.220,6200,13.200,2500,13.100,4900,13.090,9300,13.280,6400,13.290,8200,13.300,6900,13.310,9000,13.320,2017-01-24,11:30:00,00";

var hq_str_sh600153="建发股份,10.520,10.510,10.500,10.540,10.460,10.490,10.500,4834159,50730040.000,32800,10.490,60100,10.480,186000,10.470,181241,10.460,125800,10.450,56600,10.500,105500,10.510,108400,10.520,110400,10.530,139900,10.540,2017-01-24,11:30:00,00";

var hq_str_sh600005="武钢股份,0.000,3.710,3.710,0.000,0.000,0.000,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,2017-01-24,11:30:00,03";

0.110999822617

　　换成多线程之后：

import requests
import threading
import time

def get_stock(code):
    url = 'http://hq.sinajs.cn/list=' + code
    response = requests.get(url).text
    # js_info = response.read()
    print response
    
if __name__ == '__main__':
    codes = ['sh600993', 'sh000006', 'sh600658', 'sh600153', 'sh600005']
    start = time.time()
    threads = [threading.Thread(target = get_stock,args = (code,)) for code in codes]
    for t in threads:
        t.start()
    for t in threads:
        t.join()
    print time.time()-start

　　

var hq_str_sh600993="马应龙,20.020,20.090,20.060,20.060,19.950,20.040,20.060,486809,9740634.000,2100,20.040,8300,20.030,1300,20.020,2300,20.010,4100,20.000,101,20.060,10000,20.070,14400,20.080,19000,20.090,25700,20.100,2017-01-24,11:30:00,00";

var hq_str_sh600658="电子城,13.320,13.200,13.270,13.320,13.040,13.270,13.280,559733,7389992.000,30800,13.270,300,13.220,6200,13.200,2500,13.100,4900,13.090,9300,13.280,6400,13.290,8200,13.300,6900,13.310,9000,13.320,2017-01-24,11:30:00,00";

var hq_str_sh000006="地产指数,6567.8364,6574.1060,6568.6375,6577.7249,6542.6599,0,0,1486830,1392918131,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,2017-01-24,11:35:51,00";

var hq_str_sh600153="建发股份,10.520,10.510,10.500,10.540,10.460,10.490,10.500,4834159,50730040.000,32800,10.490,60100,10.480,186000,10.470,181241,10.460,125800,10.450,56600,10.500,105500,10.510,108400,10.520,110400,10.530,139900,10.540,2017-01-24,11:30:00,00";

var hq_str_sh600005="武钢股份,0.000,3.710,3.710,0.000,0.000,0.000,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,0,0.000,2017-01-24,11:30:00,03";

0.0379998683929

　　速度有了很大的提升

线程池

import requests
import threadpool
import time

def get_stock(code):
    url = 'http://hq.sinajs.cn/list=' + code
    response = requests.get(url).text
    # js_info = response.read()
    print response
    
if __name__ == '__main__':
    codes = ['sh600993', 'sh000006', 'sh600658', 'sh600153', 'sh600005']
    start = time.time()
    pool = threadpool.ThreadPool(5)
    tasks = threadpool.makeRequests(get_stock,codes)
    [pool.putRequest(task) for task in tasks]
    pool.wait()
    print time.time() - start

threadpool.ThreadPool定义了一个线程池，表示可以创建4个线程；

makeRequests创建了要开启多线程的函数，已经函数的参数以及回调函数，回调函数callback可以不写,默认是无。

[pool.putRequest(task) for task in tasks]是将所有多线程的请求扔进了线程池，等价于

for code in codes:
    pool.putRequest(code)

　pool.wait()是等待所有工作完成后退出。这里执行的数量还比较少，基本的多线程就够用了，当数量多了起来之后，线程池的效果会好一些。