Python多线程、异步+多进程爬虫实现代码
这篇文章主要介绍了Python多线程、异步+多进程爬虫实现代码,需要的朋友可以参考下
安装Tornado
省事点可以直接用grequests库,下面用的是tornado的异步client。 异步用到了tornado,根据官方文档的例子修改得到一个简单的异步爬虫类。可以参考下最新的文档学习下。
pip install tornado
异步爬虫
#!/usr/bin/env python
# -*- coding:utf-8 -*-
import time
from datetime import timedelta
from tornado import httpclient, gen, ioloop, queues
import traceback
class AsySpider(object):
"""A simple class of asynchronous spider."""
def __init__(self, urls, concurrency=10, **kwargs):
urls.reverse()
self.urls = urls
self.concurrency = concurrency
self._q = queues.Queue()
self._fetching = set()
self._fetched = set()
def fetch(self, url, **kwargs):
fetch = getattr(httpclient.AsyncHTTPClient(), 'fetch')
return fetch(url, **kwargs)
def handle_html(self, url, html):
"""handle html page"""
print(url)
def handle_response(self, url, response):
"""inherit and rewrite this method"""
if response.code == 200:
self.handle_html(url, response.body)
elif response.code == 599: # retry
self._fetching.remove(url)
self._q.put(url)
@gen.coroutine
def get_page(self, url):
try:
response = yield self.fetch(url)
print('######fetched %s' % url)
except Exception as e:
print('Exception: %s %s' % (e, url))
raise gen.Return(e)
raise gen.Return(response)
@gen.coroutine
def _run(self):
@gen.coroutine
def fetch_url():
current_url = yield self._q.get()
try:
if current_url in self._fetching:
return
print('fetching****** %s' % current_url)
self._fetching.add(current_url)
response = yield self.get_page(current_url)
self.handle_response(current_url, response) # handle reponse
self._fetched.add(current_url)
for i in range(self.concurrency):
if self.urls:
yield self._q.put(self.urls.pop())
finally:
self._q.task_done()
@gen.coroutine
def worker():
while True:
yield fetch_url()
self._q.put(self.urls.pop()) # add first url
# Start workers, then wait for the work queue to be empty.
for _ in range(self.concurrency):
worker()
yield self._q.join(timeout=timedelta(seconds=300000))
assert self._fetching == self._fetched
def run(self):
io_loop = ioloop.IOLoop.current()
io_loop.run_sync(self._run)
class MySpider(AsySpider):
def fetch(self, url, **kwargs):
"""重写父类fetch方法可以添加cookies,headers,timeout等信息"""
cookies_str = "PHPSESSID=j1tt66a829idnms56ppb70jri4; pspt=%7B%22id%22%3A%2233153%22%2C%22pswd%22%3A%228835d2c1351d221b4ab016fbf9e8253f%22%2C%22_code%22%3A%22f779dcd011f4e2581c716d1e1b945861%22%7D; key=%E9%87%8D%E5%BA%86%E5%95%84%E6%9C%A8%E9%B8%9F%E7%BD%91%E7%BB%9C%E7%A7%91%E6%8A%80%E6%9C%89%E9%99%90%E5%85%AC%E5%8F%B8; think_language=zh-cn; SERVERID=a66d7d08fa1c8b2e37dbdc6ffff82d9e|1444973193|1444967835; CNZZDATA1254842228=1433864393-1442810831-%7C1444972138" # 从浏览器拷贝cookie字符串
headers = {
'User-Agent': 'mozilla/5.0 (compatible; baiduspider/2.0; +http://www.baidu.com/search/spider.html)',
'cookie': cookies_str
}
return super(MySpider, self).fetch( # 参数参考tornado文档
url, headers=headers, request_timeout=1
)
def handle_html(self, url, html):
print(url, html)
def main():
urls = []
for page in range(1, 100):
urls.append('http://www.baidu.com?page=%s' % page)
s = MySpider(urls)
s.run()
if __name__ == '__main__':
main()
可以继承这个类,塞一些url进去,然后重写handle_page处理得到的页面。
异步+多进程爬虫
还可以再变态点,加个进程池,使用了multiprocessing模块。效率飕飕的,
#!/usr/bin/env python
# -*- coding:utf-8 -*-
import time
from multiprocessing import Pool
from datetime import timedelta
from tornado import httpclient, gen, ioloop, queues
class AsySpider(object):
"""A simple class of asynchronous spider."""
def __init__(self, urls, concurrency):
urls.reverse()
self.urls = urls
self.concurrency = concurrency
self._q = queues.Queue()
self._fetching = set()
self._fetched = set()
def handle_page(self, url, html):
filename = url.rsplit('/', 1)[1]
with open(filename, 'w+') as f:
f.write(html)
@gen.coroutine
def get_page(self, url):
try:
response = yield httpclient.AsyncHTTPClient().fetch(url)
print('######fetched %s' % url)
except Exception as e:
print('Exception: %s %s' % (e, url))
raise gen.Return('')
raise gen.Return(response.body)
@gen.coroutine
def _run(self):
@gen.coroutine
def fetch_url():
current_url = yield self._q.get()
try:
if current_url in self._fetching:
return
print('fetching****** %s' % current_url)
self._fetching.add(current_url)
html = yield self.get_page(current_url)
self._fetched.add(current_url)
self.handle_page(current_url, html)
for i in range(self.concurrency):
if self.urls:
yield self._q.put(self.urls.pop())
finally:
self._q.task_done()
@gen.coroutine
def worker():
while True:
yield fetch_url()
self._q.put(self.urls.pop())
# Start workers, then wait for the work queue to be empty.
for _ in range(self.concurrency):
worker()
yield self._q.join(timeout=timedelta(seconds=300000))
assert self._fetching == self._fetched
def run(self):
io_loop = ioloop.IOLoop.current()
io_loop.run_sync(self._run)
def run_spider(beg, end):
urls = []
for page in range(beg, end):
urls.append('http://127.0.0.1/%s.htm' % page)
s = AsySpider(urls, 10)
s.run()
def main():
_st = time.time()
p = Pool()
all_num = 73000
num = 4 # number of cpu cores
per_num, left = divmod(all_num, num)
s = range(0, all_num, per_num)
res = []
for i in range(len(s)-1):
res.append((s[i], s[i+1]))
res.append((s[len(s)-1], all_num))
print res
for i in res:
p.apply_async(run_spider, args=(i[0], i[1],))
p.close()
p.join()
print time.time()-_st
if __name__ == '__main__':
main()
多线程爬虫
线程池实现.
#!/usr/bin/env python
# -*- coding:utf-8 -*-
import Queue
import sys
import requests
import os
import threading
import time
class Worker(threading.Thread): # 处理工作请求
def __init__(self, workQueue, resultQueue, **kwds):
threading.Thread.__init__(self, **kwds)
self.setDaemon(True)
self.workQueue = workQueue
self.resultQueue = resultQueue
def run(self):
while 1:
try:
callable, args, kwds = self.workQueue.get(False) # get task
res = callable(*args, **kwds)
self.resultQueue.put(res) # put result
except Queue.Empty:
break
class WorkManager: # 线程池管理,创建
def __init__(self, num_of_workers=10):
self.workQueue = Queue.Queue() # 请求队列
self.resultQueue = Queue.Queue() # 输出结果的队列
self.workers = []
self._recruitThreads(num_of_workers)
def _recruitThreads(self, num_of_workers):
for i in range(num_of_workers):
worker = Worker(self.workQueue, self.resultQueue) # 创建工作线程
self.workers.append(worker) # 加入到线程队列
def start(self):
for w in self.workers:
w.start()
def wait_for_complete(self):
while len(self.workers):
worker = self.workers.pop() # 从池中取出一个线程处理请求
worker.join()
if worker.isAlive() and not self.workQueue.empty():
self.workers.append(worker) # 重新加入线程池中
print 'All jobs were complete.'
def add_job(self, callable, *args, **kwds):
self.workQueue.put((callable, args, kwds)) # 向工作队列中加入请求
def get_result(self, *args, **kwds):
return self.resultQueue.get(*args, **kwds)
def download_file(url):
#print 'beg download', url
requests.get(url).text
def main():
try:
num_of_threads = int(sys.argv[1])
except:
num_of_threads = 10
_st = time.time()
wm = WorkManager(num_of_threads)
print num_of_threads
urls = ['http://www.baidu.com'] * 1000
for i in urls:
wm.add_job(download_file, i)
wm.start()
wm.wait_for_complete()
print time.time() - _st
if __name__ == '__main__':
main()
这三种随便一种都有很高的效率,但是这么跑会给网站服务器不小的压力,尤其是小站点,还是有点节操为好。
数据分析咨询请扫描二维码
若不方便扫码,搜微信号:CDAshujufenxi
“最近复购率一直在下降,我们的营销力度不小啊,为什么用户还是走了?” “是不是广告投放的用户质量不高?还是我们的产品问题 ...
2025-02-21以下文章来源于数有道 ,作者数据星爷 SQL查询是数据分析工作的基础,也是CDA数据分析师一级的核心考点,人工智能时代,AI能为 ...
2025-02-19在当今这个数据驱动的时代,几乎每一个业务决策都离不开对数据的深入分析。而其中,指标波动归因分析更是至关重要的一环。无论是 ...
2025-02-18当数据开始说谎:那些年我们交过的学费 你有没有经历过这样的场景?熬了三个通宵做的数据分析报告,在会议上被老板一句"这数据靠 ...
2025-02-17数据分析作为一门跨学科领域,融合了统计学、编程、业务理解和可视化技术。无论是初学者还是有一定经验的从业者,系统化的学习路 ...
2025-02-17挖掘用户价值本质是让企业从‘赚今天的钱’升级为‘赚未来的钱’,同时让用户从‘被推销’变为‘被满足’。询问deepseek关于挖 ...
2025-02-17近来deepseek爆火,看看deepseek能否帮我们快速实现数据看板实时更新。 可以看出这对不知道怎么动手的小白来说是相当友好的, ...
2025-02-14一秒精通 Deepseek,不用找教程,不用买资料,更不用报一堆垃圾课程,所有这么去做的,都是舍近求远,因为你忽略了 deepseek 的 ...
2025-02-12自学 Python 的关键在于高效规划 + 实践驱动。以下是一份适合零基础快速入门的自学路径,结合资源推荐和实用技巧: 一、快速入 ...
2025-02-12“我们的利润率上升了,但销售额却没变,这是为什么?” “某个业务的市场份额在下滑,到底是什么原因?” “公司整体业绩 ...
2025-02-08活动介绍 为了助力大家在数据分析领域不断精进技能,我们特别举办本期打卡活动。在这里,你可以充分利用碎片化时间在线学习,让 ...
2025-02-071、闺女,醒醒,媒人把相亲的带来了。 我。。。。。。。 2、前年春节相亲相了40个, 去年春节相亲50个, 祖宗,今年你想相多少个 ...
2025-02-06在数据科学的广阔领域中,统计分析与数据挖掘占据了重要位置。尽管它们常常被视为有关联的领域,但两者在理论基础、目标、方法及 ...
2025-02-05在数据分析的世界里,“对比”是一种简单且有效的方法。这就像两个女孩子穿同一款式的衣服,效果不一样。 很多人都听过“货比三 ...
2025-02-05当我们只有非常少量的已标记数据,同时有大量未标记数据点时,可以使用半监督学习算法来处理。在sklearn中,基于图算法的半监督 ...
2025-02-05考虑一种棘手的情况:训练数据中大部分样本没有标签。此时,我们可以考虑使用半监督学习方法来处理。半监督学习能够利用这些额 ...
2025-02-04一、数学函数 1、取整 =INT(数字) 2、求余数 =MOD(除数,被除数) 3、四舍五入 =ROUND(数字,保留小数位数) 4、取绝对值 =AB ...
2025-02-03作者:CDA持证人 余治国 一般各平台出薪资报告,都会哀嚎遍野。举个例子,去年某招聘平台发布《中国女性职场现状调查报告》, ...
2025-02-02真正的数据分析大神是什么样的呢?有人认为他们能轻松驾驭各种分析工具,能够从海量数据中找到潜在关联,或者一眼识别报告中的数 ...
2025-02-01现今社会,“转行”似乎成无数职场人无法回避的话题。但行业就像座围城:外行人看光鲜,内行人看心酸。数据分析这个行业,近几年 ...
2025-01-31