Python多线程、异步+多进程爬虫实现代码


Posted in Python onFebruary 17, 2016

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。

Python 相关文章推荐
Win7上搭建Cocos2d-x 3.1.1开发环境
Jul 03 Python
Python实现高效求解素数代码实例
Jun 30 Python
win7 x64系统中安装Scrapy的方法
Nov 18 Python
python 多个参数不为空校验方法
Feb 14 Python
pytorch 固定部分参数训练的方法
Aug 17 Python
Python多线程操作之互斥锁、递归锁、信号量、事件实例详解
Mar 24 Python
python 日志 logging模块详细解析
Mar 31 Python
如何将tensorflow训练好的模型移植到Android (MNIST手写数字识别)
Apr 22 Python
Django 5种类型Session使用方法解析
Apr 29 Python
使用keras内置的模型进行图片预测实例
Jun 17 Python
Python并发爬虫常用实现方法解析
Nov 19 Python
Python利用capstone实现反汇编
Apr 06 Python
玩转python爬虫之爬取糗事百科段子
Feb 17 #Python
玩转python爬虫之正则表达式
Feb 17 #Python
玩转python爬虫之URLError异常处理
Feb 17 #Python
玩转python爬虫之cookie使用方法
Feb 17 #Python
Python 爬虫爬取指定博客的所有文章
Feb 17 #Python
Using Django with GAE Python 后台抓取多个网站的页面全文
Feb 17 #Python
python实现RSA加密(解密)算法
Feb 17 #Python
You might like
深入php 正则表达式的学习探讨
2013/06/06 PHP
PHP与javascript实现变量交互的示例代码
2013/07/23 PHP
php去掉URL网址中带有PHPSESSID的配置方法
2014/07/08 PHP
教大家制作简单的php日历
2015/11/17 PHP
简单谈谈PHP面向对象之标识对象
2017/06/27 PHP
PHP+ajax实现获取新闻数据简单示例
2018/05/08 PHP
Laravel统计一段时间间隔的数据方法
2019/10/09 PHP
Javascript Math ceil()、floor()、round()三个函数的区别
2010/03/09 Javascript
模拟jQuery中的ready方法及实现按需加载css,js实例代码
2013/09/27 Javascript
关于jquery的多个选择器的使用示例
2013/10/18 Javascript
JavaScript实现Java中Map容器的方法
2016/10/09 Javascript
js canvas实现适用于移动端的百分比仪表盘dashboard
2017/07/18 Javascript
ndm:NPM的桌面GUI应用程序
2018/10/15 Javascript
微信小程序实现跳转的几种方式总结(推荐)
2019/04/24 Javascript
如何阻止移动端浏览器点击图片浏览
2020/08/29 Javascript
梳理一下vue中的生命周期
2020/12/30 Vue.js
python入门前的第一课 python怎样入门
2018/03/06 Python
django admin 自定义替换change页面模板的方法
2019/08/23 Python
Numpy 中的矩阵求逆实例
2019/08/26 Python
python使用celery实现异步任务执行的例子
2019/08/28 Python
详解numpy.ndarray.reshape()函数的参数问题
2020/10/13 Python
纯css3显示隐藏一个div特效的具体实现
2014/02/10 HTML / CSS
美国知名的女性服饰品牌:LOFT(洛芙特)
2016/08/05 全球购物
国际象棋商店:The Chess Store
2018/07/09 全球购物
mysql有关权限的表都有哪几个
2015/04/22 面试题
计算机应用专业推荐信
2013/11/13 职场文书
小学生爱国演讲稿
2014/04/25 职场文书
领导班子个人对照检查剖析材料
2014/09/29 职场文书
机关干部三严三实心得体会
2014/10/13 职场文书
2015年秘书个人工作总结
2015/04/25 职场文书
2016年教师党员公开承诺书
2016/03/24 职场文书
导游词之吉林花园山
2019/10/17 职场文书
2020年元旦晚会策划书模板
2019/12/30 职场文书
ORM模型框架操作mysql数据库的方法
2021/07/25 MySQL
如何创建一个创建MySQL数据库中的datetime类型
2022/03/21 MySQL
MySQL视图概念以及相关应用
2022/04/19 MySQL