python多线程、异步、异步+多进程爬虫

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。

Quora的Python异步编程实践