昨天接的一个单子,要爬上万个URL,说要快,越快越好。
于是想起来异步协程爬虫,一直想着学但是没学,正好借此学一下
是爬取的lianjia的租房房屋信息,没有反爬
除去写文件的耗时,把本来20多分钟的爬取过程降到了40秒
这里主要记录异步协程的编写思路
代码贴在最后
这个接口要包含异常处理等功能,以使程序在收到意料之外的响应时,不至于过于脆弱
一个简单的GET接口实现如下:
# -*- coding: utf-8 -*-
#几个包或变量详见下面的总代码
async def get(url):
async with semaphore:
try:
async with session.get(url) as response:
return await response.text()
except aiohttp.ClientError:
logging.error('Error occurred while getting %s',url,exc_info=True)
return None
可以先简单实现一个main函数,要用async
修饰(所有涉及到网络IO的函数都需要这样),比如获得一个最初的请求,然后通过分析函数得到一批URL
然后就是实现具体功能函数,可以先用for循环来写,比如:
for url in urls: get_index(url)
注意所有涉及网络IO的函数调用都要带上await
比如2中的代码就可以改成
indextasks = [asyncio.ensure_future(get_index(url)) for url in urls]
result = await asyncio.gather(*indextasks)
注意这里的返回值不能是两个或多个,只能是一个,可以在函数中打包好再返回
返回的result是每个函数调用的结果组成的list
读者师傅可以直接套用这个模板来写
# -*- coding: utf-8 -*-
'''
导入各种包,其中:
re.findall用来进行正则匹配
csv用来写csv文件
asyncio和aiohttp用来配合进行异步协程爬虫开发
time用来记录运行时间
logging用来显示错误日志
'''
from re import findall
import csv
import asyncio
import aiohttp
import time
import logging
import operator
#定义基本url和行政区域列表
baseurl = "https://cd.lianjia.com/zufang"
block_list = ["锦江", "青羊", "武侯", "高新", "成华", "金牛", "天府新区", "高新西"]
#定义session和最大允许开启的协程数量,这个数越大,爬的越快
session = None
semaphore = asyncio.Semaphore(8)
'''
这个函数定义了一个基本的用来实现一个使用get方法获取目标网页html文本的接口,相当于requests.get
input: A URL
output: This URL's HTML
'''
async def get(url):
async with semaphore:
try:
logging.info('Getting %s',url)
async with session.get(url) as response:
return await response.text()
except aiohttp.ClientError:
logging.error('Error occurred while getting %s',url,exc_info=True)
#这个函数用来获取每个行政分区对应的URL
def get_blockurls(html):
result = []
for block in block_list:
block_url = findall(r'href="/zufang(.*?)" >'+block, html)[0]
result.append(block_url)
return result
#这个函数用来获取子区域的区域名
def get_subblock(html):
result = []
html = html.replace("\n","").replace("\r","").replace("\t","").replace(" ","")
temp = findall(r'--level3">' , html)
for t in temp:
result.append(t.split('">')[1])
return result
#这个函数用来获取每个区域的房间数量
def get_roomnum(html):
result = 0
result = findall(r'content__title--hl">(.*?)', html)[0]
return result
#这个函数获得各个房间的URL
async def get_roomurls(html, num):
result = []
pagenum = int((num - (num%30))/30) + 1
html = html.replace("\n","").replace("\r","").replace("\t","").replace(" ","")
urls = findall(r'class="content__list--item--aside"target="_blank"href="/zufang(.*?)"title="', html)
for u in urls:
result.append(baseurl+u)
for p in range(2,pagenum+1):
html = await get(baseurl+"/pg"+str(p)+"/#contentList")
if not html: continue
#这里的判断就是判断get的返回值是否为None,如果是就是请求出错,需要跳过斜面的操作以免出现其他问题
html = html.replace("\n", "").replace("\r","").replace("\t","").replace(" ","")
urls = findall(r'class="content__list--item--aside"target="_blank"href="/zufang(.*?)"title="', html)
for u in urls:
result.append(baseurl+u)
return result
#这个函数通过正则读出HTML中的信息,并写入文件
async def get_roommessage(html, bname, w2):
'''
无关紧要的正则提取和写文件操作
'''
async def get_rooms(html, num, bname, w2):
#根据数量获取指定数量的房间urls
if num < 1000:
room_urls = await get_roomurls(html, num)
else:
room_urls = await get_roomurls(html, 1000)
if not room_urls: return
for u in room_urls:
room_r = await get(u) #爬取每个房URL的HTML
if not room_r:
continue
try:
room_message = await get_roommessage(room_r, bname, w2) #筛选并写入每个房间的信息
except:
pass
async def geturls(block, bname):
blockurl = baseurl + block
block_r = await get(blockurl)
sub_blocks = get_subblock(block_r)
return sub_blocks
async def get_message_main(block, bname, w1, w2):
print("运行了main一次")
blockurl = baseurl + block #拼接成区域的完整URL
block_r = await get(blockurl) #获取这个URL的HTML
room_num = get_roomnum(block_r) #获取每个区域的房间数量
w1.writerow([bname , room_num]) #写入文件
result = await get_rooms(block_r, int(room_num), bname, w2) #爬取每个区域的房间
async def main():
global session #将session扩展为全局变量
session = aiohttp.ClientSession() #初始化获得一个session
#创建文件,写入表头
f1 = open('file1.csv','w',encoding='utf-8')
f2 = open('file2.csv','w',encoding='utf-8')
w1 = csv.writer(f1)
w2 = csv.writer(f2)
w1.writerow(['行政区域','挂网租房数量'])
#获取每个区域的url
base_r = await get(baseurl)
block_urls = get_blockurls(base_r)
#创建并运行协程
indextasks = [asyncio.ensure_future(get_message_main(block,bname,w1,w2)) for block,bname in zip(block_urls,block_list)]
result = await asyncio.gather(*indextasks)
#关闭文件和session
f1.close()
f2.close()
await session.close()
def paicsv():
data=[]
reader = csv.reader(open("file2.csv","r"))
for row in reader:
data.append(row)
data.sort()
with open("file2.csv","w") as f:
writer = csv.writer(f)
writer.writerow(['行政区域','区域','小区','房型','房源维护时间','面积','朝向','维护','入住','楼层','电梯','车位','用水','用电','燃气','采暖','租期','看房','付款方式','租金','押金','洗衣机','空调','衣柜','电视','冰箱','热水器','床','暖气','宽带','天然气'])
for row in data:
writer.writerow(row)
if __name__ == '__main__':
start = time.time()
print("Start at: " , start)
#运行main函数
asyncio.get_event_loop().run_until_complete(main())
end = time.time()
print("End at: " , end)
print("Time cost:" , end-start)
paicsv()
以上,共勉
有问题欢迎在评论区提问讨论,一起学习