Python爬虫—urllib

urllib语法

urllib.request模块

• Request(url,data)：用作url请求传参，返回的Request对象可直接传入urlopen
• urlretrieve(url, path)：直接下载url网页到本地
• urlcleanup()：清除缓存信息
• urlopen(url[, timeout])：访问url，如果设置timeout超时将抛出<urlopen error timed out>异常。返回Response对象用法如下
  • response.getcode()：响应的状态码
  • response.geturl()：当前访问的url
  • response.info()：url网页的信息
  • response.read().decode('utf-8')：网页源代码
• build_opener([proxy,request.HTTPHandler])：构建opener对象，该对象用法如下:
  • add_headers = [(),(),()...]：列表可传入请求头元组，如headers、cookies
  • open(url)：携带头参数访问url，返回Response对象

  url = 'https://www.baidu.com'
  # 必须传入元组
  headers = ('User-Agent', 'Mozilla/5.0 (Windows NT 10.0; WOW64) ...')
  cookies = ('B2352_3239RTM=207 ...')
  opener = build_opener()
  opener.add_headers = [headers, cookies]
  response = opener.open(url)
  

• install_opener(opener)：将opener对象设置为全局的，一旦设置为全局，opener.open(url) -> request.urlopen(url)

  url = 'https://www.baidu.com'
  headers = ('User-Agent', 'Mozilla/5.0 (Windows NT 10.0; WOW64) ...')
  cookies = ('B2352_3239RTM=207 ...')
  opener = build_opener()
  opener.add_headers = [headers, cookies]
  # 添加为全局opener
  request.install_opener(opener)
  # opener.open -> request.urlopen
  response = request.urlopen(url)
  

• ProxyHandler({"scheme":"ip:port”}): 与代理ip相关。返回proxy对象可传入build_opener()

urllib.parse模块

• quote(text)：对字符串url编码，可以将中文转化为url字符串

  params = '?name=玛尔扎哈&age=20&sex=girl'
  print(quote(params))
  params = '沙扬娜拉'
  print(quote(params))
  

• urlencode(form_data)：传入字典表单生成查询字符串，会对汉字、特殊字符编码

  formdata = {
      'name': '阿兹尔',  # 中文自动quote编码
      'username': 'demanwei',
      'password': '123abc',
      'extraA': '',
      'extraB': None,
  }
  print(urlencode(formdata))
  

• parse_qs(urlencoded)：urlencode的逆操作，将查询字符串解码成字典，会自动对转码后的中文解码

  urlencoded = 'name=%E9%98%BF%E5%85%B9%E5%B0%94&username=demanwei&password=123abc&extraA=&extraB=None'
  # 字典的v是一个列表
  print(parse_qs(urlencoded))
  # 通常参数k对应的v只有一个,可通过下面的方式得到字典对象
  print({k: v[0] for k, v in parse_qs(urlencoded).items()})
  

• urlparse(url)：获取url的结构信息(scheme,netloc,path,params,query,fragment)

  url = 'http://localhost:8080/user/get?id=12'
  print(urlparse(url))
  

• urlsplit(url)：和urlparse基本一样，但不含params，即(scheme,netloc,path,query,fragment)

  url = 'http://localhost:8080/user/get?id=12'
  print(urlsplit(url))
  

注意

尽量使用urlencode而非quote！

him = '玛尔扎哈'
her = '沙扬娜拉'
raw_url = 'http://localhost:8080?him={}&her={}'.format(him, her)
print(raw_url)

# 返回每个解码后的text,不要直接操作整个url!!
new_url = 'http://localhost:8080?him={}&her={}'.format(quote(him), quote(her))
print(new_url)

# 直接返回查询字符串,以后用这个
formdata = {'him': him, 'her': her}
new_url = 'http://localhost:8080?{}'.format(urlencode(formdata))
print(new_url)

GET和POST请求

GET

kw = request.quote('马云')
url = 'https://www.baidu.com/s?wd={}'.format(kw)
response = request.urlopen(url)

POST

url = 'https://www.iqianyue.com/mypost'
# POST表单参数
body = {'name':'林在超','pass':'111111'}
data = parse.urlencode(body).encode('utf-8')

req = request.Requset(url, data=data)		#Request对象
response = request.urlopen(req)
text = response.read().decode('utf-8')
print(text)

请求头

url = 'https://www.baidu.com'
headers = ('User-Agent','Mozilla/5.0 (Windows NT 10.0; WOW64)....')
cookies = ('........')

opener = request.build_opener()
opener.add_headers = [headers, cookies] # 必须传入元组
# opener设置为全局的
request.install_opener(opener)

response = opener.open(url)
print(response.getcode())

Cookie的加载与保存

from http.cookiejar import MozillaCookieJar
from urllib import request

cookiejar = MozillaCookieJar('resource/cookie.txt')
cookiejar.load(ignore_discard=True)

handler = request.HTTPCookieProcessor(cookiejar)
opener = request.build_opener(handler)
opener.open('https://httpbin.org/cookies/set?oourse=abc')

cookiejar.save(ignore_discard=True)

异常处理

• URLError：爬虫在运行的过程中，很多时候会遇到这样或那样的异常。如果没有异常处理，爬虫遇到异常时就会直接崩溃停止运行，下次再运行时又从头开始。所以，要开发一个具有顽强生命力的爬虫，必须要进行异常处理。出现原因如下:
  • 连不上服务器
  • 远程url不存在
  • 无网络
  • 触发HTMLError
• 与HTTPError的关系：两者都是异常处理的类，HTTPError是URLError的子类，前者有异常状态码与异常原因，URLError只有异常原因没有异常状态码。所以在处理的时候，不能使用URLError直接代替HTTPError，如果要替代必须判断是否有状态码属性。

try:
	request.urlopen('http:/www.ncut.edu.com')
except urllib.error.URLError as e:
	print(e)			# <urlopen error no host given>
    print(e.reason)		# no host given
    print(repr(e))		# URLError('no host given')
except Exception as e:
    print(repr(e))

代理ip的构建

• 代理的原理：在请求网站之前，先请求代理服务器，然后让代理服务器去请求目的网站，代理服务器拿到目的网站的数据之后，再转发给我们的代码

• 步骤:

  1. 使用request.ProxyHandler({"scheme":"ip:port”}),带入代理构建一个handler
  2. 使用上面的handler构建一个opener
  3. 使用opener发送一个请求

• 法1: 适合于代理ip稳定的情况，类似于用户代理ip的构建

  ip = '175.6.6.101:5000'
  
  proxy = request.ProxyHandler({'http': ip})
  opener = request.build_opener(proxy, request.HTTPHandler)
  request.install_opener(opener)
  
  response = request.urlopen('http://www.win4000.com/meitu.html')
  

• 法2:接口调用法，适合ip不稳定

  handler = request.ProxyHandler({"http":"233.241.78.43:8010"})
  opener = request.build_opener(handler)
  response = opener.open('http://www.win4000.com/meitu.html')
  

代理用户池的构建

IP池

def spider_ip():
    """ 此函数爬取IP代理网站(如快代理...),返回IP_pool """
    ip_pool = []
    pass	# 爬取过程略
	# [{'协议类型': 'ip:端口'}, {'协议类型': 'ip:端口'}, {'协议类型': 'ip:端口'}...]
	return ip_pool

ip_pool = spider_ip()
proxy = request.ProxyHandler({'http':random.choice(ip_pool)})
opener = request.build_opener(proxy, request.HTTPHandler)
request.install_opener(opener)
response = request.urlopen('https://www.baidu.com')

UA池

# 一堆user-agent
UA_pool = [
    'Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0)',
    'Opera/9.80 (Windows NT 6.1; U; zh-cn) Presto/2.9.168 Version/11.50',
    'Mozilla/5.0 (Windows NT 6.1; WOW64; rv:6.0) Gecko/20100101 Firefox/6.0'
]

headers = ('user-agent', random.choice(UA_pool))
opener = urllib.request.build_opener()
opener.add_headers = [headers]
request.install_opener(opener)
response = request.urlopen('https://www.baidu.com')