class SinaClient(object):
def __init__(self, username=None, password=None):
self.username = username
self.password = password
self.servertime = None
self.nonce = None
self.pubkey = None
self.rsakv = None
self.post_data = None
self.headers = {}
self.session = None
self.cookiejar = None
self.logger = None
self.state = False
self.initParams()
def initParams(self):
self.logger = LogClient().createLogger(
'SinaClient',
'out/log_' + time.strftime("%Y%m%d", time.localtime()) + '.log')
self.headers = dataEncode.headers
return self
def setAccount(self, username, password):
self.username = username
self.password = password
return self
def setPostData(self):
self.servertime, self.nonce, self.pubkey, self.rsakv = dataEncode.get_prelogin_info(
)
self.post_data = dataEncode.encode_post_data(self.username,
self.password,
self.servertime,
self.nonce, self.pubkey,
self.rsakv)
return self
def login(self, username=None, password=None):
self.setAccount(username, password)
self.setPostData()
login_url = r'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.15)'
session = requests.Session()
response = session.post(login_url, data=self.post_data)
json_text = response.content.decode('gbk')
res_info = json.loads(json_text)
try:
if res_info["retcode"] == "0":
self.logger.info("Login success!")
self.state = True
cookies = session.cookies.get_dict()
cookies = [key + "=" + value for key, value in cookies.items()]
cookies = "; ".join(cookies)
session.headers["Cookie"] = cookies
else:
self.logger.error("Login Failed! | " + res_info["reason"])
except Exception, e:
self.logger.error("Loading error --> " + e)
self.session = session
return session
class SinaClient(object):
def __init__(self, username=None, password=None):
self.username = username
self.password = password
self.servertime = None
self.nonce = None
self.pubkey = None
self.rsakv = None
self.post_data = None
self.headers = {}
self.session = None
self.cookiejar = None
self.logger = None
self.status = False
self.access_token = None
self.app_key = None
self.initParams()
self.timeout = 3
socket.setdefaulttimeout(3)
self.tryTimes = 8
def initParams(self):
self.logger = LogClient().createLogger('SinaWSpider',
myconf.log_out_path)
self.headers = myconf.headers
self.access_token = myconf.access_token
self.app_key = myconf.app_key
return self
def setAccount(self, username, password):
self.username = username
self.password = password
return self
def setPostData(self):
self.servertime, self.nonce, self.pubkey, self.rsakv = dataEncode.get_prelogin_info(
)
self.post_data = dataEncode.encode_post_data(self.username,
self.password,
self.servertime,
self.nonce, self.pubkey,
self.rsakv)
return self
def switchUserAgent(self, enableAgent=True):
user_agent = random.choice(myconf.agent_list)
self.headers["User-Agent"] = user_agent
return self
def switchUserAccount(self, userlist):
is_login = False
while not is_login:
self.switchUserAgent()
self.logger.info("User-Agent is: " + self.headers["User-Agent"])
user = random.choice(userlist).split("|")
self.logger.info("logining with user: "******"Cannot login to sina!")
continue
is_login = True
return self
def enableCookie(self, enableProxy=False):
self.cookiejar = cookielib.LWPCookieJar() # 建立COOKIE
cookie_support = urllib2.HTTPCookieProcessor(self.cookiejar)
opener = urllib2.build_opener(cookie_support, urllib2.HTTPHandler)
urllib2.install_opener(opener)
return self
def login(self, username=None, password=None):
self.logger.info("Start to login...")
self.setAccount(username, password)
self.setPostData()
self.enableCookie(enableProxy=True)
login_url = r'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.15)'
headers = self.headers
try:
request = urllib2.Request(login_url,
urllib.urlencode(self.post_data),
headers)
resText = urllib2.urlopen(request).read()
jsonText = json.loads(resText)
if jsonText["retcode"] == "0":
self.logger.info("Login success!")
self.status = True
cookies = ';'.join([
cookie.name + "=" + cookie.value
for cookie in self.cookiejar
])
headers["Cookie"] = cookies
else:
self.logger.error("Login Failed --> " + jsonText["reason"])
except Exception, e:
self.logger.error("Login Failed2! --> " + str(e))
self.headers = headers
return self
class SinaClient(object):
def __init__(self, username=None, password=None):
#用户输入的用户名与密码
self.username = username
self.password = password
#从prelogin.php中获取的数据
self.servertime = None
self.nonce = None
self.pubkey = None
self.rsakv = None
#请求时提交的数据列表
self.post_data = None
self.headers = {}
#用于存储登录后的session
self.session = None
self.cookiejar = None
#用于输出log信息
self.logger = None
#登录状态，初始化为False，表示未登录状态
self.status = False
#微博API必备信息
self.access_token = None
self.app_key = None
#初始时调用initParams方法，初始化相关参数
self.initParams()
self.timeout = 3
socket.setdefaulttimeout(3)
self.tryTimes = 8
#初始化参数
def initParams(self):
self.logger = LogClient().createLogger('SinaWSpider',
myconf.log_out_path)
self.headers = myconf.headers
self.access_token = myconf.access_token
self.app_key = myconf.app_key
return self
#设置username 和 password
def setAccount(self, username, password):
self.username = username
self.password = password
return self
#设置post_data
def setPostData(self):
self.servertime, self.nonce, self.pubkey, self.rsakv = dataEncode.get_prelogin_info(
)
self.post_data = dataEncode.encode_post_data(self.username,
self.password,
self.servertime,
self.nonce, self.pubkey,
self.rsakv)
return self
#使用用户代理，更换header中的User-Agent
def switchUserAgent(self, enableAgent=True):
user_agent = random.choice(myconf.agent_list)
self.headers["User-Agent"] = user_agent
return self
#用于切换用户账号，防止长时间爬取账号被禁
def switchUserAccount(self, userlist):
is_login = False
while not is_login:
self.switchUserAgent()
self.logger.info("User-Agent is: " + self.headers["User-Agent"])
user = random.choice(userlist).split("|")
self.logger.info("logining with user: "******"Cannot login to sina!")
continue
is_login = True
return self
#生成Cookie,接下来的所有get和post请求都带上已经获取的cookie
def enableCookie(self, enableProxy=False):
self.cookiejar = cookielib.LWPCookieJar() # 建立COOKIE
cookie_support = urllib2.HTTPCookieProcessor(self.cookiejar)
if enableProxy:
proxy = myconf.swithProxy()
proxy_support = urllib2.ProxyHandler(proxy) # 使用代理myconf.proxies
opener = urllib2.build_opener(proxy_support, cookie_support,
urllib2.HTTPHandler)
self.logger.info("Proxy enable, proxy is: " + str(proxy))
else:
opener = urllib2.build_opener(cookie_support, urllib2.HTTPHandler)
urllib2.install_opener(opener)
return self
#使用urllib2模拟登录过程
def login(self, username=None, password=None):
self.status = False #重新将登录状态设置为False
self.logger.info("Start to login...")
#根据用户名和密码给默认参数赋值,并初始化post_data
self.setAccount(username, password)
self.setPostData()
self.enableCookie(enableProxy=True)
#登录时请求的url
login_url = r'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.15)'
headers = self.headers
try:
request = urllib2.Request(login_url,
urllib.urlencode(self.post_data),
headers)
resText = urllib2.urlopen(request).read()
jsonText = json.loads(resText)
if jsonText["retcode"] == "0":
self.logger.info("Login success!")
self.status = True
#将cookie加入到headers中
cookies = ';'.join([
cookie.name + "=" + cookie.value
for cookie in self.cookiejar
])
headers["Cookie"] = cookies
else:
self.logger.error("Login Failed --> " + jsonText["reason"])
except Exception, e:
self.logger.error("Login Failed2! --> " + str(e))
self.headers = headers
return self
class SinaClient(object):
def __init__(self, username=None, password=None):
# 用户输入的用户名与密码
self.username = username
self.password = password
# 从prelogin.php中获取的数据
self.servertime = None
self.nonce = None
self.pubkey = None
self.rsakv = None
# 请求时提交的数据列表
self.post_data = None
self.headers = {}
# 用于存储登录后的session
self.session = None
self.cookiejar = None
# 用于输出log信息
self.logger = None
# 存储登录状态，初始状态为False
self.state = False
# 初始时调用initParams方法，初始化相关参数
self.initParams()
# 初始化参数
def initParams(self):
self.logger = LogClient().createLogger(
'SinaClient',
'out/log_' + time.strftime("%Y%m%d", time.localtime()) + '.log')
self.headers = dataEncode.headers
return self
# 设置username 和 password
def setAccount(self, username, password):
self.username = username
self.password = password
return self
# 设置post_data
def setPostData(self):
self.servertime, self.nonce, self.pubkey, self.rsakv = dataEncode.get_prelogin_info(
)
self.post_data = dataEncode.encode_post_data(self.username,
self.password,
self.servertime,
self.nonce, self.pubkey,
self.rsakv)
return self
# 使用requests库登录到 https://login.sina.com.cn
def login(self, username=None, password=None):
# 根据用户名和密码给默认参数赋值,并初始化post_data
self.setAccount(username, password)
self.setPostData()
# 登录时请求的url
login_url = r'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.15)'
session = requests.Session()
response = session.post(login_url, data=self.post_data)
json_text = response.content.decode('gbk')
res_info = json.loads(json_text)
try:
if res_info["retcode"] == "0":
self.logger.info("Login success!")
self.state = True
# 把cookies添加到headers中
cookies = session.cookies.get_dict()
cookies = [key + "=" + value for key, value in cookies.items()]
cookies = "; ".join(cookies)
session.headers["Cookie"] = cookies
else:
self.logger.error("Login Failed! | " + res_info["reason"])
except Exception, e:
self.logger.error("Loading error --> " + e)
self.session = session
return session
class SinaClient(object):
def __init__(self, username=None, password=None):
#用户输入的用户名与密码
self.username = username
self.password = password
#从prelogin.php中获取的数据
self.servertime = None
self.nonce = None
self.pubkey = None
self.rsakv = None
#请求时提交的数据列表
self.post_data = None
self.headers = {}
#用于存储登录后的session
self.session = None
self.cookiejar = None
#用于输出log信息
self.logger = None
#存储登录状态，初始状态为False
self.state = False
#初始时调用initParams方法，初始化相关参数
self.initParams()
#初始化参数
def initParams(self): #// 如：out/log_20190420.log
self.logger = LogClient().createLogger('SinaClient', 'out/log_' + time.strftime("%Y%m%d", time.localtime()) + '.log')
self.headers = dataEncode.headers
return self
#设置username 和 password
def setAccount(self, username, password):
self.username = username
self.password = password
return self
#设置post_data
def setPostData(self):
self.servertime, self.nonce, self.pubkey, self.rsakv = dataEncode.get_prelogin_info()
self.post_data = dataEncode.encode_post_data(self.username, self.password, self.servertime, self.nonce, self.pubkey, self.rsakv)
return self
#使用requests库登录到 https://login.sina.com.cn
def login(self, username=None, password=None):
#根据用户名和密码给默认参数赋值,并初始化post_data
self.setAccount(username, password)
self.setPostData()
#登录时请求的url //network中headers可以看到此url，这里把url后面的时间戳省略了
login_url = r'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.15)'
session = requests.Session()
response = session.post(login_url, data=self.post_data)
json_text = response.content.decode('gbk')
res_info = json.loads(json_text)
try:
if res_info["retcode"] == "0":
self.logger.info("Login success!")
self.state = True
#把cookies添加到headers中
cookies = session.cookies.get_dict()
cookies = [key + "=" + value for key, value in cookies.items()]
cookies = "; ".join(cookies)
session.headers["Cookie"] = cookies
else:
self.logger.error("Login Failed! | " + res_info["reason"])
#except Exception, e: #//本版本不能这样使用
except Exception as e:
self.logger.error("Loading error --> " + e)
self.session = session
return session
#生成Cookie,接下来的所有get和post请求都带上已经获取的cookie
def enableCookie(self, enableProxy=False):
#self.cookiejar = cookielib.LWPCookieJar() # 建立COOKIE //python 3.6 改成 cookiejar了
self.cookiejar = cookiejar.LWPCookieJar() # 建立COOKIE
#cookie_support = urllib2.HTTPCookieProcessor(self.cookiejar) //urllib2 在 python3.x 中被改为urllib.request
#//利用urllib库中的request的HTTPCookieProcessor对象来创建cookie处理器
cookie_support = urllib.request.HTTPCookieProcessor(self.cookiejar)
if enableProxy: #//这里enableProxy=False，应该不执行if语句
#proxy_support = urllib2.ProxyHandler({'http': 'http://122.96.59.107:843'}) # 使用代理
proxy_support = urllib.request.ProxyHandler({'http': 'http://122.96.59.107:843'}) # 使用代理
#opener = urllib2.build_opener(proxy_support, cookie_support, urllib2.HTTPHandler)
opener = urllib.request.build_opener(proxy_support, cookie_support, urllib.request.HTTPHandler)
self.logger.info("Proxy enable.")
else:
#opener = urllib2.build_opener(cookie_support, urllib2.HTTPHandler)
opener = urllib.request.build_opener(cookie_support, urllib.request.HTTPHandler) #//通过handler来构建opener
#urllib2.install_opener(opener)
urllib.request.install_opener(opener)
#//如果程序里所有的请求都使用自定义的opener，可以使用urllib.request.install_opener(),
#//将自定义的 opener 对象 定义为 全局opener，表示如果之后凡是调用urlopen，都将使用这个opener.
#使用urllib2模拟登录过程
def login2(self, username=None, password=None):
self.logger.info("Start to login...")
#根据用户名和密码给默认参数赋值,并初始化post_data
self.setAccount(username, password)
self.setPostData()
self.enableCookie()
#登录时请求的url
login_url = r'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.15)'
headers = self.headers
#request = urllib2.Request(login_url, urllib.urlencode(self.post_data), headers) //POST data should be bytes
#resText = urllib2.urlopen(request).read()
request = urllib.request.Request(login_url, urllib.parse.urlencode(self.post_data).encode("utf-8"), headers)
resText = urllib.request.urlopen(request).read()
try:
jsonText = json.loads(resText)
if jsonText["retcode"] == "0":
self.logger.info("Login success!")
self.state = True
#将cookie加入到headers中
cookies = ';'.join([cookie.name + "=" + cookie.value for cookie in self.cookiejar])
headers["Cookie"] = cookies #//dict headers在dataEncode.py中，这里给headers加上Cookie
else:
self.logger.error("Login Failed --> " + jsonText["reason"])
#//except Exception, e:
except Exception as e:
#//print e
print(e)
self.headers = headers
return self
#打开url时携带headers,此header需携带cookies
def openURL(self, url, data=None):
#req = urllib2.Request(url, data=data, headers=self.headers)
#text = urllib2.urlopen(req).read()
req = urllib.request.Request(url, data=data, headers=self.headers)
text = urllib.request.urlopen(req).read()
return text
#功能：将文本内容输出至本地
def output(self, content, out_path, save_mode="w"):
self.logger.info("Download html page to local machine. | path: " + out_path)
prefix = os.path.dirname(out_path)
if not os.path.exists(prefix):
os.makedirs(prefix)
fw = open(out_path, save_mode)
#fw.write(content)
fw.write(str(content))
fw.close()
return self
"""
防止读取出来的HTML乱码，测试样例如下
req = urllib2.Request(url, headers=headers)
text = urllib2.urlopen(req).read()
unzip(text)
"""
def unzip(self, data):
data = StringIO.StringIO(data)
gz = gzip.GzipFile(fileobj=data)
data = gz.read()
gz.close()
return data