def _get_captcha(self, lang: str):
"""
请求验证码的 API 接口，无论是否需要验证码都需要请求一次
如果需要验证码会返回图片的 base64 编码
根据 lang 参数匹配验证码，需要人工输入
:param lang: 返回验证码的语言(en/cn)
:return: 验证码的 POST 参数
"""
api = 'https://www.zhihu.com/api/v3/oauth/captcha?lang={}'.format(lang)
resp = self.session.get(api)
show_captcha = re.search(r'true', resp.text)
if show_captcha:
put_resp = self.session.put(api)
json_data = json.loads(put_resp.text)
img_base64 = json_data['img_base64'].replace(r'\n', '')
img_data = base64.b64decode(img_base64)
self.logger.info('使用超级鹰识别验证码...')
ok, result = image_to_text(img_data)
if ok:
self.logger.info('成功识别验证码! ')
# 这里必须先把参数 POST 验证码接口
time.sleep(1)
res = self.session.post(api, data={
'input_text': result
}).json()
if 'success' in res.keys():
return result
raise Exception('验证码 post 失败: {}'.format(
res['error']['message']))
raise Exception('验证码识别失败: ', result)
raise Exception('获取验证码失败! ')
def _get_verifycode(self):
captcha_url = 'http://icode.renren.com/getcode.do?t=web_login&rnd=0.28838133194471105'
img_data = self.session.get(captcha_url).content
self.logger.info('使用超级鹰识别验证码...')
ok, result = image_to_text(img_data)
if ok:
self.logger.info('成功识别验证码！')
return result
raise Exception('验证码识别失败: ', result)
def _get_verifycode(self, pcid):
captcha_url = f'https://login.sina.com.cn/cgi/pin.php?r=16343619&s=0&p={pcid}'
img_data = self.session.get(captcha_url).content
self.logger.info('使用超级鹰识别验证码...')
ok, result = image_to_text(img_data)
if ok:
self.logger.info('成功识别验证码！')
return result
raise Exception('验证码识别失败: ', result)
def _get_verifycode(self):
captcha_url = 'https://www.qichamao.com/usercenter/varifyimage?'
img_data = self.session.get(captcha_url).content
self.logger.info('使用超级鹰识别验证码...')
ok, result = image_to_text(img_data)
if ok:
self.logger.info('成功识别验证码！')
return result
raise Exception('验证码识别失败: ', result)
def _get_verifycode(self):
timestamp = int(time.time() * 1000)
captcha_url = f'https://api.qimai.cn/account/getVerifyCodeImage?{timestamp}'
img_data = self.session.get(captcha_url).content
self.logger.info('使用超级鹰识别验证码...')
ok, result = image_to_text(img_data)
if ok:
self.logger.info('成功识别验证码！')
return result
raise Exception('验证码识别失败: ', result)
def _get_verifycode(self):
captcha_url = 'https://login.anjuke.com/general/captcha?h=45×tamp={}'.format(
int(time.time()) * 1000)
img_data = self.session.get(captcha_url).content
self.logger.info('使用超级鹰识别验证码...')
ok, result = image_to_text(img_data)
if ok:
self.logger.info('成功识别验证码！')
return result
raise Exception('验证码识别失败: ', result)
def _get_verifycode(self, code_string):
"""
使用超级鹰识别验证码
:param pcid:
:return:
"""
captcha_url = f'https://passport.baidu.com/cgi-bin/genimage?{code_string}'
img_data = self.session.get(captcha_url).content
self.logger.info('使用超级鹰识别验证码...')
ok, result = image_to_text(img_data)
if ok:
self.logger.info('成功识别验证码！')
return result
raise Exception('验证码识别失败: ', result)
