def get_qrcode_info(request):
"""
获取二维码相关信息
:param request:
:return:
"""
ret_data = {}
try:
qrurl = "https://qrlogin.taobao.com/qrcodelogin/generateQRCode4Login.do"
res = get_response_by_requests(qrurl, headers=DEFAULT_HEADERS).json()
if res.get("success"):
url = res.get("url", "")
qrcode_url = "https:" + url if not url.startswith("http") else url
qrcode_body = get_response_by_requests(
qrcode_url, headers=DEFAULT_HEADERS).content
qrcode_info = {
"lgToken": res.get("lgToken", ""),
"adToken": res.get("adToken", ""),
"qrcode_url": qrcode_url,
"qrcode_body": bytes.decode(b64encode(qrcode_body))
}
ret_data["msg"] = "获取二维码相关信息成功"
ret_data["data"] = qrcode_info
else:
add_ajax_error_json(ret_data, "获取二维码相关信息失败")
except Exception:
add_ajax_error_json(ret_data, "获取二维码相关信息出错")
else:
add_ajax_ok_json(ret_data)
finally:
return JsonResponse(ret_data)
def http_request(self, url, method="GET", data=None, headers=None, cookies=None,
to_json=False, get_str=True, charset="utf-8", get_cookies=False):
"""
封装HTTP请求
:param url:
:param data:
:param method:
:param headers:
:param cookies:
:param to_json:
:param get_str:
:param charset:
:param get_cookies:
:return:
"""
try:
cookies_dic = {}
if headers is None:
headers = self.headers or {}
if cookies is None:
cookies = self.cookies or {}
if isinstance(cookies, list):
cookies = {cookie["name"]: cookie["value"] for cookie in cookies}
if method == "GET":
if get_cookies:
resp = get_response_by_requests(url, headers=headers, cookie_jar=cookies)
cookies_dic = resp.cookies.get_dict()
content = resp.content
else:
content = get_content_by_requests(url, headers=headers, cookie_jar=cookies)
elif method == "POST":
if get_cookies:
resp = get_response_by_requests_post(url, headers=headers, cookie_jar=cookies)
cookies_dic = resp.cookies.get_dict()
content = resp.content
else:
content = get_content_by_requests_post(url, data=data, headers=headers, cookie_jar=cookies)
else:
self.logger.error("暂不支持该请求方法")
return
if not get_str:
if get_cookies:
return {"result": content, "cookies": cookies_dic}
return content
page = content.decode(charset)
if not to_json:
if get_cookies:
return {"result": page, "cookies": cookies_dic}
return page
if get_cookies:
return {"result": json_loads(page), "cookies": cookies_dic}
return json_loads(page)
except Exception:
self.logger.exception("请求出错: url:%s" % url)
return
def __get_qr(self):
index_url = 'https://mail.qq.com/cgi-bin/loginpage'
index_r = get_response_by_requests(index_url, self.headers)
index_text = index_r.text
iframe_url = Selector(
text=index_text).xpath("//iframe/@src").extract_first("")
iframe_r = get_response_by_requests(iframe_url, self.headers)
cookies = iframe_r.cookies.get_dict()
appid = self.appid_pattern.search(iframe_url).group(1)
daid = self.daid_pattern.search(iframe_url).group(1)
qr_url = "https://ssl.ptlogin2.qq.com/ptqrshow?" \
"appid={0}&e=2&l=M&s=3&d=72&v=4&t={1}&" \
"daid={2}&pt_3rd_aid=0".format(appid, random(), daid)
qr = http_get(qr_url, self.headers, cookies=cookies)
cookies.update(qr.cookies.get_dict())
return qr.content, cookies, appid, daid
def ask_qrcode_status(request):
"""
获取扫描二维码状态
:param request:
:return:
"""
ret_data = {}
succ = False
need_refresh = False
try:
args = request.POST
username = args["username"]
account_type = args["account_type"]
lg_token = args.get("lg_token", "")
check_url_base = "https://qrlogin.taobao.com/qrcodelogin/qrcodeLoginCheck.do?" \
"lgToken={lgToken}&defaulturl=https%3A%2F%2Fwww.taobao.com%2F"
check_url = check_url_base.format(lgToken=lg_token)
res_json = get_response_by_requests(check_url,
headers=DEFAULT_HEADERS).json()
session = req_session()
msg = "通过扫描二维码登录失败"
code = res_json.get("code")
if code == "10000":
msg = "请先扫描二维码"
elif code == "10001":
msg = "扫描成功后，请确认登录"
succ = True
elif code == "10004":
msg = "二维码已失效，请重试"
need_refresh = True
elif code == "10006":
redirect_url = res_json.get("url")
resp = session.get(redirect_url,
headers=DEFAULT_HEADERS,
verify=False)
if resp.status_code == 200:
msg = "登录成功"
cookies = session.cookies.get_dict(domain='.taobao.com')
cookies_str = json_dumps(cookies)
# 将登录成功的cookies信息存入ssdb，供爬虫端使用
ssdb_connect = get_ssdb_conn()
key = username + ACCOUNT_CRAWLING_QRCODE_COOKIES_SSDB_SUFFIX + account_type
ssdb_connect.setx(key, cookies_str, DATA_EXPIRE_TIME)
succ = True
else:
msg = res_json.get("msg", "通过扫描二维码登录失败")
except Exception:
msg = "获取扫描二维码状态出错"
if succ:
add_ajax_ok_json(ret_data)
else:
ret_data["need_refresh"] = need_refresh
add_ajax_error_json(ret_data, msg)
return JsonResponse(ret_data)
def get_captcha(self, response, username="", cookie_dict=None):
"""
获取验证码并识别，返回识别的验证码, cookies
"""
# cookiejar = get_cookiejar_from_response(response)
headers = get_headers_from_response(response)
url = "http://uac.10010.com/portal/Service/CreateImage?t=" + get_js_time(
)
resp = get_response_by_requests(url,
headers=headers,
cookie_jar=cookie_dict)
return resp.content, resp.cookies.get_dict()
def send_sms_code(request):
"""
登录发送短信验证码
:param request:
:return:
"""
ret_data = {}
try:
args = request.POST
session = request.session
if args.get("is_first", False) == "true":
username = args["username"].strip()
account_type = args["account_type"]
key = username + ACCOUNT_CRAWLING_SMS_HEADERS_SSDB_SUFFIX + account_type
ssdb_conn = get_ssdb_conn()
headers_data = ssdb_conn.get(key)
if not headers_data:
add_ajax_error_json(ret_data, "获取短信验证码失败")
return JsonResponse(ret_data)
headers_dict = json_loads(headers_data)
send_url = headers_dict.get("url", "")
session["send_url"] = send_url
session["last_send_time"] = time()
# 第一次会自动发送,默认为发送成功
res_json = {"stat": "ok", "info": {"sent": True}}
else:
last_send_time = session.get("last_send_time", 0)
need_sleep_time = max(last_send_time + SMS_SLEEP_TIME + 2 -
time(), 0) if last_send_time else 0
sleep(need_sleep_time)
send_url = session.get("send_url")
res_json = get_response_by_requests(
send_url, headers=DEFAULT_HEADERS).json()
if res_json.get("stat") == "ok" and res_json.get("info",
{}).get("sent"):
add_ajax_ok_json(ret_data)
else:
error_msg = res_json.get("info", {}).get("errorMessage")
add_ajax_error_json(ret_data, error_msg or "发送短信验证码失败")
except Exception:
add_ajax_error_json(ret_data, "发送短信验证码出错")
return JsonResponse(ret_data)
def get_unisecid_request(self, response):
cookiejar = get_cookiejar_from_response(response)
headers = get_headers_from_response(response)
the_time = get_js_time()
url = "https://uac.10010.com/oauth2/genqr?" + the_time
r = get_response_by_requests(url,
headers=headers,
cookie_jar=cookiejar)
cookie = r.headers.get('Set-Cookie')
result = {}
k_v_list = cookie.split(';')
name, value = k_v_list[0].split('=')
result['name'] = name
result['value'] = value
for k_v in islice(k_v_list, 1, None):
k, v = k_v.split('=', 1)
result[k] = v
return result
def get_need_captcha_response(self, response, username, pwd_type="02"):
"""
询问是否需要验证码
"""
cookiejar = get_cookiejar_from_response(response)
headers = get_headers_from_response(response)
the_time = get_js_time()
form_data = {
'userName':
username,
'pwdType':
pwd_type,
'_':
int(the_time) + 1,
'callback':
"jQuery1720" + str(randint(1E16, 1E17 - 1)) + "_" + the_time
}
url = "http://uac.10010.com/portal/Service/CheckNeedVerify?" + urlencode(
form_data)
return get_response_by_requests(url,
headers=headers,
cookie_jar=cookiejar)
def get_ckPwd_request(self, response, username, cookie_dict=None):
meta = response.meta
sleep(self._get_sms_send_sleep_time(meta))
the_time = get_js_time()
form_data = {
'mobile':
username,
'req_time':
the_time,
'_':
int(the_time) + 1,
'callback':
"jQuery1720" + str(randint(1E16, 1E17 - 1)) + "_" + the_time
}
# cookiejar = get_cookiejar_from_response(response)
headers = get_headers_from_response(response)
url = "https://uac.10010.com/portal/Service/SendCkMSG?" + urlencode(
form_data)
resp = get_response_by_requests(url,
headers=headers,
cookie_jar=cookie_dict)
meta["last_sms_time"] = time()
text = resp.text
return ('resultCode:"0000"' in text), text