jd_spider_requests.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332
  1. import random
  2. import sys
  3. import time
  4. from jdlogger import logger
  5. from timer import Timer
  6. import requests
  7. from util import parse_json, get_session, get_sku_title,send_wechat
  8. from config import global_config
  9. from concurrent.futures import ProcessPoolExecutor
  10. class JdSeckill(object):
  11. def __init__(self):
  12. # 初始化信息
  13. self.session = get_session()
  14. self.sku_id = global_config.getRaw('config', 'sku_id')
  15. self.seckill_num = 2
  16. self.seckill_init_info = dict()
  17. self.seckill_url = dict()
  18. self.seckill_order_data = dict()
  19. self.timers = Timer()
  20. self.default_user_agent = global_config.getRaw('config', 'DEFAULT_USER_AGENT')
  21. def reserve(self):
  22. """
  23. 预约
  24. """
  25. self.__reserve()
  26. def seckill(self):
  27. """
  28. 抢购
  29. """
  30. self.__seckill()
  31. def wati_some_time(self):
  32. time.sleep(random.randint(100, 300) / 1000)
  33. def seckill_by_proc_pool(self, work_count=5):
  34. """
  35. 多进程进行抢购
  36. work_count:进程数量
  37. """
  38. with ProcessPoolExecutor(work_count) as pool:
  39. for i in range(work_count):
  40. pool.submit(self.seckill)
  41. def __reserve(self):
  42. """
  43. 预约
  44. """
  45. self.login()
  46. while True:
  47. try:
  48. self.make_reserve()
  49. except Exception as e:
  50. logger.info('预约发生异常!', e)
  51. self.wati_some_time()
  52. def __seckill(self):
  53. """
  54. 抢购
  55. """
  56. self.login()
  57. while True:
  58. try:
  59. self.request_seckill_url()
  60. self.request_seckill_checkout_page()
  61. self.submit_seckill_order()
  62. except Exception as e:
  63. logger.info('抢购发生异常,稍后继续执行!', e)
  64. self.wati_some_time()
  65. def login(self):
  66. for flag in range(1, 3):
  67. try:
  68. targetURL = 'https://order.jd.com/center/list.action'
  69. payload = {
  70. 'rid': str(int(time.time() * 1000)),
  71. }
  72. resp = self.session.get(
  73. url=targetURL, params=payload, allow_redirects=False)
  74. if resp.status_code == requests.codes.OK:
  75. logger.info('校验是否登录[成功]')
  76. logger.info('用户:{}'.format(self.get_username()))
  77. return True
  78. else:
  79. logger.info('校验是否登录[失败]')
  80. logger.info('请重新输入cookie')
  81. time.sleep(1)
  82. continue
  83. except Exception as e:
  84. logger.info('第【%s】次失败请重新获取cookie', flag)
  85. time.sleep(1)
  86. continue
  87. sys.exit(1)
  88. def make_reserve(self):
  89. """商品预约"""
  90. logger.info('商品名称:{}'.format(get_sku_title()))
  91. url = 'https://yushou.jd.com/youshouinfo.action?'
  92. payload = {
  93. 'callback': 'fetchJSON',
  94. 'sku': self.sku_id,
  95. '_': str(int(time.time() * 1000)),
  96. }
  97. headers = {
  98. 'User-Agent': self.default_user_agent,
  99. 'Referer': 'https://item.jd.com/{}.html'.format(self.sku_id),
  100. }
  101. resp = self.session.get(url=url, params=payload, headers=headers)
  102. resp_json = parse_json(resp.text)
  103. reserve_url = resp_json.get('url')
  104. self.timers.start()
  105. while True:
  106. try:
  107. self.session.get(url='https:' + reserve_url)
  108. logger.info('预约成功,已获得抢购资格 / 您已成功预约过了,无需重复预约')
  109. if global_config.getRaw('messenger', 'enable') == 'true':
  110. success_message = "预约成功,已获得抢购资格 / 您已成功预约过了,无需重复预约"
  111. send_wechat(success_message)
  112. break
  113. except Exception as e:
  114. logger.error('预约失败正在重试...')
  115. def get_username(self):
  116. """获取用户信息"""
  117. url = 'https://passport.jd.com/user/petName/getUserInfoForMiniJd.action'
  118. payload = {
  119. 'callback': 'jQuery'.format(random.randint(1000000, 9999999)),
  120. '_': str(int(time.time() * 1000)),
  121. }
  122. headers = {
  123. 'User-Agent': self.default_user_agent,
  124. 'Referer': 'https://order.jd.com/center/list.action',
  125. }
  126. resp = self.session.get(url=url, params=payload, headers=headers)
  127. try_count = 5
  128. while not resp.text.startswith("jQuery"):
  129. try_count = try_count - 1
  130. if try_count > 0:
  131. resp = self.session.get(url=url, params=payload, headers=headers)
  132. else:
  133. break
  134. self.wati_some_time()
  135. # 响应中包含了许多用户信息,现在在其中返回昵称
  136. # jQuery2381773({"imgUrl":"//storage.360buyimg.com/i.imageUpload/xxx.jpg","lastLoginTime":"","nickName":"xxx","plusStatus":"0","realName":"xxx","userLevel":x,"userScoreVO":{"accountScore":xx,"activityScore":xx,"consumptionScore":xxxxx,"default":false,"financeScore":xxx,"pin":"xxx","riskScore":x,"totalScore":xxxxx}})
  137. return parse_json(resp.text).get('nickName')
  138. def get_seckill_url(self):
  139. """获取商品的抢购链接
  140. 点击"抢购"按钮后,会有两次302跳转,最后到达订单结算页面
  141. 这里返回第一次跳转后的页面url,作为商品的抢购链接
  142. :return: 商品的抢购链接
  143. """
  144. url = 'https://itemko.jd.com/itemShowBtn'
  145. payload = {
  146. 'callback': 'jQuery{}'.format(random.randint(1000000, 9999999)),
  147. 'skuId': self.sku_id,
  148. 'from': 'pc',
  149. '_': str(int(time.time() * 1000)),
  150. }
  151. headers = {
  152. 'User-Agent': self.default_user_agent,
  153. 'Host': 'itemko.jd.com',
  154. 'Referer': 'https://item.jd.com/{}.html'.format(self.sku_id),
  155. }
  156. while True:
  157. resp = self.session.get(url=url, headers=headers, params=payload)
  158. resp_json = parse_json(resp.text)
  159. if resp_json.get('url'):
  160. # https://divide.jd.com/user_routing?skuId=8654289&sn=c3f4ececd8461f0e4d7267e96a91e0e0&from=pc
  161. router_url = 'https:' + resp_json.get('url')
  162. # https://marathon.jd.com/captcha.html?skuId=8654289&sn=c3f4ececd8461f0e4d7267e96a91e0e0&from=pc
  163. seckill_url = router_url.replace(
  164. 'divide', 'marathon').replace(
  165. 'user_routing', 'captcha.html')
  166. logger.info("抢购链接获取成功: %s", seckill_url)
  167. return seckill_url
  168. else:
  169. logger.info("抢购链接获取失败,稍后自动重试")
  170. self.wati_some_time()
  171. def request_seckill_url(self):
  172. """访问商品的抢购链接(用于设置cookie等"""
  173. logger.info('用户:{}'.format(self.get_username()))
  174. logger.info('商品名称:{}'.format(get_sku_title()))
  175. self.timers.start()
  176. self.seckill_url[self.sku_id] = self.get_seckill_url()
  177. logger.info('访问商品的抢购连接...')
  178. headers = {
  179. 'User-Agent': self.default_user_agent,
  180. 'Host': 'marathon.jd.com',
  181. 'Referer': 'https://item.jd.com/{}.html'.format(self.sku_id),
  182. }
  183. self.session.get(
  184. url=self.seckill_url.get(
  185. self.sku_id),
  186. headers=headers,
  187. allow_redirects=False)
  188. def request_seckill_checkout_page(self):
  189. """访问抢购订单结算页面"""
  190. logger.info('访问抢购订单结算页面...')
  191. url = 'https://marathon.jd.com/seckill/seckill.action'
  192. payload = {
  193. 'skuId': self.sku_id,
  194. 'num': self.seckill_num,
  195. 'rid': int(time.time())
  196. }
  197. headers = {
  198. 'User-Agent': self.default_user_agent,
  199. 'Host': 'marathon.jd.com',
  200. 'Referer': 'https://item.jd.com/{}.html'.format(self.sku_id),
  201. }
  202. self.session.get(url=url, params=payload, headers=headers, allow_redirects=False)
  203. def _get_seckill_init_info(self):
  204. """获取秒杀初始化信息(包括:地址,发票,token)
  205. :return: 初始化信息组成的dict
  206. """
  207. logger.info('获取秒杀初始化信息...')
  208. url = 'https://marathon.jd.com/seckillnew/orderService/pc/init.action'
  209. data = {
  210. 'sku': self.sku_id,
  211. 'num': self.seckill_num,
  212. 'isModifyAddress': 'false',
  213. }
  214. headers = {
  215. 'User-Agent': self.default_user_agent,
  216. 'Host': 'marathon.jd.com',
  217. }
  218. resp = self.session.post(url=url, data=data, headers=headers)
  219. return parse_json(resp.text)
  220. def _get_seckill_order_data(self):
  221. """生成提交抢购订单所需的请求体参数
  222. :return: 请求体参数组成的dict
  223. """
  224. logger.info('生成提交抢购订单所需参数...')
  225. # 获取用户秒杀初始化信息
  226. self.seckill_init_info[self.sku_id] = self._get_seckill_init_info()
  227. init_info = self.seckill_init_info.get(self.sku_id)
  228. default_address = init_info['addressList'][0] # 默认地址dict
  229. invoice_info = init_info.get('invoiceInfo', {}) # 默认发票信息dict, 有可能不返回
  230. token = init_info['token']
  231. data = {
  232. 'skuId': self.sku_id,
  233. 'num': self.seckill_num,
  234. 'addressId': default_address['id'],
  235. 'yuShou': 'true',
  236. 'isModifyAddress': 'false',
  237. 'name': default_address['name'],
  238. 'provinceId': default_address['provinceId'],
  239. 'cityId': default_address['cityId'],
  240. 'countyId': default_address['countyId'],
  241. 'townId': default_address['townId'],
  242. 'addressDetail': default_address['addressDetail'],
  243. 'mobile': default_address['mobile'],
  244. 'mobileKey': default_address['mobileKey'],
  245. 'email': default_address.get('email', ''),
  246. 'postCode': '',
  247. 'invoiceTitle': invoice_info.get('invoiceTitle', -1),
  248. 'invoiceCompanyName': '',
  249. 'invoiceContent': invoice_info.get('invoiceContentType', 1),
  250. 'invoiceTaxpayerNO': '',
  251. 'invoiceEmail': '',
  252. 'invoicePhone': invoice_info.get('invoicePhone', ''),
  253. 'invoicePhoneKey': invoice_info.get('invoicePhoneKey', ''),
  254. 'invoice': 'true' if invoice_info else 'false',
  255. 'password': '',
  256. 'codTimeType': 3,
  257. 'paymentType': 4,
  258. 'areaCode': '',
  259. 'overseas': 0,
  260. 'phone': '',
  261. 'eid': global_config.getRaw('config', 'eid'),
  262. 'fp': global_config.getRaw('config', 'fp'),
  263. 'token': token,
  264. 'pru': ''
  265. }
  266. return data
  267. def submit_seckill_order(self):
  268. """提交抢购(秒杀)订单
  269. :return: 抢购结果 True/False
  270. """
  271. url = 'https://marathon.jd.com/seckillnew/orderService/pc/submitOrder.action'
  272. payload = {
  273. 'skuId': self.sku_id,
  274. }
  275. self.seckill_order_data[self.sku_id] = self._get_seckill_order_data()
  276. logger.info('提交抢购订单...')
  277. headers = {
  278. 'User-Agent': self.default_user_agent,
  279. 'Host': 'marathon.jd.com',
  280. 'Referer': 'https://marathon.jd.com/seckill/seckill.action?skuId={0}&num={1}&rid={2}'.format(
  281. self.sku_id, self.seckill_num, int(time.time())),
  282. }
  283. resp = self.session.post(
  284. url=url,
  285. params=payload,
  286. data=self.seckill_order_data.get(
  287. self.sku_id),
  288. headers=headers)
  289. resp_json = parse_json(resp.text)
  290. # 返回信息
  291. # 抢购失败:
  292. # {'errorMessage': '很遗憾没有抢到,再接再厉哦。', 'orderId': 0, 'resultCode': 60074, 'skuId': 0, 'success': False}
  293. # {'errorMessage': '抱歉,您提交过快,请稍后再提交订单!', 'orderId': 0, 'resultCode': 60017, 'skuId': 0, 'success': False}
  294. # {'errorMessage': '系统正在开小差,请重试~~', 'orderId': 0, 'resultCode': 90013, 'skuId': 0, 'success': False}
  295. # 抢购成功:
  296. # {"appUrl":"xxxxx","orderId":820227xxxxx,"pcUrl":"xxxxx","resultCode":0,"skuId":0,"success":true,"totalMoney":"xxxxx"}
  297. if resp_json.get('success'):
  298. order_id = resp_json.get('orderId')
  299. total_money = resp_json.get('totalMoney')
  300. pay_url = 'https:' + resp_json.get('pcUrl')
  301. logger.info(
  302. '抢购成功,订单号:{}, 总价:{}, 电脑端付款链接:{}'.format(order_id,total_money,pay_url)
  303. )
  304. if global_config.getRaw('messenger', 'enable') == 'true':
  305. success_message = "抢购成功,订单号:{}, 总价:{}, 电脑端付款链接:{}".format(order_id, total_money, pay_url)
  306. send_wechat(success_message)
  307. return True
  308. else:
  309. logger.info('抢购失败,返回信息:{}'.format(resp_json))
  310. if global_config.getRaw('messenger', 'enable') == 'true':
  311. error_message = '抢购失败,返回信息:{}'.format(resp_json)
  312. send_wechat(error_message)
  313. return False