spider_get_and_check_dlt.py 9.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300
  1. # -*-coding: utf-8 -*-
  2. import os
  3. import sys
  4. sys.path.append(os.path.join(os.getcwd().split('auto_news_scheduler')[0], 'auto_news_scheduler'))
  5. import threading
  6. from datetime import datetime
  7. import time
  8. import httpx
  9. from tools_mongo_handle import MongoHandle
  10. from tools_logs_handle import LogsHandle
  11. from tools_send_email import SendEmail
  12. class GetData(object):
  13. def __init__(self, get_num=9999999):
  14. self.get_num = get_num
  15. self.url = 'https://webapi.sporttery.cn/gateway/lottery/getHistoryPageListV1.qry?gameNo=85&provinceId=0&pageSize={}&isVerify=1&pageNo=1'.format(get_num)
  16. self.logs_handle = LogsHandle()
  17. self.email_subject = 'dlt'
  18. self.email_title = '超级大乐透最新一期开奖查询对比'
  19. self.email_text = '获取数据时间:\n{0}\n{1}\n\n\n\n'.format(datetime.now().strftime("%Y-%m-%d %H:%M:%S"), ('-' * 90))
  20. self.logs_handle = LogsHandle()
  21. self.now_day = time.strftime('%Y-%m-%d', time.localtime())
  22. db = 'dlt'
  23. collection = 'dlt_' + self.now_day
  24. self.mongo = MongoHandle(db=db, collection=collection, del_db=False, del_collection=False, auto_remove=0)
  25. def main(self):
  26. data_list = self.req()
  27. result_data = self.data_handle(data_list)
  28. return result_data
  29. def req(self):
  30. resp = httpx.get(self.url)
  31. if resp.status_code != 200:
  32. print('state code: {}'.format(resp.status_code))
  33. log_detail = '访问失败, 状态码:{},url:{}'.format(resp.status_code, self.url)
  34. self.logs_handle.logs_write('auto_get_and_check_dlt', log_detail, 'error', False)
  35. exit(0)
  36. resp_json = resp.json()
  37. value = resp_json.setdefault('value')
  38. data_list = value.setdefault('list')
  39. if not data_list:
  40. self.logs_handle.logs_write('auto_get_and_check_dlt', '返回的数据为空, 获取数据失败', 'error', False)
  41. return
  42. print('已获取数据')
  43. return data_list
  44. def data_handle(self, data_list):
  45. result_data = []
  46. for d in data_list:
  47. numbers = d.setdefault('lotteryUnsortDrawresult')
  48. try:
  49. if len(numbers.split(' ')) < 7:
  50. continue
  51. except Exception as e:
  52. print('numbers: {}, err: {}'.format(numbers, e))
  53. continue
  54. red_list = numbers.split(' ')[:5]
  55. blue_list = numbers.split(' ')[5:]
  56. red_list.sort()
  57. blue_list.sort()
  58. try:
  59. # 切开红球,蓝球数组
  60. red1 = red_list[0]
  61. red2 = red_list[1]
  62. red3 = red_list[2]
  63. red4 = red_list[3]
  64. red5 = red_list[4]
  65. blue1 = blue_list[0]
  66. blue2 = blue_list[1]
  67. except Exception as e:
  68. print('红球或蓝球数据丢失')
  69. result_data.append({
  70. 'serial': d.setdefault('lotteryDrawNum'),
  71. 'red1': red1 or '',
  72. 'red2': red2 or '',
  73. 'red3': red3 or '',
  74. 'red4': red4 or '',
  75. 'red5': red5 or '',
  76. 'blue1': blue1 or '',
  77. 'blue2': blue2 or '',
  78. 'drawPdfUrl': d.setdefault('drawPdfUrl'),
  79. 'date': d.setdefault('lotteryDrawTime'),
  80. 'pool': d.setdefault('poolBalanceAfterdraw')
  81. })
  82. if result_data:
  83. return result_data
  84. else:
  85. self.logs_handle.logs_write('auto_get_and_check_dlt', '返回的数据为空, 获取数据失败', 'error', False)
  86. exit(0)
  87. class CheckMyDLT(object):
  88. def __init__(self, data):
  89. self.my_dlt = [
  90. ['10', '11', '16', '17', '18', '11', '12'],
  91. ['02', '03', '11', '12', '23', '05', '06'],
  92. ['07', '09', '15', '17', '22', '09', '11'],
  93. ['05', '06', '07', '34', '35', '02', '09'],
  94. ['09', '10', '11', '21', '22', '04', '05']
  95. ]
  96. self.data = data
  97. def main(self):
  98. print('开始数据对比')
  99. prepare_send_text, prepare_send_subject = self.process_text()
  100. self.send_data(prepare_send_subject, prepare_send_text)
  101. def process_text(self):
  102. text = ''
  103. serial_text = None
  104. subject = None
  105. for data in self.data:
  106. red_list = [data['red1'], data['red2'], data['red3'], data['red4'], data['red5']]
  107. blue_list = [data['blue1'], data['blue2']]
  108. # 只查询一期时, subject显示, 如果查询多期,则subject不显示
  109. if len(data) == 1:
  110. subject = '{}'.format(data['serial'])
  111. # 组成每期数据的text
  112. serial_text = 'serial: {}\t\tlottery draw date: {}\t\tbonus pool: {} RMB\n{}\nlottery draw num: {} + {}\n'.format(data['serial'], data['date'], data['pool'], '*' * 90,
  113. red_list, blue_list)
  114. for my_num in self.my_dlt:
  115. my_red_list = my_num[:5]
  116. my_blue_list = my_num[5:]
  117. # 使用列表推导式找出两个列表中都存在的元素
  118. red_common_elements = [element for element in red_list if element in my_red_list]
  119. blue_common_elements = [element for element in blue_list if element in my_blue_list]
  120. # 计算相等元素的数量
  121. red_equal_count = len(red_common_elements)
  122. blue_equal_count = len(blue_common_elements)
  123. serial_text += 'my nums: {} + {}\t\tred hit: {}\tblue hit: {}\n'.format(my_red_list, my_blue_list, red_equal_count, blue_equal_count)
  124. text += serial_text
  125. text += '{}\n\n\n\n'.format('*' * 90)
  126. return text, subject
  127. def send_data(self, subject, text):
  128. title = '超级大乐透最新一期开奖查询对比'
  129. SendEmail(subject, title, text).send()
  130. class SaveToDB(object):
  131. def __init__(self, data):
  132. self.logs_handle = LogsHandle()
  133. self.now_day = time.strftime('%Y-%m-%d', time.localtime())
  134. db = 'dlt'
  135. collection = 'dlt_' + self.now_day
  136. self.mongo = MongoHandle(db=db, collection=collection, del_db=False, del_collection=True, auto_remove=0)
  137. self.data = data
  138. def save_data(self):
  139. print('开始保存数据')
  140. for data in self.data:
  141. data_to_insert = {
  142. "serial": data.setdefault('serial'),
  143. "red1": data.setdefault('red1'),
  144. "red2": data.setdefault('red2'),
  145. "red3": data.setdefault('red3'),
  146. "red4": data.setdefault('red4'),
  147. "red5": data.setdefault('red5'),
  148. "blue1": data.setdefault('blue1'),
  149. "blue2": data.setdefault('blue2'),
  150. "date": data.setdefault('date'),
  151. "pool": data.setdefault('pool'),
  152. "drawPdfUrl": data.setdefault('drawPdfUrl'),
  153. "create_time": int(time.time()),
  154. "create_datetime": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
  155. }
  156. self.mongo.collection.insert_one(data_to_insert)
  157. print('数据已储存, 共储存数据{}条'.format(len(self.data)))
  158. class DLT(object):
  159. def start(self, n):
  160. # # 获取数据
  161. G = GetData(n)
  162. data = G.main()
  163. return data
  164. def check(self, data):
  165. # # 读取数据并发送到邮件
  166. Check = CheckMyDLT(data)
  167. Check.main()
  168. def mongo(self, data):
  169. # 存 mongodb
  170. Mongo = SaveToDB(data)
  171. Mongo.save_data()
  172. def main(self):
  173. L = LogsHandle()
  174. L.logs_write('auto_get_and_check_dlt', 'dlt任务开始', 'start', False)
  175. data = self.start(30)
  176. if data:
  177. tasks = [
  178. self.check,
  179. self.mongo
  180. ]
  181. threads = []
  182. for i in tasks:
  183. thread = threading.Thread(target=i, args=(data,))
  184. threads.append(thread)
  185. thread.start()
  186. for thread in threads:
  187. thread.join()
  188. L.logs_write('auto_get_and_check_dlt', 'dlt任务结束', 'start', False)
  189. print('done')
  190. else:
  191. L.logs_write('auto_get_and_check_dlt', '获取数据失败', 'error', False)
  192. class Luanch(object):
  193. def start(self, n):
  194. # # 获取数据
  195. G = GetData(n)
  196. data = G.main()
  197. return data
  198. def check(self, data):
  199. # # 读取数据并发送到邮件
  200. Check = CheckMyDLT(data)
  201. Check.main()
  202. def mongo(self, data):
  203. # 存 mongodb
  204. Mongo = SaveToDB(data)
  205. Mongo.save_data()
  206. def main(self):
  207. Logs = LogsHandle()
  208. Logs.logs_write('auto_get_and_check_dlt', 'dlt任务开始', 'start', False)
  209. data = self.start(30)
  210. if data:
  211. tasks = [
  212. self.check,
  213. self.mongo
  214. ]
  215. threads = []
  216. for i in tasks:
  217. thread = threading.Thread(target=i, args=(data,))
  218. threads.append(thread)
  219. thread.start()
  220. for thread in threads:
  221. thread.join()
  222. Logs.logs_write('auto_get_and_check_dlt', 'dlt任务结束', 'start', False)
  223. print('done')
  224. else:
  225. Logs.logs_write('auto_get_and_check_dlt', '获取数据失败', 'error', False)
  226. # if __name__ == '__main__':
  227. # L = Luanch()
  228. # L.main()
  229. # ## 单独获取数据
  230. # G = GetData()
  231. # data = G.main()
  232. # re_data = data[::-1]
  233. # save_txt = ''
  234. # for item in re_data:
  235. # save_txt += f'[[{item["red1"]}, {item["red2"]}, {item["red3"]}, {item["red4"]}, {item["red5"]}], [{item["blue1"]}, {item["blue2"]}]],\n'
  236. #
  237. # with open('dlt.txt', 'w') as f:
  238. # f.write(save_txt)