gea_his_spider.py 7.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219
  1. # -*- coding: utf-8 -*-
  2. # Author : Charley
  3. # Python : 3.10.8
  4. # Date : 2026/2/3 14:45
  5. import inspect
  6. import requests
  7. import user_agent
  8. from loguru import logger
  9. from mysql_pool import MySQLConnectionPool
  10. from tenacity import retry, stop_after_attempt, wait_fixed
  11. """
  12. https://grading11.com/cert
  13. 2511000001-2511000020
  14. 2512000001-2512000945
  15. 2601000001-2601000779
  16. """
  17. logger.remove()
  18. logger.add("./logs/{time:YYYYMMDD}.log", encoding='utf-8', rotation="00:00",
  19. format="[{time:YYYY-MM-DD HH:mm:ss.SSS}] {level} {message}",
  20. level="DEBUG", retention="7 day")
  21. def after_log(retry_state):
  22. """
  23. retry 回调
  24. :param retry_state: RetryCallState 对象
  25. """
  26. # 检查 args 是否存在且不为空
  27. if retry_state.args and len(retry_state.args) > 0:
  28. log = retry_state.args[0] # 获取传入的 logger
  29. else:
  30. log = logger # 使用全局 logger
  31. if retry_state.outcome.failed:
  32. log.warning(
  33. f"Function '{retry_state.fn.__name__}', Attempt {retry_state.attempt_number} Times")
  34. else:
  35. log.info(f"Function '{retry_state.fn.__name__}', Attempt {retry_state.attempt_number} succeeded")
  36. @retry(stop=stop_after_attempt(5), wait=wait_fixed(1), after=after_log)
  37. def get_proxys(log):
  38. """
  39. 获取代理配置
  40. :param log: 日志对象
  41. :return: 代理字典
  42. """
  43. tunnel = "x371.kdltps.com:15818"
  44. kdl_username = "t13753103189895"
  45. kdl_password = "o0yefv6z"
  46. try:
  47. proxies = {
  48. "http": "http://%(user)s:%(pwd)s@%(proxy)s/" % {"user": kdl_username, "pwd": kdl_password, "proxy": tunnel},
  49. "https": "http://%(user)s:%(pwd)s@%(proxy)s/" % {"user": kdl_username, "pwd": kdl_password, "proxy": tunnel}
  50. }
  51. return proxies
  52. except Exception as e:
  53. log.error(f"Error getting proxy: {e}")
  54. raise e
  55. @retry(stop=stop_after_attempt(5), wait=wait_fixed(1), after=after_log)
  56. def get_cert_data(log, cert_id, sql_pool):
  57. """
  58. 获取证书数据并保存到数据库
  59. :param log: 日志对象
  60. :param cert_id: 证书ID
  61. :param sql_pool: 数据库连接池
  62. """
  63. log.debug(f'{inspect.currentframe().f_code.co_name} -> 开始获取证书数据:{cert_id}')
  64. headers = {
  65. "content-type": "application/json;charset=UTF-8",
  66. "user-agent": user_agent.generate_user_agent()
  67. }
  68. url = "https://grading11.com/api/vendure/getData"
  69. data = [{
  70. "customQuery": {
  71. "query": "ordercard-query",
  72. "metadata": {
  73. # "certNumber": "2511000001"
  74. "certNumber": str(cert_id)
  75. }
  76. }
  77. }]
  78. try:
  79. response = requests.post(url, headers=headers, json=data, timeout=22, proxies=get_proxys(log))
  80. # response = requests.post(url, headers=headers, json=data, timeout=22)
  81. # print(response.text)
  82. response.raise_for_status()
  83. resp_json = response.json()
  84. orderCard = resp_json.get("orderCard", {})
  85. if not orderCard:
  86. log.error(f"Error getting orderCard: {resp_json}")
  87. sql_pool.update_one("UPDATE gea_task SET state = 3 WHERE cert_id = %s", (cert_id,))
  88. return
  89. # 提取评分数据
  90. totalGrade = orderCard.get("totalGrade")
  91. centeringTotalSubGrade = orderCard.get("centeringTotalSubGrade")
  92. surfacesTotalSubGrade = orderCard.get("surfacesTotalSubGrade")
  93. edgesTotalSubGrade = orderCard.get("edgesTotalSubGrade")
  94. cornersTotalSubGrade = orderCard.get("cornersTotalSubGrade")
  95. createdAt = orderCard.get("createdAt")
  96. date_graded = createdAt[:10] if createdAt else None
  97. remark = orderCard.get("remark")
  98. # 提取图片链接
  99. try:
  100. front_image = orderCard.get("frontSlabbedImage", {}).get("source")
  101. except AttributeError:
  102. front_image = None
  103. try:
  104. back_image = orderCard.get("backSlabbedImage", {}).get("source")
  105. except AttributeError:
  106. back_image = None
  107. # 提取卡片信息
  108. """
  109. 2023 Pokémon Trading Card Game English
  110. 【SVP】POKÉMON X VAN GOGH MUSEUM 085
  111. Pikachu with Grey Felt Hat
  112. PROMO
  113. """
  114. tag_card = orderCard.get("card", {})
  115. cardName = tag_card.get("cardName") # Pikachu with Grey Felt Hat
  116. cardNumber = tag_card.get("cardNumber") # 085
  117. brand = tag_card.get("brand", {}).get("name") # Pokémon Trading Card Game
  118. releaseYear = tag_card.get("releaseYear") # 2023
  119. language = tag_card.get("language") # en
  120. try:
  121. cardSet = tag_card.get("cardSet", {}).get("name") # 【SVP】POKÉMON X VAN GOGH MUSEUM
  122. except AttributeError:
  123. cardSet = None
  124. try:
  125. rarity = tag_card.get("rarity", {}).get("name") # PROMO
  126. except AttributeError:
  127. rarity = None
  128. data_dict = {
  129. "cert_id": cert_id,
  130. "total_grade": totalGrade,
  131. "centering_grade": centeringTotalSubGrade,
  132. "surfaces_grade": surfacesTotalSubGrade,
  133. "edges_grade": edgesTotalSubGrade,
  134. "corners_grade": cornersTotalSubGrade,
  135. "date_graded": date_graded,
  136. "remark": remark,
  137. "front_image": front_image,
  138. "back_image": back_image,
  139. "card_name": cardName,
  140. "card_number": cardNumber,
  141. "brand": brand,
  142. "release_year": releaseYear,
  143. "language": language,
  144. "card_set": cardSet,
  145. "rarity": rarity
  146. }
  147. # print(data_dict)
  148. sql_pool.insert_one_or_dict(table="gea_record", data=data_dict)
  149. sql_pool.update_one("UPDATE gea_task SET state = 1 WHERE cert_id = %s", (cert_id,))
  150. except requests.RequestException as e:
  151. log.error(f"Request failed for cert_id {cert_id}: {e}")
  152. sql_pool.update_one(
  153. "UPDATE gea_task SET state = 2 WHERE cert_id = %s",
  154. (cert_id,)
  155. )
  156. raise
  157. except Exception as e:
  158. log.error(f"Error processing cert_id {cert_id}: {e}")
  159. sql_pool.update_one(
  160. "UPDATE gea_task SET state = 2 WHERE cert_id = %s",
  161. (cert_id,)
  162. )
  163. raise
  164. @retry(stop=stop_after_attempt(100), wait=wait_fixed(3600), after=after_log)
  165. def gea_his_main(log):
  166. """
  167. 主函数
  168. :param log: logger对象
  169. """
  170. log.info(
  171. f'开始运行 {inspect.currentframe().f_code.co_name} 爬虫任务....................................................')
  172. # 配置 MySQL 连接池
  173. sql_pool = MySQLConnectionPool(log=log)
  174. if not sql_pool.check_pool_health():
  175. log.error("数据库连接池异常")
  176. raise RuntimeError("数据库连接池异常")
  177. try:
  178. sql_id_list = sql_pool.select_all("SELECT cert_id FROM gea_task WHERE state != 1")
  179. sql_id_list = [item[0] for item in sql_id_list]
  180. log.info(f'共 {len(sql_id_list)} 个证书待处理')
  181. for cert_id in sql_id_list:
  182. try:
  183. get_cert_data(log, cert_id, sql_pool)
  184. except Exception as e:
  185. log.error(f'loop -> get_cert_data, error: {e}')
  186. sql_pool.update_one("UPDATE gea_task SET state = 2 WHERE cert_id = %s", (cert_id,))
  187. except Exception as e:
  188. log.error(f'{inspect.currentframe().f_code.co_name} error: {e}')
  189. finally:
  190. log.info(f'爬虫程序 {inspect.currentframe().f_code.co_name} 运行结束,等待下一轮的采集任务............')
  191. if __name__ == '__main__':
  192. gea_his_main(logger)