RedisDB.py 2.4 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889
  1. # -*- coding: utf-8 -*-
  2. """
  3. Created on 2024-02-27
  4. ---------
  5. @summary: redis 去重
  6. ---------
  7. @author: Lzz
  8. """
  9. import hashlib
  10. import redis
  11. import setting
  12. class RedisFilter:
  13. def __init__(self, url=None, expire_time=None):
  14. if not url:
  15. url = setting.REDIS_URL
  16. self.redis_db = redis.StrictRedis.from_url(url)
  17. self._ex = expire_time or 86400 * 365 * 1 # 1年 = 86400 * 365 * 1
  18. def __repr__(self):
  19. return "<RedisFilter: {}>".format(self.redis_db)
  20. def exists(self, key):
  21. """全量检索"""
  22. if self.redis_db.exists(key) > 0:
  23. return True
  24. return False
  25. def add(self, keys):
  26. """
  27. 添加数据
  28. @param keys: 检查关键词在 redis 中是否存在,支持列表批量
  29. @return: list / 单个值(添加失败返回False, 添加成功返回True)
  30. """
  31. is_list = isinstance(keys, list)
  32. keys = keys if is_list else [keys]
  33. is_added = []
  34. for key in keys:
  35. pkey = "pylist_" + self.fingerprint(key)
  36. if not self.exists(pkey):
  37. is_added.append(self.redis_db.set(pkey, 1, ex=self._ex))
  38. else:
  39. is_added.append(False)
  40. return is_added if is_list else is_added[0]
  41. def get(self, keys):
  42. """
  43. 检查数据是否存在
  44. @param keys: list / 单个值
  45. @return: list / 单个值 (存在返回True 不存在返回False)
  46. """
  47. is_list = isinstance(keys, list)
  48. keys = keys if is_list else [keys]
  49. is_exist = []
  50. for key in keys:
  51. pkey = "pylist_" + self.fingerprint(key)
  52. is_exist.append(self.exists(pkey))
  53. # 判断数据本身是否重复
  54. temp_set = set()
  55. for i, key in enumerate(keys):
  56. if key in temp_set:
  57. is_exist[i] = True
  58. else:
  59. temp_set.add(key)
  60. return is_exist if is_list else is_exist[0]
  61. def fingerprint(self, *args):
  62. """
  63. @summary: 获取唯一的64位值,获取唯一数据指纹
  64. ---------
  65. @param args: 去重数据集合
  66. ---------
  67. @result: 5580c91ea29bf5bd963f4c08dfcacd983566e44ecea1735102bc380576fd6f30
  68. """
  69. args = sorted(args)
  70. sha256 = hashlib.sha256()
  71. for arg in args:
  72. sha256.update(str(arg).encode())
  73. return sha256.hexdigest()