item.py 3.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119
  1. """
  2. @author: luojunhui
  3. """
  4. import time
  5. default_single_video_table_fields = {
  6. "platform": "gzh",
  7. "article_title": None,
  8. "content_trace_id": None,
  9. "read_cnt": 0,
  10. "article_index": None,
  11. "out_account_name": None,
  12. "article_url": None,
  13. "url_unique_md5": None,
  14. "category": None,
  15. "publish_timestamp": None,
  16. "out_account_id": None,
  17. "cover_url": None,
  18. "crawler_timestamp": int(time.time()),
  19. "source_account": 1,
  20. "article_publish_type": None,
  21. "like_cnt": 0,
  22. "bad_status": 0,
  23. "tags": None,
  24. "video_oss_path": None,
  25. }
  26. default_account_table_fields = {
  27. "platform": 'Not NULL',
  28. "account_id": 'Not NULL',
  29. "account_name": 'Not NULL',
  30. "max_cursor": None,
  31. "account_init_date": None,
  32. "status": 0,
  33. "priority": 0,
  34. }
  35. default_candidate_account_table_fields = {
  36. "platform": 'Not NULL',
  37. "account_id": 'Not NULL',
  38. "account_name": 'Not NULL',
  39. "crawler_date": 'Not NULL',
  40. "title_list": "[]"
  41. }
  42. class Item(object):
  43. """
  44. format save to article meta table or single video source table
  45. """
  46. def __init__(self):
  47. self.item = {}
  48. def add(self, key, value):
  49. """
  50. add key value to item
  51. """
  52. self.item[key] = value
  53. def check_video_item(self):
  54. """
  55. check video item
  56. """
  57. fields = list(default_single_video_table_fields.keys())
  58. for field in fields:
  59. if self.item.get(field, None) is not None:
  60. continue
  61. else:
  62. self.item[field] = default_single_video_table_fields[field]
  63. def check_article_item(self):
  64. """
  65. check article item
  66. """
  67. return
  68. def check_account_item(self):
  69. """
  70. check account item
  71. """
  72. fields = list(default_account_table_fields.keys())
  73. for key in fields:
  74. if self.item.get(key, None) is not None:
  75. continue
  76. elif default_account_table_fields[key] == 'Not NULL':
  77. raise ValueError(f"{key} is not None, please check your account item")
  78. else:
  79. self.item[key] = default_account_table_fields[key]
  80. def check_candidate_account_item(self):
  81. """
  82. check association item
  83. """
  84. fields = list(default_candidate_account_table_fields.keys())
  85. for field in fields:
  86. if self.item.get(field, None) is not None:
  87. continue
  88. elif default_candidate_account_table_fields[field] == 'Not NULL':
  89. raise ValueError(f"{field} is not None, please check your account item")
  90. else:
  91. self.item[field] = default_candidate_account_table_fields[field]
  92. def check(self, source):
  93. """
  94. check item
  95. """
  96. match source:
  97. case "video":
  98. self.check_video_item()
  99. case "article":
  100. self.check_article_item()
  101. case "account":
  102. self.check_account_item()
  103. case "candidate_account":
  104. self.check_candidate_account_item()