item.py 1.5 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071
  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. "audit_status": 0,
  26. "category_status": 0
  27. }
  28. class Item(object):
  29. """
  30. format save to article meta table or single video source table
  31. """
  32. def __init__(self):
  33. self.item = {}
  34. def add(self, key, value):
  35. """
  36. add key value to item
  37. """
  38. self.item[key] = value
  39. def check_video_item(self):
  40. """
  41. check video item
  42. """
  43. fields = list(default_single_video_table_fields.keys())
  44. for field in fields:
  45. if self.item.get(field, None) is not None:
  46. continue
  47. else:
  48. self.item[field] = default_single_video_table_fields[field]
  49. def check_article_item(self):
  50. """
  51. check article item
  52. """
  53. return
  54. def check(self, source):
  55. """
  56. check item
  57. """
  58. match source:
  59. case "video":
  60. self.check_video_item()
  61. case "article":
  62. self.check_article_item()