company_court_open_announcement.py 4.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136
  1. # -*- coding: utf-8 -*-
  2. # @Time : 2023/7/20 16:27
  3. # @Author : XuJiakai
  4. # @File : company_court_open_announcement
  5. # @Software: PyCharm
  6. import os
  7. from data_clean.api.hbase_api import bulk_get
  8. from data_clean.dim_handle_registry import get_dim_handle
  9. from data_clean.exception.ruler_validation_exception import RulerValidationException
  10. from data_clean.utils.base_utils import *
  11. from data_clean.utils.case_utils import case_no_year_datetime
  12. from data_clean.utils.date_utils import str_2_date_time, get_update_time, establish_state_time
  13. from data_clean.utils.party_name_verify_utils import person_name_verify
  14. from data_clean.utils.str_utils import json_str_2_list
  15. # 必须命名为dim_handle
  16. dim_handle = get_dim_handle(os.path.basename(__file__))
  17. # @dim_handle.registry_prefix_func
  18. async def prefix_func(dim_data: list):
  19. print("前置程序:", dim_data)
  20. raise ValueError("前置程序错误")
  21. pass
  22. @dim_handle.registry_postfix_func()
  23. async def post_func(dim_data: list):
  24. # print("后置程序:", dim_data)
  25. for r in dim_data:
  26. r['update_time'] = get_update_time()
  27. pass
  28. pass
  29. @dim_handle.registry_row_func
  30. async def party_intersect(row_data: dict) -> dict:
  31. # 判断当事人有交叉
  32. plaintiff_info = json_str_2_list(row_data['plaintiff_info'], "name")
  33. defendant_info = json_str_2_list(row_data['defendant_info'], "name")
  34. inter = list(set(plaintiff_info).intersection(set(defendant_info)))
  35. if len(inter) == 0:
  36. return row_data
  37. else:
  38. raise RulerValidationException("ccoa_001", "当事人有交叉:%s" % inter)
  39. pass
  40. async def _get_max_establish_date(company_ids: list):
  41. company_ids = [i for i in company_ids if i]
  42. res = await bulk_get('ng_rt_company', company_ids)
  43. res = [str_2_date_time(i['ESTIBLISH_TIME']) for i in res if 'ESTIBLISH_TIME' in i and i['ESTIBLISH_TIME']]
  44. if not res:
  45. return None
  46. return max(res)
  47. pass
  48. # 开庭时间相关过滤
  49. @dim_handle.registry_row_func
  50. async def open_ann_date(row_data: dict) -> dict:
  51. import datetime
  52. now = datetime.datetime.now()
  53. delta = datetime.timedelta(days=730) # 两年后
  54. max_date = now + delta
  55. start_date = get_or_none(row_data, 'start_date')
  56. case_no = get_or_none(row_data, 'case_no')
  57. if case_no is None and start_date is None:
  58. raise RulerValidationException("ccoa_007", "案号和开庭时间均为空")
  59. if start_date is None:
  60. return row_data
  61. try:
  62. this_date = str_2_date_time(row_data['start_date'])
  63. if this_date < establish_state_time:
  64. raise RulerValidationException("ccoa_002", "开庭时间早于建国时间:%s" % row_data['start_date'])
  65. if this_date > max_date:
  66. raise RulerValidationException("ccoa_006", "开庭时间在两年后:%s" % row_data['start_date'])
  67. part_keyno = json_str_2_list(row_data['plaintiff_info'], 'litigant_id') + json_str_2_list(
  68. row_data['defendant_info'], 'litigant_id')
  69. part_keyno = [i for i in part_keyno if i and len(i) == 32]
  70. max_establish_date = await _get_max_establish_date(part_keyno)
  71. if max_establish_date and this_date < max_establish_date:
  72. raise RulerValidationException("ccoa_004", "开庭时有公司未成立,最晚一个公司成立日期:%s,开庭时间:%s" % (
  73. max_establish_date, row_data['start_date']))
  74. case_no_year_dt = case_no_year_datetime(case_no)
  75. if case_no_year_dt and this_date < case_no_year_dt:
  76. raise RulerValidationException("ccoa_005", "案号大于开庭时间年份,案号:%s,开庭时间:%s" % (
  77. case_no, row_data['start_date']))
  78. except RulerValidationException as ex:
  79. if case_no is None:
  80. raise ex
  81. row_data['start_date'] = None
  82. pass
  83. return row_data
  84. pass
  85. @dim_handle.registry_row_func
  86. async def party_unknown(row_data: dict) -> dict:
  87. # 过滤当事人名字异常,Z某某、xxx
  88. plaintiff_info = json_str_2_list(row_data['plaintiff_info'], "name")
  89. defendant_info = json_str_2_list(row_data['defendant_info'], "name")
  90. li = plaintiff_info + defendant_info
  91. for i in li:
  92. if not person_name_verify(i):
  93. raise RulerValidationException("ccoa_003", "人名不符合规范:%s" % i)
  94. return row_data
  95. async def test():
  96. res = await _get_max_establish_date(
  97. ['bc702f0f5202342a9c1c75fbf9be9aff', 'b79d862faef595f33b166562bb3c18b6', '24cb269450f9262051dfcaa3dc389844'])
  98. print(res)
  99. pass
  100. if __name__ == '__main__':
  101. import asyncio
  102. asyncio.run(test())
  103. pass