search_winhc_summary.py 1.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172
  1. # -*- coding: utf-8 -*-
  2. # @Time : 2022/11/29 15:11
  3. # @Author : XuJiakai
  4. # @File : search_winhc_summary
  5. # @Software: PyCharm
  6. from sdk import get_es_sdk
  7. from utils import map_2_json_str
  8. es_sdk = get_es_sdk("new")
  9. skip_item = ['司法拍卖', '经营异常', '企业年报', '产品信息', '历史变更', '抽查检查', '软件著作权', '作品著作权', '失信信息', '双随机抽查', '融资信息', '股东信息', '行政许可',
  10. '专利', '行政处罚', '主要成员', '商标', '被执行人', '终本案件', '限制消费', '询价评估']
  11. skip_item.clear()
  12. def _summary_format(summary: map):
  13. from utils.dim_name_mapping import get_summary_map
  14. result_data = {}
  15. dim_name = get_summary_map()
  16. for i in summary:
  17. if i not in dim_name:
  18. continue
  19. info = dim_name[i]
  20. item_name = info['item_name']
  21. if item_name in result_data:
  22. result_data[item_name] = result_data[item_name] + summary[i]
  23. pass
  24. else:
  25. result_data[item_name] = summary[i]
  26. for i in skip_item:
  27. if i in result_data:
  28. del result_data[i]
  29. # print(result_data.keys())
  30. # print(map_2_json_str(result_data))
  31. return result_data
  32. pass
  33. def search_summary(company_id: str):
  34. dsl = {
  35. "size": 1000,
  36. "query": {
  37. "term": {
  38. "_id": {
  39. "value": company_id
  40. }
  41. }
  42. }
  43. }
  44. res = es_sdk.query(index='ng_rt_summary_company', doc_type='_doc', dsl=dsl)
  45. res_map = {}
  46. for i in res:
  47. m = i['summary']
  48. for j in m:
  49. res_map[j] = m[j]
  50. # print(map_2_json_str(res_map))
  51. return _summary_format(res_map)
  52. if __name__ == '__main__':
  53. search_summary(company_id='64406e27d43838e78aae09d8096ef7ff')
  54. pass