feat: 添加当事人补充后的日志记录

- 添加当事人补充后的日志记录
- 其它方面优化

data_clean/api/http_api.py

@@ -93,8 +93,9 @@ class HttpSessionReuse:
 
     async def release(self):
         for i in range(len(self._session_pool)):
-            await self._session_pool[i].close()
-            log.info('close client session, session index :{}', i)
+            if self._session_pool[i]:
+                await self._session_pool[i].close()
+                log.info('close client session, session index :{}', i)
         pass
 
 

data_clean/handle/company_court_open_announcement.py

@@ -12,10 +12,10 @@ from data_clean.env.const import mongo_table_prefix
 from data_clean.exception.ruler_validation_exception import RulerValidationException
 from data_clean.utils.base_utils import *
 from data_clean.utils.case_utils import case_no_year_datetime
+from data_clean.utils.case_utils import get_case_party
 from data_clean.utils.date_utils import str_2_date_time, get_update_time, establish_state_time
 from data_clean.utils.party_name_verify_utils import person_name_list_verify, remove_null_party
 from data_clean.utils.str_utils import json_str_2_list
-from data_clean.utils.case_utils import get_case_party
 
 # 必须命名为dim_handle
 dim_handle = get_dim_handle(os.path.basename(__file__))
@@ -131,9 +131,6 @@ async def open_ann_date(row_data: dict) -> dict:
 @dim_handle.registry_row_func
 async def party_unknown(row_data: dict) -> dict:
     # 过滤当事人名字异常，Z某某、xxx
-    # row_data['plaintiff_info'] = remove_null_party(row_data['plaintiff_info'])
-    # row_data['defendant_info'] = remove_null_party(row_data['defendant_info'])
-    # row_data['litigant_info'] = remove_null_party(row_data['litigant_info'])
 
     plaintiff_info = json_str_2_list(row_data['plaintiff_info'], "name")
     defendant_info = json_str_2_list(row_data['defendant_info'], "name")
@@ -142,12 +139,40 @@ async def party_unknown(row_data: dict) -> dict:
     if not flag:
         result = await get_case_party(row_data['case_no'], source='open_court')
         if result:
-            row_data['plaintiff_info'] = remove_null_party(result['plaintiff_info'])
-            row_data['defendant_info'] = remove_null_party(result['defendant_info'])
-            row_data['litigant_info'] = remove_null_party(result['litigant_info'])
-            row_data['plaintiff'] = result['plaintiff']
-            row_data['defendant'] = result['defendant']
-            row_data['litigant'] = result['litigant']
+            rowkey = row_data['rowkey'] if 'rowkey' in row_data else None
+            tmp_plaintiff_info = remove_null_party(result['plaintiff_info'])
+            tmp_defendant_info = remove_null_party(result['defendant_info'])
+            tmp_litigant_info = remove_null_party(result['litigant_info'])
+            tmp_plaintiff = result['plaintiff']
+            tmp_defendant = result['defendant']
+            tmp_litigant = result['litigant']
+
+            await insert_one(mongo_table_prefix + 'info_cooa_party_set_value', {
+                "rowkey": rowkey,
+                "org_party": {
+                    'plaintiff_info': row_data['plaintiff_info'],
+                    'defendant_info': row_data['defendant_info'],
+                    'litigant_info': row_data['litigant_info'],
+                    'plaintiff': row_data['plaintiff'],
+                    'defendant': row_data['defendant'],
+                    'litigant': row_data['litigant'],
+                },
+                "new_party": {
+                    'plaintiff_info': tmp_plaintiff_info,
+                    'defendant_info': tmp_defendant_info,
+                    'litigant_info': tmp_litigant_info,
+                    'plaintiff': tmp_plaintiff,
+                    'defendant': tmp_defendant,
+                    'litigant': tmp_litigant,
+                }
+            })
+
+            row_data['plaintiff_info'] = tmp_plaintiff_info
+            row_data['defendant_info'] = tmp_defendant_info
+            row_data['litigant_info'] = tmp_litigant_info
+            row_data['plaintiff'] = tmp_plaintiff
+            row_data['defendant'] = tmp_defendant
+            row_data['litigant'] = tmp_litigant
             pass
         else:
             raise RulerValidationException("ccoa_003", "人名不符合规范：%s" % error_name)
@@ -158,14 +183,17 @@ async def party_unknown(row_data: dict) -> dict:
 
 async def test():
     res = await _get_max_establish_date(
-        ['bc702f0f5202342a9c1c75fbf9be9aff', 'b79d862faef595f33b166562bb3c18b6', '24cb269450f9262051dfcaa3dc389844'])
+        ['b13955ebaa99b06456803041e73c3888', ])
     print(res)
 
     pass
 
 
 if __name__ == '__main__':
-    import asyncio
+    row_data = {"rowkey": "a"}
 
-    asyncio.run(test())
+    rowkey = row_data['rowkey'] if 'rowkey' in row_data else None
+    print(rowkey)
+    # import asyncio
+    # asyncio.run(test())
     pass

data_clean/utils/asyncio_es_scan.py

@@ -45,4 +45,5 @@ class AsyncioElasticsearchFastScan:
             log.info("Processing a batch of data: {} - {}", (i + 1) * size, (i + 2) * size)
             yield res["hits"]["hits"]
         log.info("scan successful ! ")
+        await self.es.close()
         pass