初步添加自动审核功能

db/mysql.py

@@ -412,17 +412,19 @@ class ZxIeReview(Base):
     pk_ie_review = Column(INTEGER(11), primary_key=True, comment='自动审核主键')
     pk_phhd = Column(INTEGER(11), nullable=False, comment='报销案子主键')
     success = Column(BIT(1))
+    integrity = Column(BIT(1))
     has_settlement = Column(BIT(1))
     has_discharge = Column(BIT(1))
     has_cost = Column(BIT(1))
     full_page = Column(BIT(1))
     page_description = Column(String(255), comment='具体缺页描述')
+    consistency = Column(BIT(1), comment='三项资料一致性。0：不一致；1：一致')
     name_match = Column(CHAR(1), server_default=text("'0'"),
-                        comment='三项资料姓名是否一致。0：不一致；1：一致；2：结算单不一致；3：出院记录不一致；4：费用清单不一致；5：与报销申请对象不一致')
+                        comment='三项资料姓名是否一致。0：都不一致；1：一致；2：结算单不一致；3：出院记录不一致；4：费用清单不一致；5：与报销申请对象不一致')
     admission_date_match = Column(CHAR(1), server_default=text("'0'"),
-                                  comment='三项资料入院日期是否一致。0：不一致；1：一致；2：结算单不一致；3：出院记录不一致；4：费用清单不一致')
+                                  comment='三项资料入院日期是否一致。0：都不一致；1：一致；2：结算单不一致；3：出院记录不一致；4：费用清单不一致')
     discharge_date_match = Column(CHAR(1), server_default=text("'0'"),
-                                  comment='三项资料出院日期是否一致。0：不一致；1：一致；2：结算单不一致；3：出院记录不一致；4：费用清单不一致')
+                                  comment='三项资料出院日期是否一致。0：都不一致；1：一致；2：结算单不一致；3：出院记录不一致；4：费用清单不一致')
     create_time = Column(DateTime, server_default=text("CURRENT_TIMESTAMP"), comment='创建时间')
     creator = Column(String(255), comment='创建人')
     update_time = Column(DateTime, server_default=text("CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP"),

photo_review/auto_photo_review.py

@@ -16,7 +16,7 @@ from rapidfuzz import process, fuzz
 from sqlalchemy import update
 
 from db import MysqlSession
-from db.mysql import BdYljg, BdYlks, ZxIeResult, ZxIeCost, ZxIeDischarge, ZxIeSettlement, ZxPhhd, ZxPhrec
+from db.mysql import BdYljg, BdYlks, ZxIeResult, ZxIeCost, ZxIeDischarge, ZxIeSettlement, ZxPhhd, ZxPhrec, ZxIeReview
 from log import HOSTNAME
 from photo_review import PHHD_BATCH_SIZE, SLEEP_MINUTES, HOSPITAL_ALIAS, HOSPITAL_FILTER, DEPARTMENT_ALIAS, \
     DEPARTMENT_FILTER
@@ -390,6 +390,7 @@ def settlement_task(pk_phhd, settlement_list, identity):
     settlement_data["medical_expenses_str"] = handle_original_data(parse_money_result[0])
     settlement_data["medical_expenses"] = parse_money_result[1]
     save_or_update_ie(ZxIeSettlement, pk_phhd, settlement_data)
+    return settlement_data
 
 
 def discharge_task(pk_phhd, discharge_record, identity):
@@ -462,6 +463,7 @@ def discharge_task(pk_phhd, discharge_record, identity):
         if best_match:
             discharge_data["pk_ylks"] = best_match[2]
     save_or_update_ie(ZxIeDischarge, pk_phhd, discharge_data)
+    return discharge_data
 
 
 def cost_task(pk_phhd, cost_list, identity):
@@ -477,9 +479,10 @@ def cost_task(pk_phhd, cost_list, identity):
     cost_data["discharge_date"] = handle_date(cost_data["discharge_date_str"])
     cost_data["medical_expenses"] = handle_decimal(cost_data["medical_expenses_str"])
     save_or_update_ie(ZxIeCost, pk_phhd, cost_data)
+    return cost_data
 
 
-def photo_review(pk_phhd):
+def photo_review(pk_phhd, name):
     settlement_list = []
     discharge_record = []
     cost_list = []
@@ -499,15 +502,80 @@ def photo_review(pk_phhd):
 
     # 同一批图的标识
     identity = int(time.time())
-    settlement_task(pk_phhd, settlement_list, identity)
+    settlement_data = settlement_task(pk_phhd, settlement_list, identity)
-    discharge_task(pk_phhd, discharge_record, identity)
+    discharge_data = discharge_task(pk_phhd, discharge_record, identity)
-    cost_task(pk_phhd, cost_list, identity)
+    cost_data = cost_task(pk_phhd, cost_list, identity)
+
+    review_result = {
+        'pk_phhd': pk_phhd,
+    }
+    # 三项资料完整性判断
+    # 三项资料缺项判断
+    if (settlement_data['personal_account_payment'] + settlement_data['personal_cash_payment']
+            < settlement_data['medical_expenses']):
+        review_result['has_settlement'] = True
+    # TODO:出院记录和费用清单暂时没想好怎么判断
+    review_result['has_discharge'] = True
+    review_result['has_cost'] = True
+    # 三项资料缺页判断
+    # TODO:缺页需要对页码进行抽取，暂未训练相关模型
+    review_result['full_page'] = True
+
+    if (review_result['has_settlement'] and review_result['has_discharge'] and review_result['has_cost']
+            and review_result['full_page']):
+        review_result['integrity'] = True
+
+    # 三项资料一致性判断
+    # 姓名一致性
+    unique_name = {settlement_data['name'], discharge_data['name'], cost_data['name']}
+    if len(unique_name) == 1:
+        review_result['name_match'] = '1' if name == settlement_data['name'] else '5'
+    elif len(unique_name) == 2:
+        if settlement_data['name'] != discharge_data['name'] and settlement_data['name'] != cost_data['name']:
+            review_result['name_match'] = '2'
+        elif discharge_data['name'] != settlement_data['name'] and discharge_data['name'] != cost_data['name']:
+            review_result['name_match'] = '3'
+        else:
+            review_result['name_match'] = '4'
+    else:
+        review_result['name_match'] = '0'
+
+    # 住院日期一致性
+    if (settlement_data['admission_date'] == discharge_data['admission_date']
+            and settlement_data['discharge_date'] == discharge_data['discharge_date']):
+        review_result['admission_date_match'] = '1'
+    else:
+        review_result['admission_date_match'] = '0'
+
+    # 出院日期一致性
+    unique_discharge_date = {settlement_data['discharge_date'], discharge_data['discharge_date'],
+                             cost_data['discharge_date']}
+    if len(unique_discharge_date) == 1:
+        review_result['discharge_date_match'] = '1'
+    elif len(unique_discharge_date) == 2:
+        if (settlement_data['discharge_date'] != discharge_data['discharge_date']
+                and settlement_data['discharge_date'] != cost_data['discharge_date']):
+            review_result['discharge_date_match'] = '2'
+        elif (discharge_data['discharge_date'] != settlement_data['discharge_date']
+              and discharge_data['discharge_date'] != cost_data['discharge_date']):
+            review_result['discharge_date_match'] = '3'
+        else:
+            review_result['discharge_date_match'] = '4'
+    else:
+        review_result['discharge_date_match'] = '0'
+
+    if (review_result['name_match'] == '1' and review_result['admission_date_match'] == '1'
+            and review_result['discharge_date_match'] == '1'):
+        review_result['consistency'] = True
+
+    review_result['success'] = review_result['integrity'] and review_result['consistency']
+    save_or_update_ie(ZxIeReview, pk_phhd, review_result)
 
 
 def main():
     while 1:
         session = MysqlSession()
-        phhds = (session.query(ZxPhhd.pk_phhd)
+        phhds = (session.query(ZxPhhd.pk_phhd, ZxPhhd.cXm)
                  .join(ZxPhrec, ZxPhhd.pk_phhd == ZxPhrec.pk_phhd, isouter=True)
                  .filter(ZxPhhd.exsuccess_flag == "1")
                  .filter(ZxPhrec.pk_phrec.isnot(None))
@@ -524,7 +592,7 @@ def main():
                 pk_phhd = phhd.pk_phhd
                 logging.info(f"开始识别：{pk_phhd}")
                 start_time = time.time()
-                photo_review(pk_phhd)
+                photo_review(pk_phhd, phhd.cXm)
 
                 # 识别完成更新标识
                 session = MysqlSession()