添加医院名的分析处理

photo_review/photo_review.py

@@ -19,7 +19,8 @@ from photo_review import PATIENT_NAME, ADMISSION_DATE, DISCHARGE_DATE, MEDICAL_E
 from ucloud import ufile
 from util import image_util, util
 from util.data_util import handle_date, handle_decimal, parse_department, handle_name, \
-    handle_insurance_type, handle_original_data, handle_hospital, handle_department, handle_id, handle_age, parse_money
+    handle_insurance_type, handle_original_data, handle_hospital, handle_department, handle_id, handle_age, parse_money, \
+    parse_hospital
 
 
 # 合并信息抽取结果
@@ -242,11 +243,13 @@ def discharge_task(pk_phhd, discharge_record, identity):
     if hospitals:
         hospital_like_conditions = []
         for hospital in hospitals:
-            if hospital in HOSPITAL_ALIAS:
-                for hospital_alias in HOSPITAL_ALIAS[hospital]:
+            parsed_hospitals = parse_hospital(hospital)
+            for parsed_hospital in parsed_hospitals:
+                if parsed_hospital in HOSPITAL_ALIAS:
+                    for hospital_alias in HOSPITAL_ALIAS[parsed_hospital]:
                         hospital_like_conditions.append(BdYljg.name.like(f'%{hospital_alias}%'))
                 else:
-                hospital_like_conditions.append(BdYljg.name.like(f'%{hospital}%'))
+                    hospital_like_conditions.append(BdYljg.name.like(f'%{parsed_hospital}%'))
         session = MysqlSession()
         yljg = session.query(BdYljg.pk_yljg, BdYljg.name).filter(or_(*hospital_like_conditions)).limit(1).one_or_none()
         session.close()

util/data_util.py

@@ -173,3 +173,16 @@ def handle_age(string):
     string = string.split("岁")[0]
     num = re.sub(r'\D', '', string)
     return num[-3:]
+
+
+# 分析医院
+def parse_hospital(string):
+    result = []
+    if not string:
+        return result
+
+    string_without_brackets = string.replace(")", "").replace("）", "").replace("(", " ").replace("（", " ")
+    string_without_company = string_without_brackets.replace("有限公司", "")
+    simple_chinese_string = string_without_company.replace("醫", "医")
+    result += simple_chinese_string.split(" ")
+    return result