添加医保类型的区分

photo_review/util/data_util.py

@@ -85,16 +85,25 @@ def parse_department(string):
     result = []
     if not string:
         return result
-    result.append(handle_department(string))
+    string = re.sub(r'\([^()]*\)|\[[^\[\]]*\]|\{[^\{\}]*\}|（[^（）]*）|[^⺀-鿿]', '', string)[:255]
+    if string == "科":
+        return result
+    result.append(string)
     string_without_num = re.sub(r'\d|一|二|三|四|五|六|七|八|九|十', '', string)
+    if string == "科":
+        return result
     if string_without_num != string:
-        result.append(handle_department(string_without_num))
-    string_without_brackets = re.sub(r'\([^()]*\)|\[[^\[\]]*\]|\{[^\{\}]*\}|（[^（）]*）', "", string_without_num)
-    if string_without_brackets != string_without_num:
-        result.append(handle_department(string_without_brackets))
-    pure_string = string_without_brackets.split("科")[0] + "科"
-    if pure_string != string_without_brackets:
-        result.append(handle_department(pure_string))
+        result.append(string_without_num)
+    pure_string = string_without_num.split("科")[0] + "科"
+    if string == "科":
+        return result
+    if pure_string != string_without_num:
+        result.append(pure_string)
+    pure_string_without_io = pure_string.replace("内", "").replace("外", "")
+    if string == "科":
+        return result
+    if pure_string_without_io != pure_string:
+        result.append(pure_string)
     return result
 
 
@@ -109,7 +118,19 @@ def handle_name(string):
 def handle_insurance_type(string):
     if not string:
         return ""
-    return string.replace(":", "").replace("：", "")[:255]
+    worker_insurance_keys = ["社保", "城保", "职"]
+    villager_insurance_keys = ["农保", "居民"]
+    migrant_worker_insurance_keys = ["农民工"]
+    no_insurance_keys = ["自费", "全费"]
+    if any(key in string for key in worker_insurance_keys):
+        return "职工医保"
+    if any(key in string for key in villager_insurance_keys):
+        return "居民医保"
+    if any(key in string for key in migrant_worker_insurance_keys):
+        return "农民工医保"
+    if any(key in string for key in no_insurance_keys):
+        return "无医保"
+    return ""
 
 
 # 处理原始数据