adfd

zyz/hosipatl_name.py

@@ -35,9 +35,15 @@ def name_short():
              "光学", "国际", "连锁", "综合", "专科",]
     location = ["街道", "社区",]
     city_list = get_city_names()
-    stop_words = names + location + city_list
+    stop_words = names + location
     new_names = []
+    third_names = []
     for name in name_list:
+        city_tag = "flag"
+        for city in city_list:
+            if city in name:
+                city_tag = city
+            name = re.sub(city, '', name)
         for word in stop_words:
             name = re.sub(word, '', name)
         # 去除\t
@@ -48,12 +54,18 @@ def name_short():
         name = re.sub(r'\(.*?\)', '', name)
         # 去除 左英文括号，右中文括号
         name = re.sub(r'\(.*?\）', '', name)
+        if city_tag != "flag":
+            third_names.append(city_tag+name)
+        else:
+            third_names.append("无")
         new_names.append(name)
     df = pd.DataFrame()
     df['old_name'] = name_list
     df['new_name'] = new_names
+    df['core_name'] = third_names
+
     print(df.head(6))
-    df.to_csv("/home/gmuser/"+"12.csv",index=None)
+    df.to_csv("/home/gmuser/"+"hospital_name_21.csv",index=None)