add

local/hospital_name_short.py

@@ -30,7 +30,7 @@ def name_short():
     name_list = pd.DataFrame(list(result))[0].values.tolist()
     db.close()
     print(name_list[:10])
-    names = ["美容院","门诊部","医疗", "门诊", "研究所", "有限", "公司", "医学", "诊所", "中心","美容","医院","整形","外科",
+    names = ["中医","美容院","门诊部","医疗", "门诊", "研究所", "有限", "公司", "医学", "诊所", "中心","美容","医院","整形","外科",
              "集团", "卫生", "机构", "专业","皮肤","管理",
              "光学", "国际", "连锁", "综合", "专科","市","\*","\•"]
     location = ["街道", "社区",]
@@ -52,29 +52,33 @@ def name_short():
                 pass
             else:
                 first_names.append(name)
-                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
-                name = re.sub(r'\t', "", name)
-                # 去除 中文括号( )
-                name = re.sub(r'\（.*?\）', '', name)
-                # 去除 英文括号( )
-                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" and name != "":
-                    third_names.append(city_tag+name)
-                else:
-                    third_names.append("无")
-                if name != "":
+                if "第" in name:
                     second_names.append(name)
+                    third_names.append(name)
                 else:
-                    second_names.append(first_names[-1])
+                    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
+                    name = re.sub(r'\t', "", name)
+                    # 去除 中文括号( )
+                    name = re.sub(r'\（.*?\）', '', name)
+                    # 去除 英文括号( )
+                    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" and name != "":
+                        third_names.append(city_tag+name)
+                    else:
+                        third_names.append("无")
+                    if name != "":
+                        second_names.append(name)
+                    else:
+                        second_names.append(first_names[-1])
 
     df = pd.DataFrame()
     df['old_name'] = first_names