Merge branch 'zhao22' into 'master'

修改医院名称脚本

See merge request !88

local/hospital_name_short.py

@@ -32,40 +32,50 @@ def name_short():
     print(name_list[:10])
     names = ["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",
              "医美", "集团", "卫生", "机构", "专业","皮肤","管理",
-             "光学", "国际", "连锁", "综合", "专科",]
+             "光学", "国际", "连锁", "综合", "专科","市","*"]
     location = ["街道", "社区",]
     city_list = get_city_names()
+    city_list += ["香港","澳门","韩国"]
+
+    for word in ["市中心","丰台","科","东方"]:
+        if word in city_list:
+            city_list.remove(word)
+
     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
-        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":
-            third_names.append(city_tag+name)
-        else:
-            third_names.append("无")
-        new_names.append(name)
+        for test_word in ["测试","测试1","ceshi","ceshi1"]:
+            if test_word in name:
+                pass
+            else:
+                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":
+                    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("/tmp/"+"name_22.csv",index=None)
+    df.to_csv("/tmp/"+"name_28.csv",index=None)
 
 
 
@@ -89,7 +99,7 @@ def name_process(name):
     project_tags = ["口腔","植发","牙","皮肤","眼","外科","美容","整形","烧伤","胸","丰胸","美胸","祛痘","祛斑","脱毛",
                     "创伤","除疤","半永久","纹绣","纹眉"]
     names = ["医疗","门诊","研究所","有限","公司","医学","诊所","中心","医美","集团","卫生","机构","专业",
-             "光学","国际","连锁","综合","专科",""]
+             "光学","国际","连锁","综合","专科","市"]
     location = ["街道","社区",]
     stop_words = project_tags + names + location
     for word in stop_words: