PostgreSQL INSERT ON CONFLICT不存在则插入，存在则更新

近期有一个需求，向一张数据库表插入数据，如果是新数据则执行插入动作，如果插入的字段和已有字段重复，则更新该行对应的部分字段

1. 创建测试表

create table meta_data (
        id serial,
        user_id varchar(128) DEFAULT NULL,
        file_name varchar(1024) DEFAULT NULL,
        file_path varchar(1024) DEFAULT NULL,
        update_time TIMESTAMP DEFAULT NULL,
        UNIQUE (user_id,file_name)
    );


postgres=# \d meta_data
                                      Table "public.meta_data"
   Column    |            Type             |                       Modifiers                        
-------------+-----------------------------+--------------------------------------------------------
 id          | integer                     | not null default nextval('meta_data_id_seq'::regclass)
 user_id     | character varying(128)      | default NULL::character varying
 file_name   | character varying(1024)     | default NULL::character varying
 file_path   | character varying(1024)     | default NULL::character varying
 update_time | timestamp without time zone | 
Indexes:
    "meta_data_user_id_file_name_key" UNIQUE CONSTRAINT, btree (user_id, file_name)

2. 插入两条测试数据

INSERT INTO meta_data (
               user_id,
               file_name,
               file_path,
               UPDATE_TIME )
     VALUES ( 'user_id01',
              'file_name01',
              '/usr/local/file_name01',
              now())
ON CONFLICT (user_id, file_name) DO UPDATE SET file_path = EXCLUDED.file_path, UPDATE_TIME = EXCLUDED.UPDATE_TIME;    

INSERT INTO meta_data (
               user_id,
               file_name,
               file_path,
               UPDATE_TIME )
     VALUES ( 'user_id02',
              'file_name02',
              '/usr/local/file_name02',
              now())
ON CONFLICT (user_id, file_name) DO UPDATE SET file_path = EXCLUDED.file_path, UPDATE_TIME = EXCLUDED.UPDATE_TIME;

postgres=# select * from meta_data;
 id |  user_id  |  file_name  |       file_path        |        update_time         
----+-----------+-------------+------------------------+----------------------------
  1 | user_id01 | file_name01 | /usr/local/file_name01 | 2019-09-23 17:14:52.39878
  2 | user_id02 | file_name02 | /usr/local/file_name02 | 2019-09-23 17:14:53.118192
(2 rows)

3. 插入第三条测试数据，注意插入的字段user_id和file_name和第二条语句对应的字段是重复的

INSERT INTO meta_data (
               user_id,
               file_name,
               file_path,
               UPDATE_TIME )
     VALUES ( 'user_id02',
              'file_name02',
              '/usr/local/file_name03',
              now())
ON CONFLICT (user_id, file_name) DO UPDATE SET file_path = EXCLUDED.file_path, UPDATE_TIME = EXCLUDED.UPDATE_TIME;

postgres=# select * from meta_data;
 id |  user_id  |  file_name  |       file_path        |        update_time         
----+-----------+-------------+------------------------+----------------------------
  1 | user_id01 | file_name01 | /usr/local/file_name01 | 2019-09-23 17:14:52.39878
  2 | user_id02 | file_name02 | /usr/local/file_name03 | 2019-09-23 17:16:52.457696
(2 rows)

可以看到新插入的第三条语句其实是更新了已存在的第二条记录

PostgreSQL INSERT ON CONFLICT不存在则插入，存在则更新

猜你喜欢