如何获取最新记录

Question

我有一张这样的表：

+--------+-------------+----------+---------+
| PK(id) | FK(user_id) | ctime    | comment |
+--------+-------------+----------+---------+
| 1      | 1           | 20170101 | "Haha"  |
+--------+-------------+----------+---------+
| 2      | 2           | 20170102 | "Nope"  |
+--------+-------------+----------+---------+
| 3      | 2           | 20170104 | "Geez"  |
+--------+-------------+----------+---------+
| 4      | 1           | 20170110 | "Gone"  |
+--------+-------------+----------+---------+

我希望检索每个FK(user_id) 的最新记录，如下所示：

+--------+-------------+----------+---------+
| PK(id) | FK(user_id) | ctime    | comment |
+--------+-------------+----------+---------+
| 3      | 2           | 20170104 | "Geez"  |
+--------+-------------+----------+---------+
| 4      | 1           | 20170110 | "Gone"  |
+--------+-------------+----------+---------+

所以我尝试了SELECT DISTINCT T.* FROM my_table AS T 和SELECT DISTINCT T.user_id FROM my_table AS T； 但是，它们不起作用！

我尝试了 GROUP BY 语句，如下所示，

SELECT T.* FROM my_table AS T GROUP BY `user_id` DESC

它确实像我期望的那样完美。 于是我开始想办法把SQL翻译成Django！

首先，我尝试使用 RawSQL：

from django.db.models.expressions import RawSQL

def _select_latest_rows(model_class, target_column_name, query_set=None):
    query_set = query_set or model_class.objects
    table_name = model_class._meta.db_table
    raw_sql = '''
    SELECT * FROM %s GROUP BY %s
    '''
    return query_set.annotate(val=RawSQL(raw_sql, (table_name, target_column_name,)))

每当我调用此函数时，Django 都会弹出以下错误。

django.db.utils.ProgrammingError: (1064, "You have an error in your SQL syntax
blah blah blah...

然后我检查了我的查询字符串：

print(_select_latest_rows(model_class, target_column_name, query_set).query)

然后它返回：

SELECT T.`id`, T.`user_id`, T.`ctime`, T.`comment`, (SELECT * FROM my_table GROUP BY user_id) AS `val` FROM my_table

好吧，我不得不说我对 SQL 语法不太熟悉，到目前为止，我还没有弄清楚这个查询字符串中错误的位置:(

我必须按照this answer 中的建议在模型的管理器实例中调用raw() 函数吗？

model_class.objects.raw('SELECT T.* FROM my_table AS T GROUP BY `user_id` DESC LIMIT 10 OFFSET 0')

或者我不应该在开头使用 GROUP BY 语句？

Answer 1

Oracle/PostgreSQL：

尝试使用下面的SQL，它会满足你的要求。

select id,user_id,ctime,comment
from  (select   id,user_id,ctime,comment
                rank() over(partition by user_id order by ctime desc) rn
        from your_table_name
      ) s
where rn = 1;

Answer 2

最简单的方法是：

select * 
from my_table t
where ctime = (select max(ctime) from my_table where user_id = t.user_id);

但是，如果user_id有多个相同的日期，您也可以在内部查询中使用limit子句

select * 
from my_table t
where id = (select id 
            from my_table 
            where  user_id = t.user_id
            order by ctime desc 
            LIMIT 1);