Question

我有一个带有性能数据表的mysql数据库。每个数据表条目都用于特定的计数器。

DATA
----
dat_date     BIGINT
value     BIGINT
dat_date  DATETIME(3)
FK_dt_id  BIGINT

对于每个条目，我需要计算速度和加速度。所以我想创建一个SELECT，它将选择一行PREVIOUS行。 FK_dt_id是计数器类型

的外键

我试过这个：

SELECT          d1.dat_date,
            d1.value v1,
            d1.PK_dat_id
    FROM data d1
    INNER JOIN (SELECT * FROM data d2 ORDER BY d2.dat_date) d2
    ON (d2.dat_date < d1.dat_date
            AND d2.FK_dt_id = d1.FK_dt_id)

它有效，但我无法获得以前的主键，只有它的价值。有什么想法吗？

Answer 1

使用用户变量的最简单（不是必要的最简单）方法（使用“纯粹的”MySQL语法）是一个小技巧：

select @pk_data_id as prev_pk_data_id
     , @pk_data_id := (case 
           when @fk_counter_id = fk_counter_id then a.pk_data_id 
           else 0
       end) as pk_data_id
     , @fk_counter_id := a.fk_counter_id as fk_counter_id
from
    (select @pk_data_id := 0, @fk_counter_id := 0) as init
    , data as a
order by a.fk_counter_id, a.pk_data_id

获得此行集后，您可以将其与data表格结合使用。

我将这个结果放在一个临时表中，稍后再使用它;像这样的东西：

drop table if exists temp_tbl;
create temporary table temp_tbl
    select @pk_data_id as prev_pk_data_id
         , @pk_data_id := (case 
               when @fk_counter_id = fk_counter_id then a.pk_data_id 
               else 0
           end) as pk_data_id
         , @fk_counter_id := a.fk_counter_id as fk_counter_id
    from
        (select @pk_data_id := 0, @fk_counter_id := 0) as init
        , data as a
    order by a.fk_counter_id, a.pk_data_id;
alter table temp_tbl
    add index dId (pk_data_id),
    add index pdId (prev_pk_data_id),
    add index cId (fk_counter_id);
-- Now use the temp table to get what you need
select d1.*
     , d2.pk_data_id as prev_pk_data_id
     , d2.data_value as prev_data_value
     , d2.data_datetime as prev_data_datetime
from data as d1
     -- If you don't use the temp table, substitute 'temp_tbl' with
     -- the query from above
     inner join temp_tbl as a 
         on d1.pk_data_id = a.pk_data_id 
            and d1.fk_counter_id = a.fk_counter_id
     left join data as d2
         on a.prev_pk_data_id = d2.pk_data_id
            and a.fk_counter_id = d2.fk_counter_id

按日期时间选择以前的数据输入

1 个答案: