SQL Server初学者：如何从另一个表更新表数据？

Question

我还在学习SQL。我有关于更新表的问题。我如何根据其他表更新我的表？请看下面的例子：

create table batch_info (
    batch_key int identity(1, 1) not null primary key, 
    batch_num int not null, 
    batch_name varchar(50) null, 
    batch_start datetime null, 
    batch_end datetime null, 
    table_name varchar(50) null, 
    insert_rows int null 
) 
go 

insert into batch_info (batch_num, batch_name, batch_start) 
    values ('32', 'Batch_to_insert_rows', '20110414') 
go 

select * from batch_info 

create table ref_table (
    bat_num int not null, 
    bat_end_date datetime null, 
    bat_table_name varchar(50) null, 
    bat_ins_rows int null, 
) 
go 

insert into ref_table 
    values ('32','20110414 02:12:00.000','Table1','10'), 
      ('32','20110414 02:12:00.000','Table2','33'), 
      ('32','20110414 02:12:00.000','Table3','12') 

select * from ref_table 

--How can I update batch_info table to get this info? 
select 
bi.batch_key, 
bi.batch_num, 
bi.batch_name, 
bi.batch_start, 
rt.bat_end_date, 
rt.bat_table_name, 
rt.bat_ins_rows 
from batch_info as bi 
inner join ref_table as rt on bi.batch_num = rt.bat_num 

编辑：一批重点是代孕所以它应该是渐进的，而不是（1，1，1）在我的连接查询。

Answer 1

假设batch_num和bat_end_date的组合是唯一的，那么这应该做的伎俩：

update batch_info 
set  batch_end = rt.bat_end_date, 
     table_name = rt.bat_table_name, 
     insert_rows = rt.bat_ins_rows 
from batch_info bi 
     inner join ref_table rt on rt.bat_num = bi.batch_num 
     inner join 
     (
      select bat_num, 
        MIN(bat_end_date) as min_bat_end_date 
      from ref_table 
     ) oldest on oldest.bat_num = rt.bat_num 
        and oldest.min_bat_end_date = rt.bat_end_date 
where bi.batch_end is null 

insert into batch_info 
(batch_num, batch_name, batch_start, batch_end, table_name, insert_rows) 
select bi.batch_num, 
     bi.batch_name, 
     bi.batch_start, 
     rt.bat_end_date, 
     rt.bat_table_name, 
     rt.bat_ins_rows 
from batch_info bi 
     inner join ref_table rt on rt.bat_num = bi.batch_num 
where not exists 
     (
      select * 
      from  batch_info e 
      where e.batch_num = bi.batch_num 
        and e.batch_end = rt.bat_end_date 
     ) 

如果值的另一组合建立的独特性，则这些可以在第一个查询添加到oldest，和e在第二。

我同意@Jim，这应该被分成两个表。执行此操作的难度是此设计将产生的困难的预兆。您试图将两种不同的东西（批次和批次运行）放到一张表中，这绝不是一个好主意。

Answer 2

insert into batch_info 
SELECT batch_num, batch_name, batch_start, 
bat_end, bat_table_name, bat_ins_rows 
FROM batch_info inner join ref_table 
on batch_info.batch_num = ref_table.bat_num 

delete from batch_info 
where batch_end is null 

编辑：INSERT正在生成的所有匹配行&重新插入。新的身份值将被添加到这些。现有行（其他字段为空）用于INSERT语句&，后面使用第二个语句删除。

Answer 3

我打算告诉你Merge Statement，但我不认为这可以做到这一点。我想你会被卡住，使用Cursor。

有了这个说法并不知道你现有的要求，它真的应该是两个表。把所有东西都放到一张桌子里，就是使桌子设计非规范化。

Answer 4

使batch_info.batch_key字段自动增量。

create table batch_info (
batch_key int identity(1, 1) not null primary key, -- <<- make autoincrement. 
...