对比库表结构，生成SQL

网上找了一圈对比库的工具，能手工生成差别的SQL的工具没有，改造了一下网上的sql

1, 获取字段名的类型

create FUNCTION [dbo].[getColType](@tab varchar(100),@col varchar(100))
Returns nvarchar(100)
BEGIN
	Declare @lx VARCHAR(100)
	Declare @Length INT
    Declare @Scale int
	Declare @code nvarchar(32)

SELECT 

    @lx       = b.name,
    @Length       = COLUMNPROPERTY(a.id,a.name,'PRECISION'),
    @Scale   = isnull(COLUMNPROPERTY(a.id,a.name,'Scale'),0)

FROM
    syscolumns a
left join
    systypes b
on
    a.xusertype=b.xusertype
inner join
    sysobjects d
on
    a.id=d.id  and d.xtype='U' and  d.name<>'dtproperties'
left join
    syscomments e
on
    a.cdefault=e.id
left join
sys.extended_properties   g
on
    a.id=G.major_id and a.colid=g.minor_id
left join 

sys.extended_properties f
on
    d.id=f.major_id and f.minor_id=0
where
    d.name=@tab    --如果只查询指定表,加上此条件
	AND a.name=@col
 SET @code=@lx
 IF @lx LIKE 'decimal'
  SET   @code =@lx+'('+ CAST(@Length AS VARCHAR(10))+','+CAST(@Scale AS VARCHAR(10)) +')'
 IF  @lx LIKE '%varchar'
 BEGIN
   IF @Length=-1
    SET   @code =@lx+'(max)'
   ELSE
     SET   @code =@lx+'('+ CAST(@Length AS varchar(10))+')'
 end
Return @code
end

　2 对比生成SQL，这个存储过程特定情况的，网友要用，需要手工改造一下

create proc
 [dbo].[p_comparestructure]
@dbname1
varchar(250),--要比较的数据库名1
@dbname2
varchar(250)
--要比较的数据库名2
as
create table #tb1(表名1
varchar(250),字段名
varchar(250),序号
int,标识
bit,主键
bit,类型
varchar(250),
占用字节数
int,长度
int,小数位数
int,允许空
bit,默认值
 sql_variant,字段说明 sql_variant) 

create table #tb2(表名2
varchar(250),字段名
varchar(250),序号
int,标识
bit,主键
bit,类型
varchar(250),
占用字节数
int,长度
int,小数位数
int,允许空
bit,默认值
 sql_variant,字段说明 sql_variant) 

--得到数据库1的结构
exec('insert
 into #tb1 SELECT
表名=d.name,字段名=a.name,序号=a.colid,
标识=case
 when a.status=0x80 then 1 else 0 end,
主键=case
 when exists(SELECT 1 FROM '+@dbname1+'..sysobjects
 where xtype=''PK''
 and parent_obj=a.id and name in (
SELECT
 name FROM '+@dbname1+'..sysindexes
 WHERE indid in(
SELECT
 indid FROM '+@dbname1+'..sysindexkeys
 WHERE id = a.id AND colid=a.colid
)))
 then 1 else 0 end,
类型=b.name,占用字节数=a.length,长度=a.prec,小数位数=a.scale,允许空=a.isnullable,
默认值=isnull(e.text,''''),字段说明=isnull(g.[value],'''')
FROM
 '+@dbname1+'..syscolumns
 a
left
 join '+@dbname1+'..systypes
 b on a.xtype=b.xusertype
inner
 join '+@dbname1+'..sysobjects
 d on a.id=d.id  and d.xtype=''U''
 and  d.name <>''dtproperties''
left
 join '+@dbname1+'..syscomments
 e on a.cdefault=e.id
left
 join sys.extended_properties g
ON
a.ID=g.major_id
 AND a.COLID=g.minor_id
order
 by a.id,a.colorder')

--得到数据库2的结构
exec('insert
 into #tb2 SELECT
表名=d.name,字段名=a.name,序号=a.colid,
标识=case
 when a.status=0x80 then 1 else 0 end,
主键=case
 when exists(SELECT 1 FROM '+@dbname2+'..sysobjects
 where xtype=''PK''
 and parent_obj=a.id and name in (
SELECT
 name FROM '+@dbname2+'..sysindexes
 WHERE indid in(
SELECT
 indid FROM '+@dbname2+'..sysindexkeys
 WHERE id = a.id AND colid=a.colid
)))
 then 1 else 0 end,
类型=b.name,占用字节数=a.length,长度=a.prec,小数位数=a.scale,允许空=a.isnullable,
默认值=isnull(e.text,''''),字段说明=isnull(g.[value],'''')
FROM
 '+@dbname2+'..syscolumns a left
 join '+@dbname2+'..systypes
 b on a.xtype=b.xusertype inner
 join '+@dbname2+'..sysobjects
 d on a.id=d.id  and d.xtype=''U''
 and  d.name <>''dtproperties''left
 join '+@dbname2+'..syscomments
 e on a.cdefault=e.id left
 join sys.extended_properties g
ON
a.ID=g.major_id
 AND a.COLID=g.minor_id
order
 by a.id,a.colorder')
--and not exists (select 1 from #tb2 where 表名2=a.表名1)
SELECT * INTO #diff  FROM (SELECT 比较结果,表名1,字段名1, CAST(默认值1 AS VARCHAR(300)) 默认值11 FROM (
select 比较结果=case when a.表名1
is null and b.序号=1
then '线上库缺少表：'+b.表名2
when b.表名2
is null and a.序号=1
then '线下库缺少表:'+a.表名1
when a.字段名
is null and exists(select 1
from #tb1
where 表名1=b.表名2)
then '线上库 ['+b.表名2+'] 缺少字段：'+b.字段名
when b.字段名
is null and exists(select 1
from #tb2
where 表名2=a.表名1)
then '线下库缺少字段：'+a.表名1+':'+a.字段名
when a.标识
 <>b.标识 then '标识不同'
when a.主键
 <>b.主键 then '主键设置不同'
when a.类型
 <>b.类型 then '字段类型不同'
when a.占用字节数
 <>b.占用字节数 then '占用字节数'
when a.长度
 <>b.长度 then '长度不同'
when a.小数位数
 <>b.小数位数 then '小数位数不同'
when a.允许空
 <>b.允许空 then '是否允许空不同'
when a.默认值
 <>b.默认值 then '默认值不同'
when a.字段说明
 <>b.字段说明 then '字段说明不同'
else '' end,
 a.表名1, a.字段名 字段名1,b.表名2, b.字段名 字段名2,a.默认值  默认值1,b.默认值 默认值2
from #tb1
 a
full join #tb2
 b on a.表名1=b.表名2
and a.字段名=b.字段名
where a.表名1
is null or a.字段名
is null or b.表名2
is null or b.字段名
is null
or a.标识
 <>b.标识 or a.主键
 <>b.主键 or a.类型
 <>b.类型
or a.占用字节数
 <>b.占用字节数 or a.长度
 <>b.长度 or a.小数位数
 <>b.小数位数
or a.允许空
 <>b.允许空 or a.默认值
 <>b.默认值 or a.字段说明
 <>b.字段说明
 ) t1 WHERE (t1.比较结果 LIKE ('%线下库%') OR  t1.比较结果 LIKE ('%默认值%') ) AND t1.比较结果 NOT like  '%Sequence_20151207' ) t;

 SELECT * FROM  #diff
DECLARE @diff    VARCHAR(300)
DECLARE @tabname   VARCHAR(300)
DECLARE @colname   VARCHAR(300)
DECLARE @defaultvalue  VARCHAR(300)

DECLARE cursors CURSOR
FOR   SELECT * FROM #diff  --查询集合  

OPEN cursors
FETCH NEXT FROM cursors INTO @diff, @tabname,@colname,@defaultvalue
WHILE @@fetch_status = 0
BEGIN
    --遍历集合
    IF @diff LIKE '%线下库缺少表%'
	BEGIN
     PRINT ' select * into  '+@tabname+' from [b_Online]..'+@tabname
	 PRINT 'GO'
	END
    IF @diff LIKE '%线下库缺少字段%'
	BEGIN
     PRINT ' alter table '+@tabname+' add  '+@colname+' '+dbo.getColType(@tabname,@colname)
	 PRINT 'GO'
	END
    IF @diff LIKE '%默认值不同%'
	BEGIN
	IF @defaultvalue<>''
	BEGIN
	 PRINT ' alter table   ['  + @tabname + ']  ADD  DEFAULT '  + @defaultvalue + '  FOR  ['  + @colname + ']'
	 PRINT 'GO'
	END
	END 

   --EXEC(@sql)
   FETCH NEXT FROM cursors INTO  @diff, @tabname,@colname,@defaultvalue
END
 CLOSE cursors
 deallocate cursors

　执行：  exec p_comparestructure 'b_Online','b_Offline'   生成的SQL在Print 信息里。