菜鸟学sql，Oracle数据库结构比较

转载自：http://www.cnblogs.com/tianqing/archive/2008/06/25/1229419.html

做oracle下数据库结构比较，比较具体详细差异，例如：2个库具体有那些表不同，不同在什么地方，主键、长度、类型......,要求是用一句sql实现，对于我这个sql菜鸟来说，的确是个挑战。好在有个SQL Server例子可以借鉴。先贴一下oracle下的sql

--------------------------------------------------------------------------------------
--oracle数据库帐套比较，
--在本例中，databaseA数据库，
          --databaseB是数据库。
--在实际的运行过程中，请依据需要修改本sql
--------------------------------------------------------------------------------------
select (case when a.表名1 is null and b.序号=1 then '库1缺少表：'||b.表名2
                     when b.表名2 is null and a.序号=1 then '库2缺少表:'||a.表名1
                     when a.字段名 is null and exists(select 1 from (select table_name as 表名1  from DBA_TAB_COLUMNS where upper(DBA_TAB_COLUMNS.owner) =upper('databaseB')) where 表名1=b.表名2) then '库1 ['||b.表名2||'] 缺少字段：'||b.字段名
                     when b.字段名 is null and exists(select 1 from (select table_name as 表名2  from DBA_TAB_COLUMNS where upper(DBA_TAB_COLUMNS.owner) =upper('databaseA')) where 表名2=a.表名1) then '库2 ['||a.表名1||'] 缺少字段：'||a.字段名
                     when a.主键<>b.主键 then '主键不同'
                     when a.类型<>b.类型 then '类型不同'
                     when a.字段长度<>b.字段长度 then '字段长度不同'
                     when a.精度<>b.精度 then '精度不同'
                     when a.小数位<>b.小数位 then '小数位不同'
                     when a.允许空<>b.允许空 then '允许空不同'
                else '' end) result,a.*,b.*
from (select a.table_name as 表名1, a.column_name as 字段名,b.constraint_type as 主键, a.data_type as 类型,column_id as 序号, a.data_length as 字段长度, a.data_precision as 精度,  a.data_scale as 小数位, a.nullable as 允许空
   from all_tab_columns a left join (select d.constraint_type,e.column_name,e.table_name from all_constraints d,all_cons_columns e
 where  d.owner  =  e.owner  and d.table_name  =  e.table_name and
  d.constraint_name  =  e.constraint_name and d.constraint_type  = 'P') b on a.COLUMN_NAME = b.column_name and a.TABLE_NAME = b.table_name
   where  a.owner  = upper('databaseB') order by 表名1 asc) a

full join (select a.table_name as 表名2, a.column_name as 字段名,b.constraint_type as 主键, a.data_type as 类型,column_id as 序号, a.data_length as 字段长度, a.data_precision as 精度,  a.data_scale as 小数位, a.nullable as 允许空
   from all_tab_columns a left join (select d.constraint_type,e.column_name,e.table_name from all_constraints d,all_cons_columns e
 where  d.owner  =  e.owner  and d.table_name  =  e.table_name and
  d.constraint_name  =  e.constraint_name and d.constraint_type  = 'P') b on a.COLUMN_NAME = b.column_name and a.TABLE_NAME = b.table_name
   where  a.owner  = upper('databaseA') order by 表名2 asc) 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.类型<>b.类型 or a.字段长度<>b.字段长度 or a.精度<>b.精度 or a.小数位<>b.小数位 or a.允许空<>b.允许空 or a.主键<>b.主键
order by nvl(a.表名1,b.表名2),nvl(a.序号,b.序号)

执行结果显示的差异信息比较详细。可以依据这个来修改数据库，比较方便。
现在分析下这句sq的语法。
1.将两个数据库oracle叫schema下的所有表结构select出来，即：

所有表结构
select a.table_name as 表名1, a.column_name as 字段名,b.constraint_type as 主键, a.data_type as 类型,column_id as 序号, a.data_length as 字段长度, a.data_precision as 精度,  a.data_scale as 小数位, a.nullable as 允许空
  from all_tab_columns a left join (select d.constraint_type,e.column_name,e.table_name from all_constraints d,all_cons_columns e
  where  d.owner  =  e.owner  and d.table_name  =  e.table_name and
   d.constraint_name  =  e.constraint_name and d.constraint_type  = 'P') b on a.COLUMN_NAME = b.column_name and a.TABLE_NAME = b.table_name
  where  a.owner  = upper('databaseB') order by 表名1 asc

通过上述sql就可以将某个数据库下的表结构select出来，通过left join 关联列的主键。
2. 将要比较的两个库的表结构用full join 关联起来。
3. 同在用case来比较具体不同信息。然后在加上where条件。

综述，比较oracle中，2个数据库的表结构基本就用上述sql比较了，效率有些慢，朋友们有好的建议可以谈谈。谢谢。