亚洲精品成a人在线观看,亚洲男人天堂2018av,国产成人亚洲综合一区

left join如何去除重复数据

Thu, 03 Nov 2011 02:52:00 GMT

--大概思�\�Q�如果是从主表一直left join 子表 ,distinct业务层次最底层的子表ID

--wz_xqjhd 物资需求计划单

--wz_zbbd 物资招标标段

--WZ_ZBBDQB 招标标段�{�报

--wz_yzbfa 物资预中标方�?/span>

--wz_qbzb 物资�{�报主表

--wz_zbjg 物资中标�l�果

--jhtm 计划条目

--jhtmfp 计划条目分配

--以上表关�p?�Q�wz_xqjhd(1:n)wz_zbbd wz_xqjhd(1:n) wz_zbjg

-- wz_yzbfa(1:n)wz_zbbd wz_qbzb(1:n)WZ_ZBBDQB wz_zbbd(1:n)WZ_ZBBDQB WZ_ZBBDQB相当于中间表

-- wz_xqjhd(1:n) wz_zbjg

-- jhtmfp 为wz_jhtm与wz_zbbd的中间表

-- wz_jhtm 条目一�?相当于是业务的最低层 �q�个时候可以distinct去过滤最底层数据

select distinct(jhtm.jhtm_id) ,xqjhd.*,zbbd.*,jhtm.* from wz_xqjhd xqjhd

left join wz_zbbd zbbd on zbbd.xqjhd_id = xqjhd.xqjhd_id

left join WZ_ZBBDQB zbbdqb on zbbdqb.zbbd_id=zbbd.zbbd_id

left join wz_yzbfa yzbfa on yzbfa.yzbfa_id=zbbd.yzbfa_id

left join wz_qbzb qbzb on qbzb.qb_id=zbbdqb.dbqb_id

left join wz_zbjg zbjg on zbjg.Xqjhd_Id=xqjhd.xqjhd_id

left join wz_jhtm jhtm on jhtm.xqjhd_id=xqjhd.xqjhd_id

left join wz_jhtmfp jhtmfp on jhtmfp.jhtm_id=jhtm.jhtm_id and jhtmfp.zbbd_id=zbbd.zbbd_id;

以上是从主表兌��子表一层一层来的，下面是从子表开始：

select distinct(jhtm.jhtm_id) ,xqjhd.*,zbbd.*,jhtm.* from wz_jhtm jhtm

left join wz_jhtmfp jhtmfp on jhtmfp.jhtm_id=jhtm.jhtm_id

left join wz_zbbd zbbd on jhtmfp.zbbd_id = zbbd.zbbd_id

left join WZ_ZBBDQB zbbdqb on zbbdqb.zbbd_id=zbbd.zbbd_id

left join wz_yzbfa yzbfa on yzbfa.yzbfa_id=zbbd.yzbfa_id

left join wz_qbzb qbzb on qbzb.qb_id=zbbdqb.dbqb_id

left join wz_xqjhd xqjhd on xqjhd.xqjhd_id=jhtm.xqjhd_id

left join wz_zbjg zbjg on zbjg.Xqjhd_Id=xqjhd.xqjhd_id

从主表开始left join�q�是子表开始，具体得看业务逻辑�Q�参考的具体层次

孤飞�?/a> 2011-11-03 10:52 发表评论

ORACLE导出导入

Wed, 02 Nov 2011 11:42:00 GMT

一. 创徏表�ƈ插入数据

SQL> create table t1 as select * from scott.dept;

Table created.

SQL> select * from t1;

    DEPTNO DNAME          LOC
---------- -------------- -------------
        10 ACCOUNTING     NEW YORK
        20 RESEARCH       DALLAS
        30 SALES          CHICAGO
        40 OPERATIONS     BOSTON

�? 导出

SQL> exp -- 找不到该命��o
SP2-0042: unknown command "exp" - rest of line ignored.
SQL> host exp -- 在exp命��o前加上host

提示输入
username: scott/tiger@orcl --如果是管理员则加�?as sysdba

Enter array fetch buffer size: 4096 > 回�R

Export file: Expdat.dmp > e:\testexp.dmp -- 导出文�g存放位置

(2)U(sers), or (3)(ables): (2)U > 回�R -- 导出方式, �q�里只列出用��h��式和表方�? 如果是以��理员��n�? 则有三种方式, ��x��个数据库.

Export grants (yes/no): yes > 回�R --是否导出权限

Export table data (yes/no): yes >回�R -- 是否导出表中数据

compress extents (yes/no): yes >回�R -- 是否用压�~�区

... �q�样��把scott的所有对象导��Z��.

�? 导入

SQL> imp -- 找不到该命��o
SP2-0042: unknown command "imp" - rest of line ignored.
SQL> host imp -- 在imp命��o前加上host

提示输入
username: scott/tiger@orcl --如果是管理员则加�?as sysdba

Import file: EXPDAT.DMP > e:\testexp.dmp -- 导入文�g存放位置

Enter insert buffer size (minimum is 8192) 30720) >回�R

List contents of import file only(yes/no): no >回�R

ignore create error due to object existence (yes/no): no > yes -- 如果对象已经存在, 则忽�?br />
import grants (yes/no): yes >回�R -- 是否导入权限

import table data (yes/no): yes >回�R -- 是否导入表中数据

import entire export file (yes/no): no >no

username: scott -- 从哪个用户导�?

Enter table(T) or partition(T:P) names. Null list means all tales for user

Enter table(T) or partition(T:P) name or . if done: t1 -- T 表示普通表, T:P 表示表空�?

孤飞�?/a> 2011-11-02 19:42 发表评论

Oracle drop database

Wed, 02 Nov 2011 07:24:00 GMT

�?0g以前�Q�要��d��删除数据库，只有两个�Ҏ��Q�一个是利用DBCA囑�Ş化工具删除数据库�Q�另外一个就是关闭数据库后，手工删除数据文�g、控制文件以及日志文件的�Ҏ��?nbsp;

�?0g开始，Oracle提供了DROP DATABASE的语法，使得数据库的删除变得非常的简单�?nbsp;

不过DROP DATABASE�q�是有一定的限制条�g的：

SQL> conn / as sysdba
Connected.
SQL> drop database;
drop database
*
ERROR at line 1:
ORA-01586: database must be mounted EXCLUSIVE and not open for this operation

SQL> alter database close;

Database altered.

SQL> drop database;
drop database
*
ERROR at line 1:
ORA-12719: operation requires database is in RESTRICTED mode

SQL> alter system enable restricted session;

System altered.

SQL> drop database;

Database dropped.

Disconnected from Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 - 64bit Production
With the Partitioning, OLAP and Data Mining options
SQL> exit
[oracle@bjtest ~]$ ps -ef|grep test102
oracle 15805 14913 0 06:13 pts/1 00:00:00 grep test102

数据库必��d��于MOUNT状态，且设�|�了RESTRICTED SESSION。执行了DROP DATABASE命��o后，Oracle自动删除控制文�g�Q�已�l�控制文件中记录的数据文件和在线重做日志文�g�Q�如果数据库使用了SPFILE�Q�那么SPFILE文�g也会删除。这个命令不会删除归��文件和备䆾文�g�?nbsp;

在alert文�g中，可以看到drop database对应的操作：

drop database
Mon May 18 06:13:01 2009
Deleted file /data/oradata/test102/system01.dbf
Deleted file /data/oradata/test102/undotbs1.dbf
Deleted file /data/oradata/test102/sysaux01.dbf
Deleted file /data/oradata/test102/redo01.log
Deleted file /data/oradata/test102/redo02.log
Deleted file /data/oradata/test102/redo03.log
Deleted file /data/oradata/test102/redo04.log
Deleted file /data/oradata/test102/redo05.log
Deleted file /data/oradata/test102/redo06.log
Deleted file /data/oradata/test102/redo07.log
Deleted file /data/oradata/test102/temp01.dbf
Instance terminated by USER, pid = 15795
Deleted file /data/oradata/test102/control01.ctl
Completed: drop database
Mon May 18 06:13:08 2009
Shutting down instance (abort)
License high water mark = 1

孤飞�?/a> 2011-11-02 15:24 发表评论

Oracle转义字符

Wed, 26 Oct 2011 14:18:00 GMT

1、oracle �Ҏ��字符转义
关键词： oracle 转义

环境�Q�oracle 9i plsql

在plsql里边执行:
update userinfo set pageurl='myjsp?page=1&pagesize=10' where id='test'
�q�条sql语句往数据库的pageurl字段放进��M��一个url地址�Q�但是执行的时候却�q��那么理想�Q�因��其中有一个oracle的特�D�字�W�，需要进行�{义，那就是字�W?&'.

怎么处理上例中的�Ҏ��字符�Q?br />两个办法�Q?br /> 1) update userinfo set pageurl='myjsp?page=1'||'&'||'pagesize=10' where id='test'
2) update userinfo set pageurl='myjsp?page=1'||chr(38)||'pagesize=10' where id='test'

其中||是连字符, chr(38)跟ASCII的字�W��{码是一致的�?br />plsql中还可以set define off来关闭特�D�字�W�，�q�可以用show define来查看有些特�D�定义的字符�?/div>

2、oracle 中如何�{�?�Ҏ��字符
�?: 如何转义下划�U?_

select * from ng_values where name like 'lady_%'

jieguo �l�果昄�� lady_test,lady_test,lady1

正确�l�果应该�?lady_test,lady_test

不包�?lady1

请各位给��{义方�?3ks

answer:

select ... from ... where ... like '/_%' escape '/';

3�?br />insert into t(col) values(chr(ascii('&'))) �Q?/div>

(�Ҏ��一)
�?插入�Ҏ��字元'&'
SQL> SHOW DEFINE
define "&" (hex 26) ? <--- DEFINE的default值是 ‘&’
SQL> SET DEFINE OFF
SQL> SHOW DEFINE
define OFF
SQL> INSERT INTO VALUES ('AT&T');
/
1 row created

(�Ҏ��?
SQL> SHOW ESCAPE
escape OFF <--- ESCAPE的default值是 OFF
SQL> SET ESCAPE ON
SQL> SHOW ESCAPE
escape "\" (hex 5c)
SQL> INSERT INTO temp_table VALUES ('select * from emp where ename = \&1');

1 row created.

�Q?br />几个��试�Ҏ��Q?br />SELECT 'myjsp?page=1&pagesize=10' FROM dual;
SELECT 'myjsp?page=1&pagesize=10' FROM dual;

SQL> create table a (b varchar2(10));

Table created

SQL> insert into a values('_a');

1 row inserted

SQL> select B from A where instr(b,'_a')>0;

B
----------
_a
select B from A where b like '%\_a%' escape '\'

先启动�{义符
set escape on
select B from A where b like '%\_a%' escape '\';

select B from A where b like '%\_a%' escape '\'

孤飞�?/a> 2011-10-26 22:18 发表评论

关于navicat�q�接oracle �?ORA-12737 set CHS16GBK错误的解�?

Fri, 01 Jul 2011 02:54:00 GMT

1、原因：
ORA-12737: Instant Client Light: unsupported server character set CHS16GBK
Google�Q�解释如下：
ORA-12737:
Instant Client Light: unsupported server character set string
Cause: The character set specified is not allowed for this operation or is invalid. Instant Client Light has only minimal character sets.
Action: Do not use Instant Client Light for this character set

2、首先，我们打开“tools”-->"options"菜单�Q�见到如下界面，依据OCI library(oci.dll) 路径�Q�导航到 navicat oci 目录下，备䆾里面的文�Ӟ��通过在该目录新徏bak文�g夹，��?#8220;%dir%/Navicat Lite\instantclient_10_2”路径下的所有dll文�g剪切到bak文�g夹）�?br />

3、在oracle 客户端安装目录bin/目录下，扑ֈ�如下几个.dll文�g
4、将上面所列文件放�|�到“%dir%/Navicat Lite\instantclient_10_2” 目录下，重新启动navcat卛_��?br />
5、oracle10的客��L��dll文�g替换地址(请复制下面地址)�Q?br />http://www.tkk7.com/Files/liudawei/关于navicat�q�接oracle%20�?20ORA-12737%20set%20CHS16GBK错误的解�?zip

孤飞�?/a> 2011-07-01 10:54 发表评论

Oracle创徏外键(删除外键)

Thu, 30 Jun 2011 11:42:00 GMT

--删除外键
ALTER TABLE WZ_JHTM DROP CONSTRAINT F_WZ_JHTM_XQJHD;

--创徏外键
ALTER TABLE WZ_JHTM add CONSTRAINT "F_WZ_JHTM_XQJHD" FOREIGN KEY ("XQJHD_ID")
REFERENCES "BID_ADMIN"."WZ_XQJHD" ("XQJHD_ID") ENABLE

孤飞�?/a> 2011-06-30 19:42 发表评论

Thu, 30 Jun 2011 11:17:00 GMT

CREATE OR REPLACE PROCEDURE p_del_table
--功能�Q��@环删除表数据�Q�避免��用大量回滚段
--设计�Q?br />--�~�码�Q?br />--日期�Q?006�q?6�?7�?br />(
iTabName       VARCHAR2        --输入要删除的表名
)
AS
vSql          VARCHAR2(2000); --动态SQL
vTabName      VARCHAR2(100);
BEGIN
--vTabName:=TRIM(iTabName);
vTabName:='vtestjhtm';
LOOP
--循环删除数据�Q�每�ơ删�?000条然后commit
vSql:='DELETE FROM '||vTabName||' WHERE ROWNUM<1000';
--执行动态SQL
EXECUTE IMMEDIATE vSql;
COMMIT;
--删除的表数据为空�Ӟ��退�?br />EXIT WHEN SQL%ROWCOUNT=0;
END LOOP;
--提示删除成功
dbms_output.put_line('DELETE TABLE '||vTabName||' Successfully!');
EXCEPTION
WHEN OTHERS THEN
ROLLBACK;
--提示删除��p�|�Q��ƈ提示错误信息
dbms_output.put_line('DELETE TABLE '||vTabName||' Fail!');
dbms_output.put_line('Error Message is: '||substr(SQLERRM,1,200));
END;

孤飞�?/a> 2011-06-30 19:17 发表评论

Mon, 27 Jun 2011 07:29:00 GMT

Select r.table_name table_name, p.table_name ref_table_name
From User_Constraints r, User_Cons_Columns p
Where r.constraint_type = 'R'
And r.r_constraint_name = p.constraint_name
And p.table_name = upper('&ref_talbe')

孤飞�?/a> 2011-06-27 15:29 发表评论

Oracle数据SQL整理

Wed, 01 Jun 2011 06:28:00 GMT

一.查看表结�?br />USER_TAB_COLUMNS视图中，如scott模式下查看dept表结构：
20:52:08 SQL> SELECT column_name, data_type FROM user_tab_columns WHERE table_name='DEPT';

�?Oracle查询是否解锁与锁�?br />1.查询

      SELECT /*+ rule */
s.username,
decode(l.type, 'TM', 'TABLE LOCK', 'TX', 'ROW LOCK', l.type) LOCK_LEVEL,
o.owner,
o.object_name,
o.object_type,
s.sid,
s.serial#,
s.terminal,
s.machine,
s.program,
s.osuser
FROM v$session s, v$lock l, dba_objects o
WHERE l.sid = s.sid
   AND l.id1 = o.object_id(+)
   AND s.username is NOT NULL

2.解锁(�Ҏ��需要kill 550为的sid 3761为对应的serial#)
    alter system kill session '550,3761';
    commit;

�?增加修改字段字段注释表注�?br />alter table 表名 add   字段 char(1);
alter table 表名 modify 字段 char(1);
comment on column 表名.字段�?is '状�?;
commnet on table 表名 is '表名'

孤飞�?/a> 2011-06-01 14:28 发表评论

left join �?left outer join 的区�?

Mon, 30 May 2011 01:05:00 GMT

摘要: left join �?left outer join 的区�? 通俗的讲�Q?nbsp; A left join B 的连接的记录��C��A表的记录数同 A ... 阅读全文

孤飞�?/a> 2011-05-30 09:05 发表评论

数据库递归查询

Mon, 09 May 2011 08:28:00 GMT

DB2的递归查询

在db2可以使用sql语句来进行递归查询,��是使用with语句

1.先徏一个树形表:

create table tst (
id integer,
parentId int,
name varchar(20))

2.插入数据

insert into tst values
(1,0,'a'),
(2,0,'b'),
(3,1,'c'),
(4,1,'d'),
(5,4,'d'),
(6,5,'d')

3.使用递归查询

with rpl (id,parentId,name) as
(
select id,parentId,name from tst where parentId=1
union all
select child.id,child.parentId,child.name from rpl parent, tst child where parent.id=child.parentId
)
select * from rpl

Oracle的递归查询

攉��的几条在oracle中通过connect by prior来实现递归查询

Start with...Connect By子句递归查询一般用于一个表�l�护树�Ş�l�构的应用�?br /> 创徏�C�Z��表：
CREATE TABLE TBL_TEST
(
ID NUMBER,
NAME VARCHAR2(100 BYTE),
PID NUMBER DEFAULT 0
);

插入��试数据�Q?br /> INSERT INTO TBL_TEST(ID,NAME,PID) VALUES('1','10','0');
INSERT INTO TBL_TEST(ID,NAME,PID) VALUES('2','11','1');
INSERT INTO TBL_TEST(ID,NAME,PID) VALUES('3','20','0');
INSERT INTO TBL_TEST(ID,NAME,PID) VALUES('4','12','1');
INSERT INTO TBL_TEST(ID,NAME,PID) VALUES('5','121','2');

从Root往树末梢递归
select * from TBL_TEST
start with id=1
connect by prior id = pid

从末梢往树ROOT递归
select * from TBL_TEST
start with id=5
connect by prior pid = id

===============================================================================================================

有一张表   t
字段�Q?nbsp;
parent
child
两个字段的关�p�L��父子关系

写一个sql语句�Q�查询出指定父下面的所有的�?nbsp;

比如

a   b
a   c
a   e
b   b1
b   b2
c   c1
e   e1
e   e3
d   d1

指定parent=a�Q�选出
a   b
a   c
a   e
b   b1
b   b2
c   c1
e   e1
e   e3

SQL语句�Q?nbsp;
select   parent,child   from   test   start   with   parent='a'
connect   by   prior   child=parent

================================================================================================

connect by 是结构化查询中用到的�Q�其基本语法是：
select ... from tablename start by cond1
connect by cond2
where cond3;
��单说来是��一个树状结构存储在一张表里，比如一个表中存在两个字�D?
id,parentid那么通过表示每一条记录的parent是谁�Q�就可以形成一个树状结构�?
用上�q�语法的查询可以取得�q�棵树的所有记录�?
其中COND1是根�l�点的限定语句，当然可以攑֮�限定条�g�Q�以取得多个根结点，实际��是多棵树�?
COND2是连接条�Ӟ��其中用PRIOR表示上一条记录，比如 CONNECT BY PRIOR ID=PRAENTID��是说上一条记录的ID是本条记录的PRAENTID�Q�即本记录的父亲是上一条记录�?
COND3是过滤条�Ӟ��用于对返回的所有记录进行过滤�?/font>

PRIOR和START WITH关键字是可选项
PRIORY�q�算�W�必��L��|�在�q�接关系的两列中某一个的前面。对于节炚w��的父子关�p�，PRIOR
�q�算�W�在一侧表�C�父节点�Q�在另一侧表�C�子节点�Q�从而确定查找树�l�构是的��序是自��向下还�?br /> 自底向上。在�q�接关系中，除了可以使用列名外，�q�允�怋�用列表达式�?strong>START WITH 子句�?br /> 可选项�Q�用来标识哪个节点作为查找树型结构的根节炏V��若该子句被省略�Q�则表示所有满��x��?br /> 条�g的行作�ؓ根节炏V�?br /> 完整的例子如SELECT PID,ID,NAME FROM T_WF_ENG_WFKIND START WITH PID =0 CONNECT BY PRIOR ID = PID

以上主要是针对上层对下层的顺向递归查询而��用start with ... connect by prior ...�q�种方式,但有时在需求需要的时�?可能会需要由下层向上层的逆向递归查询,此是语句��有所变化:例如要实�?select * from table where id in ('0','01','0101','0203','0304') ;现在��x��0304的上一�U?3�l�递归出来,0203的上一�U?2�l�递归出来,�?1现在已经是存在的,最高层�?.而这张table不仅仅这些数�?但我现在只需�?'0','01','0101','0203','0304','02','03')�q�些数据,此时语句可以�q�样写SELECT PID,ID,NAME FROM V_WF_WFKIND_TREE WHERE ID IN (SELECT DISTINCT(ID) ID FROM V_WF_WFKIND_TREE CONNECT BY PRIOR PID = ID START WITH ID IN ('0','01','0101','0203','0304') );

其中START WITH ID IN里面的��g��可以替换SELECT 子查询语�?

注意�׃��层向下层递归与下层向上层递归的区别在于START WITH...CONNECT BY PRIOR...的先后顺序以�?ID = PID �?PID = ID 的微��变�?

=============================================================
connect by prior start with �l�常会被用到一个表中存在递归关系的时候。比如我们经�怼��一个比较复杂的目录树存储到一个表中。或者将一些部门存储到一个表中，而这些部门互相有隶属关系。这个时候你��׃��用到connect by prior start with�?/font>

典型的��用方法就是：
select * from table connect by prior cur_id=parent_id start with cur_id=???
例如�Q?br /> a            b
1           0
2           1
3           1
4           2
5           3

如果��x��找a=2及其下面的所有数据，则：
select * from table connect by prior a=b start with a=2
a           b
2           1
4           2

�q�些只是基础�Q�皮毛。其实只要你灉|��的构造查询语句。可以得出意想不到的�l�果。比如生成树每一个�\径�?br /> 但是�q�些记录�l�成的树必须正常才可以。如果有互�ؓ父子的情况，��׃��出现循环错误�Q?/font>

select * from tb_cus_area_cde

--子取�?/strong>
select * from tb_cus_area_cde a
CONNECT BY PRIOR a.c_snr_area=a.c_area_cde START WITH a.c_area_cde='1040101'

--父取�?br /> select * from tb_cus_area_cde a
CONNECT BY PRIOR     a.c_area_cde=a.c_snr_area START WITH a.c_snr_area is null

注意�Q�在用这个函数的时候，statement的参数要�?ResultSet.TYPE_SCROLL_INSENSITIVE   而不能用 ResultSet.TYPE_SCROLL_SENSITIVE,在这里再把这两个之间的区别讲�Ԍ��

1.TYPE_FORWORD_ONLY,只可向前滚动�Q?nbsp;

2.TYPE_SCROLL_INSENSITIVE,双向滚动�Q�但不及时更斎ͼ��是如果数据库里的数据修改过�Q��ƈ不在ResultSet中反应出来�?nbsp;

3.TYPE_SCROLL_SENSITIVE�Q�双向滚动，�q�及时跟�t�数据库的更�?以便更改ResultSet中的数据

======================================
10g树�Ş查询�Ҏ��CONNECT_BY_ISCYCLE

     �?0g中Oracle提供了新的伪列：CONNECT_BY_ISCYCLE�Q�通过�q�个伪列�Q�可以判断是否在树�Ş查询的过�E�中构成了��@环，�q�个伪列只是在CONNECT BY NOCYCLE方式下有效�?/font>

　　�q�一��描�q�C��下解决问题的思�\�?/font>

　　CONNECT_BY_ISCYCLE的实现和前面两篇文章中CONNECT_BY_ROOT和CONNECT_BY_ISLEAF的实现完全不同�?/font>

　　因�ؓ要实现CONNECT_BY_ISCYCLE�Q�就必须先实现CONNECT BY NOCYCLE�Q�而在9i中是没有�Ҏ��实现�q�个功能的�?/font>

　　也就是说�Q�首先要实现自己的树形查询的功能�Q�而仅�q�第一点，��是一个异常困隄��问题�Q�何况后面还要实现NOCYCLE�Q�最后再加上一个ISCYCLE的判断�?/font>

　　所以�ȝ��来说�Q�这个功能的实现比前面两个功能要复杂得多。由于树形查询的LEVEL是不固定的，所以采用链接的方式实现�Q�基本上是不现实的。换句话��_��用纯SQL的方式来实现树�Ş查询的功能基本上不可行。而�ؓ了解册��个功能，只能通过PL/SQL配合SQL来实现�?/font>

　　仍然是首先构造一个例子：

SQL> CREATE TABLE T_TREE (ID NUMBER, FATHER_ID NUMBER, NAME VARCHAR2(30));
　　表已创徏�?
　　SQL> INSERT INTO T_TREE VALUES (1, 0, 'A');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (2, 1, 'BC');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (3, 1, 'DE');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (4, 1, 'FG');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (5, 2, 'HIJ');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (6, 4, 'KLM');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (7, 6, 'NOPQ');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (0, 0, 'ROOT');
　　已创�?1 行�?
　　SQL> INSERT INTO T_TREE VALUES (4, 7, 'FG');
　　已创�?1 行�?
　　SQL> COMMIT;
　　提交完成�?
　　SQL> SELECT * FROM T_TREE;
　　ID FATHER_ID NAME
　　---------- ---------- ------------------------------
　　1 0 A
　　2 1 BC
　　3 1 DE
　　4 1 FG
　　5 2 HIJ
　　6 4 KLM
　　7 6 NOPQ
　　0 0 ROOT
　　4 7 FG
　　已选择9行�?

　　上面构造了两种树�Ş查询循环的情况，一�U�是当前记录的自循环�Q�另一�U�是树�Ş查询的某个子节点是当前节点的��先节点�Q�从而构成了循环。在�q�个例子中，记录ID�?和ID�?且FATHER_ID�{�于7的两条记录分别构成了上述的两�U��@环的情况�?/font>

　　下面��来看看CONNECT_BY_ISCYCLE和CONNECT BY NOCYCLE的功能：

SQL> SELECT *
　　2 FROM T_TREE
　　3 START WITH ID = 0
　　4 CONNECT BY PRIOR ID = FATHER_ID;
　　ERROR:
　　ORA-01436: 用户数据中的 CONNECT BY 循环
　　未选定�?
　　SQL> SELECT *
　　2 FROM T_TREE
　　3 START WITH ID = 1
　　4 CONNECT BY PRIOR ID = FATHER_ID;
　　ERROR:
　　ORA-01436: 用户数据中的 CONNECT BY 循环
　　未选定�?

　　�q�就是不使用CONNECT BY NOCYCLE的情况，查询会报错，指出树�Ş查询中出现��@环，�?0g中可以��用CONNECT BY NOCYCLE的方式来避免错误的��生：

SQL> SELECT *
　　2 FROM T_TREE
　　3 START WITH ID = 0
　　4 CONNECT BY NOCYCLE PRIOR ID = FATHER_ID;
　　ID FATHER_ID NAME
　　---------- ---------- ------------------------------
　　0 0 ROOT
　　1 0 A
　　2 1 BC
　　5 2 HIJ
　　3 1 DE
　　4 1 FG
　　6 4 KLM
　　7 6 NOPQ
　　已选择8行�?

　　使用CONNECT BY NOCYCLE�Q�Oracle自动避免循环的��生，��不产生循环的数据查询出来，下面看看CONNECT_BY_ISCYCLE的功能：

SQL> SELECT ID,
　　2 FATHER_ID,
　　3 NAME,
　　4 CONNECT_BY_ISCYCLE CYCLED
　　5 FROM T_TREE
　　6 START WITH ID = 0
　　7 CONNECT BY NOCYCLE PRIOR ID = FATHER_ID;
　　ID FATHER_ID NAME CYCLED
　　---------- ---------- ------------------------------ ----------
　　0 0 ROOT 1
　　1 0 A 0
　　2 1 BC 0
　　5 2 HIJ 0
　　3 1 DE 0
　　4 1 FG 0
　　6 4 KLM 0
　　7 6 NOPQ 1
　　已选择8行�?

　　可以看到�Q�CONNECT_BY_ISCYCLE伪列指出循环在树形查询中发生的位�|��?/font>

　　��Z��实现CONNECT_BY_ISCYCLE��必��d��实现CONNECT BY NOCYCLE方式�Q�而这�?i中是没有现成的办法的�Q�所以这里尝试��用PL/SQL来自己实现树形查询的功能�?/font>

SQL> CREATE OR REPLACE FUNCTION F_FIND_CHILD(P_VALUE VARCHAR2) RETURN VARCHAR2 AS
　　2 V_STR VARCHAR2(32767) := '/' || P_VALUE;
　　3
　　4 PROCEDURE P_GET_CHILD_STR (P_FATHER IN VARCHAR2, P_STR IN OUT VARCHAR2) AS
　　5 BEGIN
　　6 FOR I IN (SELECT ID FROM T_TREE WHERE FATHER_ID = P_FATHER AND FATHER_ID != ID) LOOP
　　7 IF INSTR(P_STR || '/', '/' || I.ID || '/') = 0 THEN
　　8 P_STR := P_STR || '/' || I.ID;
　　9 P_GET_CHILD_STR(I.ID, P_STR);
　　10 END IF;
　　11 END LOOP;
　　12 END;
　　13 BEGIN
　　14 P_GET_CHILD_STR(P_VALUE, V_STR);
　　15 RETURN V_STR;
　　16 END;
　　17 /

　　函数已创建�?/font>

　　构造一个函敎ͼ�在函��C��递归调用�q�程来实现树形查询的功能�?/font>

　　下面看看调用�q�个函数的结果：　

SQL> SELECT F_FIND_CHILD(0) FROM DUAL;
　　F_FIND_CHILD(0)
　　------------------------------------------------
　　/0/1/2/5/3/4/6/7
　　SQL> SELECT F_FIND_CHILD(2) FROM DUAL;
　　F_FIND_CHILD(2)
　　------------------------------------------------
　　/2/5
　　SQL> SELECT F_FIND_CHILD(4) FROM DUAL;
　　F_FIND_CHILD(4)
　　------------------------------------------------
　　/4/6/7

　　虽然目前存在的问题还有很多，但是已经基本上实��C��一个最��单的NOCYCLE的SYS_CONNECT_BY_PATH的功能�?/font>

　　有了�q�个函数作�ؓ基础�Q�就可以逐步的实现最�l�的目标了�?br />

孤飞�?/a> 2011-05-09 16:28 发表评论

唯一索引&普通烦引的一个有��的区别

Fri, 17 Dec 2010 08:07:00 GMT
今天CNOUG上，一个网友问了一个随机取记录的问题，�q�接如下�Q?

http://www.oracle.com.cn/viewthread.php?tid=130433

在测试过�E�，却引��h��对Oracle的烦引的机制有了一�ơ生动的回顾�?

��试的本意是参数一个随机数�Q�然后根据随机数取一条记录�?

SELECT * FROM phs.t1 WHERE rn = trunc(dbms_random.value(1,100000))

Rn是烦引字�D?normal betree�Q�根据rownum来徏立�?

但是初次��试的结果，却出乎我意料�Q�居然查询结果是�Q�无记录或者随机长度记录集。多�ơ测试：

SELECT COUNT(*) FROM phs.t1 WHERE rn = trunc(dbms_random.value(1,100000));

�l�果竟然是：0,28909�Q?,0,9870�Q?3012�Q?,56789�Q?5189�Q?240…

我花�?分钟��p��决了�q�个问题�Q�但�׃��三个��时的思考，我终于想了一套说法来描述�q�里面的关系�?

解决1�Q�把sql改�ؓ

WITH tab AS(SELECT trunc(dbms_random.value(1,100000)) a FROM dual)
SELECT * FROM t1 WHERE rn =( SELECT a FROM tab)

解决2�Q�把rn的烦引改为，unique

实际看到�W�二�U�解军_��法，��大概有�U�明白的感觉了，但是惌��释一下又无法说清。说白了�Q�这里面牉|��了Oracle Intenal的东西，�q�些东西又不是公开的。我只能凭着仅有的资料去猜测�?

�?/font>normal索引下，SQL的执行计划是RANGE SCAN�Q��?/font>unique索引�Q�则�?/font>unique SCAN�?/font>
发生�q�种现象实际�?/font>SQL的执行机制以及烦引的机制有着密切的联�p�R�?/font>
通过查询�Q�可以看��索引�?/font>2747个叶子块�Q?/font>67�?/font>branch 块，height 3
SQL> SELECT s.leaf_blocks,s.blevel FROM dba_indexes s WHERE s.index_name = 'INDEX_T1_RN';

LEAF_BLOCKS BLEVEL
----------- ----------
2747 2

SQL> SELECT SUM(blocks) FROM dba_extents WHERE segment_name='INDEX_T1_RN';

SUM(BLOCKS)
-----------
2816
那么它的�l�构如图�Q?/font>

'800')this.width='800';if(this.height>'600')this.height='600';">

SQL的执行步骤如下：
1�Q?/font>SQL匚w��Q�语法语义检�?/font>,通过�?/font>LIBRARY CACHE中对象的比对�Q�进行匹配�?/font>
2�Q�对子查询，视图�{�进行重新组合和SQL改写�Q�判断对象访问的开销以及�l�果集的大小,每个对象都独立计��成本以及返回的�l�果集的大小�Q�判断不同的�q�接��序的不同开销
�Q�连接方式和�q�接��序被通盘考虑�Q��ƈ且找到开销最��的�q�接方式�q�个步骤里面包含�?/font>SQL执行计划的优化。��生执行树�Q�执行树被生成后攑֜�LIBRARY CACHE里，�?/font>SQL执行的时候，被用来驱动查�?/font>.
4�Q�分配绑定变量需要的内存�I�间�Q�绑定变量的值实现绑定。��用上一步��生的执行计划执行SQL.
5�Q�对�?/font>SELECT操作�Q�比普通的SQL多了一�?/font>FETCH步骤�Q�在�q�个步骤中，实际上的DB BLOCK的访问才会��生。在�q�个阶段�Q�将剔除不需要的数据�Q�把�l�果攑օ��l�果集，传输�l�客��L��?/font>
有了上面的知识，我现在来解释�q�个现象�?/font>
Q:��Z��么在normal的烦引下�Q?/font>sql�q�回�l�构集的大小是随机的�Q�在unique下面却能��定唯一记录�Q?/font>
A:首先�?/font>normal索引下，优化器会把执行计划解释�ؓRANGE SCAN�Q�因为它认�ؓ可能�q�回多条记录。优化器在进行执行计划编写的时候，首先��׃��计算 trunc(dbms_random.value(1,100000))�Q�由此来��定选择什么执行计划，假设本次产生的随机数�?7905�?

�q�入执行阶段�Q�系�l�找�?7905“开�?#8220;所在的索引数据块，注意�q�个“开�?#8221;�Q�也��是说它只关注数据是从那里开始的�Q�，随后�q�行遍历该块的水�q�链表寻找，执行�q��o,扑ֈ�适合rowid�Q�再去获取数据块�?

在normal模式下，�׃��是范围SCAN�Q�优化器实际�?

Rn= trunc(dbms_random.value(1,100000))

拆成�?

Rn >= trunc(dbms_random.value(1,100000)) and

Rn <= trunc(dbms_random.value(1,100000))

在unique模式下，仍然�?

Rn= trunc(dbms_random.value(1,100000))

所以，在normal模式下，遍历�q�程发现�?7905的记录，但它�q�个时候它�q�会判断是否已经辑ֈ�区间扫描的终点，又会执行Rn <= trunc(dbms_random.value(1,100000))

�q�个时候，trunc(dbms_random.value(1,100000))被重新计��！�Q�！

那么��׃��出现下面的分支情况：

'800')this.width='800';if(this.height>'600')this.height='600';">

1.�l�果>87905,优化器认为已�l�达到区间终点（索引是有序的�Q�，谓词判断�l�果 true + false = false,所以返回空�l�果集�?

2.�l�果��于�{�于87905,�q�回该记录，计算出新的谓词例�?2346�Q�再�ơ寻扑֝��Q�遍历水�q�_��表（重复上面的动作）�Q�然后又�q�入分支判断…最�l�出现分�?的时候终止�?

�q�样��p��明�ؓ什么会出现�q�种情况�?

�?/font>unique模式下，在遍历水�q�链表的时候，扑ֈ�当前值就�q�回�Q�而不用进行区间判断，或者是SCAN多条。所以结果集是正��的�?/font>
Q:��Z��么采用了with语法可以避免�q�种情况呢？
A:因�ؓwith字句�?/font>oracle内部被解释�ؓ一个内联视图或者��时表�Q�所�?/font>trunc(dbms_random.value(1,100000))只会计算一�ơ，之后的计��是针对固定的内联视囄��?

oracle 不允许这�U�写法，rn= sequences.nextval�Q�估计也是出于这个考虑吧�?
惛_��q�里�Q�就�l�束了，其实可以自己写一个random函数�Q�在里面加一个记录点�Q�来证明谓词多次改变�Q�也可以做更详细的sql trace�Q�dump dump…

或者换一个Oracle的�h来解释一下，��OK了，但我估计在Oracle china也没几个了解数据库机密的人。唉…

引自�Q�http://valen.blog.ccidnet.com/blog-htm-do-showone-uid-51502-type-blog-itemid-262825.html

孤飞�?/a> 2010-12-17 16:07 发表评论

Oracle随机抽取(�?

Fri, 17 Dec 2010 08:06:00 GMT

�?00万数据量的表中随机抽取一条数�?要达�?.02S以内

刚好我有一�?20w数据的测试表�Q�在一个很老的PC上，northwood 2.4的p4�Q�sis芯片�l��?
复制内容到剪贴板
代码:
SQL> set timing on; SQL> SELECT COUNT(1) FROM t1;   COUNT(1) ---------- 1219948 Executed in 0.015 seconds SQL> alter table T1 add rn number; Table altered Executed in 0.36 seconds SQL> update t1   2  set rn =rownum; 1219948 rows updated Executed in 129.75 seconds SQL> create index index_t1_rn on T1 (rn); Index created Executed in 51.234 seconds SQL> SQL> WITH tab AS(SELECT trunc(dbms_random.value(1,100000)) a FROM dual)   2  SELECT * FROM t1 WHERE rn =( SELECT a FROM tab)   3  / NORMAL                                  CLASS       RN ---------- --------------------------------------- ----------    3198                                     5    7384 Executed in 0.031 seconds SQL> SQL> WITH tab AS(SELECT trunc(dbms_random.value(1,100000)) a FROM dual)   2  SELECT * FROM t1 WHERE rn =( SELECT a FROM tab)   3  / NORMAL                                  CLASS       RN ---------- --------------------------------------- ----------    4760                                     12    72082 Executed in 0.047 seconds SQL> / NORMAL                                  CLASS       RN ---------- --------------------------------------- ----------    6922                                     7    30862 Executed in 0.033 seconds SQL> / NORMAL                                  CLASS       RN ---------- --------------------------------------- ----------    1727                                     5    81038 Executed in 0.019 seconds SQL> / NORMAL                                  CLASS       RN ---------- --------------------------------------- ----------    11890                                     3    65704 Executed in 0.016 seconds
100w不是一个恐怖的数据量，创徏字段索引都比较快。这�c�问题关注的焦点�Q�就是在如何减少IO上，诸如get random value的问题，实际只消耗一�ơCPU旉��Q�而CPU都几百M HZ�?br /> 最后，我用了一个with 语法来取敎ͼ�其实是我在测试过�E�发��C��个现象，暂时不表�?br />
引自�Q�http://www.oracle.com.cn/viewthread.php?tid=130433&extra=page%3D1

孤飞�?/a> 2010-12-17 16:06 发表评论

Fri, 17 Dec 2010 07:30:00 GMT

首先�W�一个是随机抽取6�?

    select * from (select * from tablename order by order by dbms_random.value) where rownum�Q?

    �q�个�Ҏ��的原理我认�ؓ应该是把表中的数据全部查询出来按照随机数�q�行排列后在从查询出来的数据中查询中6条记录，�q�个�Ҏ��我在使用的过�E�中发现�Q�如果记录一多的话查询的速度有一点点的慢

�W�二个是利用oracle的sample()或sample block�Ҏ��

选择10%的记�?br /> select * from t1 sample(10)
选择0.1%的记�?br /> select * from t1 sample(0.1)

�Ҏ��数据块选择1%的记�?br /> select * from t1 sample block(1)

使用数据块选择与��用记录行选择的区别：使用数据块选择表示��h��的采集是��Z��数据块采集的�Q�也��是说样本如果一个数据块被采集�ؓ��h��Q�则数据块里的记录全部都是样�?/p>

��h��l�计是基于统计学采集的，是有概率问题�Q�不一定完全准��，如你要取50%的记录，但实际可能返回给�?9%的记录集�Q�也可能�q�回�l�你51%的记录集

例如

如果表T1有数据块B1,B2
B1有记录R1,R2,R3,R4,R5
B2有记录R6,R7,R8,R9,R10

如果使用如下SQL选择50%的数�?br /> select * from t1 sample block(50)

则返回的�l�果可能是数据块B1的记�?br /> R1,R2,R3,R4,R5
也可能是数据块B2的记�?br /> R6,R7,R8,R9,R10
也可能不�q�回记录�?/p>

如果使用如下SQL选择50%的数�?br /> select * from t1 sample (50)

则返回的�l�果可能�?br /> R2,R3,R5,R8,R9
也可能是如下的样�?br /> R1,R3,R4,R8

应用�C�Z��Q?br /> 随机从表中取�?条记录，选取记录的概率是1%
select * from t1 sample(1) where rownum=1

随机从表中取�?0条记录，选取记录的概率是0.1%
select * from t1 sample(0.1) where rownum<=10

注：当选取的概率越低，讉K��表的记录数将��多

ORACLE参考手册中的相兌��明：

sample_clause
The sample_clause lets you instruct Oracle to select from a random sample of rows from the table, rather than from the entire table.

BLOCK
BLOCK instructs Oracle to perform random block sampling instead of random row sampling.

sample_percent
sample_percent is a number specifying the percentage of the total row or block count to be included in the sample. The value must be in the range .000001 to (but not including) 100.

Restrictions on Sampling During Queries
You can specify SAMPLE only in a query that selects from a single table. Joins are not supported. However, you can achieve the same results by using a CREATE TABLE ... AS SELECT query to materialize a sample of an underlying table and then rewrite the original query to refer to the newly created table sample. If you wish, you can write additional queries to materialize samples for other tables.

When you specify SAMPLE, Oracle automatically uses cost-based optimization. Rule-based optimization is not supported with this clause.

--------------------------------------------------------------------------------
Caution:
The use of statistically incorrect assumptions when using this feature can lead to incorrect or undesirable results.

--------------------------------------------------------------------------------

�?
Sample选项
使用sample选项的意思是指定Oracle从表中随机选择记录��h��Q�这��h��从整个表中选择更高�?

block选项
加上 BLOCK选项时表�C�随机取数据块，而不是随机取记录�?

sample_percent选项
sample_percent是指定总记录行或数据块为数据样本的癑ֈ�比数��|��q�个值只能在0.000001�?00之间�Q�且不能�{�于100

限制
只能在单表查询的SQL中指定sample选项�Q�不支持有连接的查询。但是，你可以��用CREATE TABLE ... AS SELECT查询的语法完成同��L��效果�Q�然后再采用新徏的样本表重新�~�写查询SQL�?/p>
当你指定用sample�Ӟ��不支持基于规�?rule)的优化法则，ORACLE自动使用基本成本(cost)的优化法�?br />

孤飞�?/a> 2010-12-17 15:30 发表评论

Thu, 16 Dec 2010 06:02:00 GMT

1.查询


SELECT /*+ rule */
s.username,
decode(l.type, 'TM', 'TABLE LOCK', 'TX', 'ROW LOCK', l.type) LOCK_LEVEL,
o.owner,
o.object_name,
o.object_type,
s.sid,
s.serial#,
s.terminal,
s.machine,
s.program,
s.osuser
FROM v$session s, v$lock l, dba_objects o
WHERE l.sid = s.sid
   AND l.id1 = o.object_id(+)
   AND s.username is NOT NULL

2.解锁(�Ҏ��需要kill 550为的sid 3761为对应的serial#)
    alter system kill session '550,3761';
    commit;

孤飞�?/a> 2010-12-16 14:02 发表评论

Oracle下实��C��键自增长(触发�?

Mon, 29 Nov 2010 01:46:00 GMT

drop table book;

--创徏�?nbsp;

create table book(

   bookId varchar2(4) primary key,

   name varchar2(20)

);

--创徏序列

create sequence book_seq start with 1 increment by 1;



--创徏触发�?nbsp;

create or replace trigger book_trigger

before insert on book

for each row

begin

select book_seq.nextval into :new.bookId from dual;

end ;

--��d��数据

insert into book(name)  values ('cc');

insert into book(name)  values ('dd');



commit;

孤飞�?/a> 2010-11-29 09:46 发表评论

Oracle创徏表空�?用户

Tue, 02 Nov 2010 01:42:00 GMT
01_create_tablespace

----------------------------------------------------
-- Export file for user JSYD                      --
-- Created by Administrator on 2008-9-4, 10:34:43 --
----------------------------------------------------set echo on
--set feedback off
--set verify off

--define tb_path=C:\oracle\product\10.1.0\oradata\steel_trade\

--prompt '建立表空间开�?..'

create tablespace STEEL_TRADE_DAT datafile '&tb_path\STEEL_TRADE_DAT.DBF' size 10M AUTOEXTEND ON NEXT 10240K MAXSIZE UNLIMITED extent management local uniform size 128k online;
create tablespace STEEL_TRADE_IDX datafile '&tb_path\STEEL_TRADE_IDX.DBF' size 10M AUTOEXTEND ON NEXT 10240K MAXSIZE UNLIMITED extent management local uniform size 128k online;
create temporary tablespace STEEL_TRADE_TMP tempfile '&tb_path\STEEL_TRADE_TMP.DBF' size 10m AUTOEXTEND ON NEXT 10240K MAXSIZE UNLIMITED extent management local uniform size 128k;

--prompt '建立表空间完毕！'

02_create_user

----------------------------------------------------
-- Export file for user JSYD                      --
-- Created by Administrator on 2008-9-4, 10:34:43 --
----------------------------------------------------
--prompt '创徏数据库用�?..'

--set feedback off
--set define off
--set echo off
--PROMPT 正在创徏login ---jsyd
create user bid_admin identified by a123456
    default tablespace STEEL_TRADE_DAT
    temporary tablespace STEEL_TRADE_TMP
    account unlock;

--PROMPT 正在创徏角色权限 ---jsyd

--PROMPT 正在创徏�pȝ��权限 ---jsyd
grant AUDIT ANY to bid_admin with admin option;
grant DROP USER to bid_admin with admin option;
grant ALTER USER to bid_admin with admin option;
grant ANALYZE ANY to bid_admin with admin option;
grant BECOME USER to bid_admin with admin option;
grant CREATE ROLE to bid_admin with admin option;
grant CREATE TYPE to bid_admin with admin option;
grant CREATE USER to bid_admin with admin option;
grant CREATE VIEW to bid_admin with admin option;
grant ALTER SYSTEM to bid_admin with admin option;
grant AUDIT SYSTEM to bid_admin with admin option;
grant CREATE TABLE to bid_admin with admin option;
grant DROP PROFILE to bid_admin with admin option;
grant ALTER PROFILE to bid_admin with admin option;
grant ALTER SESSION to bid_admin with admin option;
grant DROP ANY ROLE to bid_admin with admin option;
grant DROP ANY TYPE to bid_admin with admin option;
grant DROP ANY VIEW to bid_admin with admin option;
grant QUERY REWRITE to bid_admin with admin option;
grant ALTER ANY ROLE to bid_admin with admin option;
grant ALTER ANY TYPE to bid_admin with admin option;
grant ALTER DATABASE to bid_admin with admin option;
grant CREATE CLUSTER to bid_admin with admin option;
grant CREATE LIBRARY to bid_admin with admin option;
grant CREATE PROFILE to bid_admin with admin option;
grant CREATE SESSION to bid_admin with admin option;
grant CREATE SYNONYM to bid_admin with admin option;
grant CREATE TRIGGER to bid_admin with admin option;
grant DROP ANY INDEX to bid_admin with admin option;
grant DROP ANY TABLE to bid_admin with admin option;
grant GRANT ANY ROLE to bid_admin with admin option;
grant LOCK ANY TABLE to bid_admin with admin option;
grant ALTER ANY INDEX to bid_admin with admin option;
grant ALTER ANY TABLE to bid_admin with admin option;
grant CREATE ANY TYPE to bid_admin with admin option;
grant CREATE ANY VIEW to bid_admin with admin option;
grant CREATE SEQUENCE to bid_admin with admin option;
grant CREATE SNAPSHOT to bid_admin with admin option;
grant DROP TABLESPACE to bid_admin with admin option;
grant ALTER TABLESPACE to bid_admin with admin option;
grant BACKUP ANY TABLE to bid_admin with admin option;
grant CREATE ANY INDEX to bid_admin with admin option;
grant CREATE ANY TABLE to bid_admin with admin option;
grant CREATE PROCEDURE to bid_admin with admin option;
grant DELETE ANY TABLE to bid_admin with admin option;
grant DROP ANY CLUSTER to bid_admin with admin option;
grant DROP ANY LIBRARY to bid_admin with admin option;
grant DROP ANY SYNONYM to bid_admin with admin option;
grant DROP ANY TRIGGER to bid_admin with admin option;
grant EXECUTE ANY TYPE to bid_admin with admin option;
grant INSERT ANY TABLE to bid_admin with admin option;
grant SELECT ANY TABLE to bid_admin with admin option;
grant UPDATE ANY TABLE to bid_admin with admin option;
grant ALTER ANY CLUSTER to bid_admin with admin option;
grant ALTER ANY LIBRARY to bid_admin with admin option;
grant ALTER ANY TRIGGER to bid_admin with admin option;
grant COMMENT ANY TABLE to bid_admin with admin option;
grant CREATE TABLESPACE to bid_admin with admin option;
grant DROP ANY SEQUENCE to bid_admin with admin option;
grant DROP ANY SNAPSHOT to bid_admin with admin option;
grant FORCE TRANSACTION to bid_admin with admin option;
grant MANAGE TABLESPACE to bid_admin with admin option;
grant ALTER ANY SEQUENCE to bid_admin with admin option;
grant ALTER ANY SNAPSHOT to bid_admin with admin option;
grant CREATE ANY CLUSTER to bid_admin with admin option;
grant CREATE ANY LIBRARY to bid_admin with admin option;
grant CREATE ANY SYNONYM to bid_admin with admin option;
grant CREATE ANY TRIGGER to bid_admin with admin option;
grant DROP ANY DIRECTORY to bid_admin with admin option;
grant DROP ANY PROCEDURE to bid_admin with admin option;
grant RESTRICTED SESSION to bid_admin with admin option;
grant ALTER ANY PROCEDURE to bid_admin with admin option;
grant ALTER RESOURCE COST to bid_admin with admin option;
grant CREATE ANY SEQUENCE to bid_admin with admin option;
grant CREATE ANY SNAPSHOT to bid_admin with admin option;
grant DROP PUBLIC SYNONYM to bid_admin with admin option;
grant EXECUTE ANY LIBRARY to bid_admin with admin option;
grant GRANT ANY PRIVILEGE to bid_admin with admin option;
grant SELECT ANY SEQUENCE to bid_admin with admin option;
grant CREATE ANY DIRECTORY to bid_admin with admin option;
grant CREATE ANY PROCEDURE to bid_admin with admin option;
grant CREATE DATABASE LINK to bid_admin with admin option;
grant GLOBAL QUERY REWRITE to bid_admin with admin option;
grant UNLIMITED TABLESPACE to bid_admin with admin option;
grant CREATE PUBLIC SYNONYM to bid_admin with admin option;
grant DROP ROLLBACK SEGMENT to bid_admin with admin option;
grant EXECUTE ANY PROCEDURE to bid_admin with admin option;
grant FORCE ANY TRANSACTION to bid_admin with admin option;
grant SELECT ANY DICTIONARY to bid_admin with admin option;
grant ALTER ROLLBACK SEGMENT to bid_admin with admin option;
grant CREATE ROLLBACK SEGMENT to bid_admin with admin option;
grant DROP PUBLIC DATABASE LINK to bid_admin with admin option;
grant CREATE PUBLIC DATABASE LINK to bid_admin with admin option;
--exit;

孤飞�?/a> 2010-11-02 09:42 发表评论

Oracle9i 字符集与NLS_LANG搭配��试��gؕ码问题分�?

Fri, 17 Sep 2010 08:04:00 GMT
【背景介�l��?/font>
===================================================================
最�q�在使用Oracle9i数据库进行数据插入、查询、导�?出，有时会出��Cؕ码的情况�Q�具体的情�Ş有以下两�U�：

1.首次插入/昄��q��

2.首次插入/昄��正常、但把数据用工具导出为本地文�?例如TXT)文�g�Q�再在另一个客��L��中打开该文件�ƈ执行时再�ơ插入的数据昄��Zؕ码�?

遂在本地创徏两个数据库，一个�ؓAL32UTF8字符集，一个�ؓZHS16GBK字符集，配合客户端NLS_LANG的不同设�|�，��试��q��的情况及�q�行原因分析�?nbsp;
===================================================================

【测试目的�?br style="font-family: " /> ===================================================================
��试不同的数据库字符集和客户端NLS_LANG搭配下，中文字符/中文日期的插入、显�C?br style="font-family: " /> ===================================================================

【测试环境�?br style="font-family: " /> ===================================================================
Windows 2000 Professional (英文�? + Oracle 9.2.0.1.0 + SQL*Plus: Release 9.2.0.1.0

注：此次��试的客��L��应用工具为SQL*PLUS�Q�如果是使用TOAD、PL/SQL Develop之类工具�Q�得到的�l�果会和以下有些不同。推荐��用SQL*PLUS作�ؓ一切客��L��应用的测试工�?br style="font-family: " /> ===================================================================

【测试数据库�?br style="font-family: " /> ===================================================================
本机数据�?: SID: PAULLIN 登陆参数: qprod/qprod@paullin
本机数据�?�Q�SID: PAUL    登陆参数�Q�qlinpen/pengpenglin@paul
===================================================================

【测试字�W�集�?br style="font-family: " /> ===================================================================
客户端应�?操作�pȝ��字符集：GB2312

客户端NLS_LANG讄��Q?br style="font-family: " /> AMERICAN_AMERICA.US7ASCII
  AMERICAN_AMERICA.WE8MSWIN1252
  AMERICAN_AMERICA.ZHS16GBK
  AMERICAN_AMERICA.AL32UTF8
  SIMPLIFIED CHINESE_CHINA.ZHS16GBK

数据库端字符集：
PAULLIN: AL32UF8
  PAUL: ZHS16GBK
===================================================================

【测试脚本�?br style="font-family: " /> ===================================================================
--登陆数据�?nbsp;
     sqlplus qprod/qprod@paullin
     sqlplus qlinpen/pengpenglin@paul

--执行��试脚本
     drop table test;
     drop table testdate;

     create table test (id number(1), name varchar2(20));
     create table testdate (birthday date);

     insert into test values(1,'Tom');
     insert into test values(2,'张三');
     insert into test values(3,'易徏�?);
     commit;

     insert into testdate values(TO_Date( '01/08/2008 04:14:00 下午',
           'MM/DD/YYYY HH:MI:SS AM'));
     commit;

--查看��试�l�果
     select * from test;
     select * from testdate;
===================================================================

【测试一�Q�数据库端字�W�集为AL32UTF8的情��c�?br style="font-family: " /> ===================================================================
1.登陆数据库：
  C:\Documents and Settings\qlinpen.E0015609D6309>sqlplus qprod/qprod@paullin

2.查看数据库字�W�集�Q?br style="font-family: " />   SQL> select * from nls_database_parameters where parameter = 'NLS_CHARACTERSET';

3.��试内容及测试结果：
  1).客户端NLS_LANG讄��为AMERICAN_AMERICA.US7ASCII�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ��q��
     中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  2).客户端NLS_LANG讄��为AMERICAN_AMERICA.WE8MSWIN1252�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ正常
     中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  3).客户端NLS_LANG讄��为AMERICAN_AMERICA.ZHS16GBK�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ正常
      中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  4).客户端NLS_LANG讄��为AMERICAN_AMERICA.AL32UTF8�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ正常
      中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  5).客户端NLS_LANG讄��为SIMPLIFIED CHINESE_CHINA.ZHS16GBK�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文均�ؓ正常
      中文日期��试�Q�插�?昄��中文日期均�ؓ正常�Q�格式�ؓDD-MM-YY(例如�Q?8-1�?-08)

4.��试�l�论�Q?br style="font-family: " />   从测试结果来看，当数据库端字�W�集为AL32UTF8�Ӟ��能够使到中文字符被正��插�?昄��的NLS_LANG的字�W�集为：  WE8MSWIN1252、ZHS16GBK、AL32UTF8�?br style="font-family: " />
  但是要��C��文格式的旉��能够被正��插入，则NLS_LANG的LANGUAGE和TERRITORY讄��必须为：SIMPLIFIED_CHINESE_CHINA
===================================================================

【测试二�Q�数据库端字�W�集为ZHS16GBK的情��c�?br style="font-family: " /> ===================================================================
1.登陆数据库：
  C:\Documents and Settings\qlinpen.E0015609D6309>sqlplus qlinpen/pengpenglin@paul

2.查看数据库字�W�集�Q?br style="font-family: " />   SQL> select * from nls_database_parameters where parameter = 'NLS_CHARACTERSET';

3.��试内容及测试结果：
  1).客户端NLS_LANG讄��为AMERICAN_AMERICA.US7ASCII�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ��q��
      中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  2).客户端NLS_LANG讄��为AMERICAN_AMERICA.WE8MSWIN1252�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ��q��
      中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  3).客户端NLS_LANG讄��为AMERICAN_AMERICA.ZHS16GBK�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ正常
      中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  4).客户端NLS_LANG讄��为AMERICAN_AMERICA.AL32UTF8�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文字符均�ؓ��q��
      中文日期��试�Q�ORA-01855: AM/A.M. or PM/P.M. required

  5).客户端NLS_LANG讄��为SIMPLIFIED CHINESE_CHINA.ZHS16GBK�Q?br style="font-family: " />       中文字符��试�Q�插�?昄��中文均�ؓ正常
      中文日期��试�Q�插�?昄��中文日期均�ؓ正常�Q�格式�ؓDD-MM-YY(例如�Q?8-1�?-08)

4.��试�l�论�Q?br style="font-family: " />   从测试结果来看，当数据库端字�W�集为ZHS16GBK�Ӟ��能够使到中文字符被正常插�?昄��的NLS_LANG的字�W�集为：ZHS16GBK�?br style="font-family: " />
  而且要��C��文格式的旉��能够被正��插入，则NLS_LANG的LANGUAGE和TERRITORY讄��必须为：SIMPLIFIED CHINESE_CHINA
===================================================================

【测试�ȝ��?br style="font-family: " /> ===================================================================
从上面两个测试的�l�果��可以明昄��出，把数据库端的字符集设�|��ؓAL32UTF8比�vZHS16GBK更加有优�ѝ�?br style="font-family: " />
UTF8支持从客��L��应用字符集�ؓWE8MSWIN1252、ZHS16GBK、AL32UTF8的环境下�q�行中文字符的插入，而ZHS16GBK只支持客��L��应用字符集�ؓZHS16GBK环境下的的中文字�W�插入�?br style="font-family: " />
其次我们来看看日期格式�ؓ�Q?1/08/2008 04:14:00 下午的记录�ؓ什么只能在NLS_LANG的LANGUAGE和TERRITORY为SIMPLIFIED CHINESE_CHINA的情况下才能正确插入�Q?br style="font-family: " />
我们知道客户端NLS_LANG的值由3部分构成�Q�即_.�Q�而掌��日期中月䆾和日昄��的恰恰就�?lt;LANGUAGE>部分�Q�所以很明显为AMERICAN的情况下�Q�是不可能正��插入的(西方用AM、PM来表�C�Z��、下�?�?br style="font-family: " />
依次�c�L��Q�如果以后出现货币和数字格式、地区和计算星期及日期的习惯插入、�{换失败，那么我们��p��查第二个元素�?br style="font-family: " /> ===================================================================

孤飞�?/a> 2010-09-17 16:04 发表评论

Fri, 17 Sep 2010 07:48:00 GMT
今天安装Oracle客户�?监听器已�l�配�|�好 �q�且在Net Configuration Assistant中测试都正常�Q�用PLSQL��试�q�接��M��成功�Q��L��报ora12541-TNS 异常信息

查了下相关东�?br />
本��n的环境变量：

d:\oracle\product\10.2.0\client_2\bin;d:\oracle\product\10.2.0\client_1;C:\Program Files\PC Connectivity Solution\;%SystemRoot%\system32;%SystemRoot%;%SystemRoot%\System32\Wbem;C:\Program Files\Common Files\Thunder Network\KanKan\Codecs;D:\Program Files\TortoiseSVN\bin

把红色去掉后 ��然成功�?希望大家以后��到此问�?��查下相关环境变量可能�Ҏ��每个人安装不�?环境变量不同只保留bin目录其他可以��L��?

孤飞�?/a> 2010-09-17 15:48 发表评论

亚洲精品成a人在线观看,亚洲男人天堂2018av,国产成人亚洲综合一区

left join如何去除重复数据

ORACLE导出导入

Oracle drop database

Oracle转义字符

关于navicat�q�接oracle �?ORA-12737 set CHS16GBK错误的解�?

Oracle创徏外键(删除外键)

Oracle数据SQL整理

left join �?left outer join 的区�?

数据库递归查询

DB2的递归查询

Oracle的递归查询

唯一索引&普通烦引的一个有���的区别

Oracle随机抽取(�?

代码:

Oracle下实��C��键自增长(触发�?

Oracle创徏表空�?用户

Oracle9i 字符集与NLS_LANG搭配���试��gؕ码问题分�?

唯一索引&普通烦引的一个有��的区别

Oracle9i 字符集与NLS_LANG搭配��试��gؕ码问题分�?