HKEY_LOCAL_MACHINE下的SOFTWARE下的ORACLE下的HOME0的NLS_LANG,值为:SIMPLIFIED CHINESE_CHINA.ZHS16GBK。
修改系统字符集:
--linux
export NLS_LANG=AMERICAN_AMERICAN.ZHS16CGB231280
--ORACLE
alter session set NLS_LANG='SIMPLEFIED CHINESE';
--windows
set nls_lang=AMERICAN_AMERICAN.ZHS16GBK
下面是转自DBA精髓的:
作者:eygle |English 【转载时请以超链接形式标明文章出处和作者信息及本声明】
链接:http://www.eygle.com/archives/2004/09/nls_character_set_04.html
导入导出是我们常用的一个数据迁移及转化工具,因其导出文件具有平台无关性,所以在跨平台迁移中,最为常用。
在导出操作时,非常重要的是客户端的字符集设置,也就是客户端的NLS_LANG设置。
NLS_LANG参数由以下部分组成:
NLS_LANG=
|
NLS_LANG各部分含义如下: LANGUAGE指定: -Oracle消息使用的语言 -日期中月份和日显示 TERRITORY指定 -货币和数字格式 -地区和计算星期及日期的习惯 CHARACTERSET: -控制客户端应用程序使用的字符集 通常设置或者等于客户端(如Windows)代码页 或者对于unicode应用设置为UTF8 在Windows上查看当前系统的代码页可以使用chcp命令:
E:\>chcp 活动的代码页: 936 |
代码页936也就是中文字符集 GBK,在Microsoft的官方站点上,我们可以遭到关于936代码页的具体编码规则,请参考以下链接:
http://www.microsoft.com/globaldev/reference/dbcs/936.htm
我们看一个简单的测试,来了解一下这几个参数的作用:
E:\>set NLS_LANG=SIMPLIFIED CHINESE_CHINA.ZHS16GBK E:\>sqlplus "/ as sysdba" SQL*Plus: Release 9.2.0.4.0 - Production on 星期六 11月 1 22:51:59 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. 连接到: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select sysdate from dual; SYSDATE ---------- 01-11月-03 已选择 1 行。 SQL> exit 从Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production中断开 E:\>set NLS_LANG=AMERICAN_AMERICA.ZHS16GBK E:\>sqlplus "/ as sysdba" SQL*Plus: Release 9.2.0.4.0 - Production on Sat Nov 1 22:52:24 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select sysdate from dual; SYSDATE --------- 01-NOV-03 1 row selected. SQL> |
查看客户端NLS_LANG设置可以使用以下方法:
Windows使用: echo %NLS_LANG% 如: E:\>echo %NLS_LANG% AMERICAN_AMERICA.ZHS16GBK Unix使用: env|grep NLS_LANG 如: /opt/oracle>env|grep NLS_LANG NLS_LANG=AMERICAN_CHINA.ZHS16GBK Windows客户端设置,可以在注册表中更改NLS_LANG,具体键值位于: HKEY_LOCAL_MACHINE\SOFTWARE\ORACLE\HOMExx\ xx指存在多个ORACLE_HOME时系统编号。 |
导入和导出是客户端产品,同SQL*PLUS和Oralce Forms一样,因此,使用EXP/IMP工具将按照NLS_LANG定义的方式转换字符集。
导出使用的字符集将会记录在导出文件中,当文件导入时,将会检查导出时使用的字符集设置,如果这个字符集不同于导入客户端的NLS_LANG
设置,字符集将根据导入客户端NLS_LANG设置进行转换,如果必要,在数据插入数据库之前会进行进一步转换。
通常在导出时最好把客户端字符集设置得和数据库端相同,这样可以避免在导出时发生不必要的数据转换,导出文件将和数据库具有相同的字符集。
即使将来会把导出文件导入到不同字符集的数据库中,这样做也可以把转换延缓至导入时刻。
当进行数据导入时,主要存在以下两种情况:
1.源数据库和目标数据库具有相同字符集设置
这时,只需要设置NLS_LANG等于数据库字符集即可导入(前提是,导出使用的是和源数据库相同字符集,即三者相同)
2.源数据库和目标数据库字符集不同
如果我们导出时候使用的NLS_LANG是和源数据库相同的字符集,那么导入时就可以设置客户端NLS_LANG等于导出时使用的字符集,这
样转换只发生在数据库端,而且只发生一次。
例如:
如果进行从WE8MSWIN1252到UTF8的转换
1)使用NLS_LANG=AMERICAN_AMERICA.WE8MSWIN1252导出数据库。
这时创建的导出文件包含WE8MSWIN1252的数据
2)导入时使用NLS_LANG=AMERICAN_AMERICA.WE8MSWIN1252
这时转换仅发生在insert数据到UTF8的数据库中。
以上假设的转换只在目标数据库字符集是源数据库字符集的超集时才能转换。如果不同,一般就需要进行一些特殊的处理。
我们简单看一下导入的转换过程(以Oracle8i为例):
1.确定导出数据库字符集环境
通过读取导出文件头,可以获得导出文件的字符集设置
2.确定导入session的字符集,即导入Session使用的NLS_LANG环境变量
3.IMP读取导出文件
读取导出文件字符集ID,和导入进程的NLS_LANG进行比较
4.如果导出文件字符集和导入Session字符集相同,那么在这一步骤内就不需要转换
如果不同,就需要把数据转换为导入Session使用的字符集。
然而这种转换只能在单byte字符集之间进行。
我们看一个测试:
E:\nls2>set NLS_LANG=AMERICAN_AMERICA.US7ASCII 设置导入session NLS_LANG为US7ASCII E:\nls2>e:\oracle\ora8i\bin\imp eygle/eygle file=Sus7ascii-Cus7ascii-exp817.dmp fromuser=eygle touser=eygle tables=test 这个导出文件是从US7ASCII数据库导出,导出客户端NLS_LANG也是US7ASCII Import: Release 8.1.7.1.1 - Production on Fri Nov 7 00:59:22 2003 (c) Copyright 2000 Oracle Corporation. All rights reserved. Connected to: Oracle8i Enterprise Edition Release 8.1.7.1.1 - Production With the Partitioning option JServer Release 8.1.7.1.1 - Production 这时导入,在DMP文件和NLS_LANG之间不需要进行字符集转换。 Export file created by EXPORT:V08.01.07 via conventional path import done in US7ASCII character set and ZHS16GBK NCHAR character set import server uses ZHS16GBK character set (possible charset conversion) export server uses UTF8 NCHAR character set (possible ncharset conversion) . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. |
5.对于多Byte字符集的导入(如:UTF8)
需要设置导入Session字符集和导出字符集相同
否则就会遇到:IMP-16 "Required character set conversion (type %lu to %lu) not supported" 错误。
:
E:\nls2>set NLS_LANG=AMERICAN_AMERICA.ZHS16GBK 导入Session字符集设置为ZHS16GBK 导入US7ASCII的导出文件 E:\nls2>e:\oracle\ora8i\bin\imp eygle/eygle file=Sus7ascii-Cus7ascii-exp817.dmp fromuser=eygle touser=eygle Import: Release 8.1.7.1.1 - Production on Fri Nov 7 00:38:55 2003 (c) Copyright 2000 Oracle Corporation. All rights reserved. Connected to: Oracle8i Enterprise Edition Release 8.1.7.1.1 - Production With the Partitioning option JServer Release 8.1.7.1.1 - Production IMP-00016: required character set conversion (type 1 to 852) not supported IMP-00000: Import terminated unsuccessfully 在从导出文件US7ASCII到导入 NLS_LANG设置为ZHS16GBK的过程中,不支持单Byte字符集向多Byte转换,报出以上错误。 |
6.导入Session字符集应该是导出字符集的超级,否则,专有的字符将难以正确转换。
7.当数据转换为导入Session字符集设置以后,如果导入Session字符集不同于导入数据库字符集,这时还需要最后一步转换,这要求导入数据库字符
集是导入session字符集的超级,否则某些专有字符将不能正常转换。
我们继续看上面的两个过程,这里有这样两个原则:
1.如果NLS_LANG的设置和数据库相同,那么数据(在传输过程中当然是2进制码)不经过转换就直接插入数据库中。
2.如果NLS_LANG的设置和数据库不同,那么数据需要转换后才能插入数据库中。
我们再回头来看上面的第一个例子:
:
Export file created by EXPORT:V08.01.07 via conventional path import done in US7ASCII character set and ZHS16GBK NCHAR character set import server uses ZHS16GBK character set (possible charset conversion) export server uses UTF8 NCHAR character set (possible ncharset conversion) . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. 这时候经过第一步转换后的数据,US7ASCII到ZHS16GBK丢失首位,原样插入数据库,我们看到这时数据库中存放的就是错误的字符(在后面 部分我们做了详细的转换): E:\nls2>sqlplus eygle/eygle SQL*Plus: Release 9.2.0.4.0 - Production on Fri Nov 7 00:35:39 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle8i Enterprise Edition Release 8.1.7.1.1 - Production With the Partitioning option JServer Release 8.1.7.1.1 - Production SQL> select * from test; NAME -------------------- 2bJT test |
在Oracle9i中,以上情况略有不同。
原文发表于itpub技术丛书《Oracle数据库DBA专题技术精粹》,未经许可,严禁转载本文.
我们知道在导出文件中,记录着导出使用的字符集id,通过查看导出文件头的第2、3个字节,我们可以找到16进制表示的字符集ID,在Windows上,
我们可以使用UltraEdit等工具打开dmp文件,查看其导出字符集::
在Unix上我们可以通过以下命令来查看:
cat expdat.dmp | od -x | head
|
Oracle提供标准函数,对字符集名称及ID进行转换:
SQL> select nls_charset_id('ZHS16GBK') from dual; NLS_CHARSET_ID('ZHS16GBK') -------------------------- 852 1 row selected. SQL> select nls_charset_name(852) from dual; NLS_CHAR -------- ZHS16GBK 1 row selected. 十进制转换十六进制: SQL> select to_char('852','xxxx') from dual; TO_CH ----- 354 1 row selected. |
对应上面的图中第2、3字节,我们知道该导出文件字符集为ZHS16GBk.
查询数据库中有效的字符集可以使用以下脚本:
col nls_charset_id for 9999 col nls_charset_name for a30 col hex_id for a20 select nls_charset_id(value) nls_charset_id, value nls_charset_name, to_char(nls_charset_id(value),'xxxx') hex_id from v$nls_valid_values where parameter = 'CHARACTERSET' order by nls_charset_id(value) / |
输出样例如下:
NLS_CHARSET_ID NLS_CHARSET_NAME HEX_ID -------------- ------------------------------ ------------- 1 US7ASCII 1 2 WE8DEC 2 3 WE8HP 3 4 US8PC437 4 5 WE8EBCDIC37 5 6 WE8EBCDIC500 6 7 WE8EBCDIC1140 7 8 WE8EBCDIC285 8 ................... 850 ZHS16CGB231280 352 851 ZHS16MACCGB231280 353 852 ZHS16GBK 354 853 ZHS16DBCS 355 860 ZHT32EUC 35c 861 ZHT32SOPS 35d 862 ZHT16DBT 35e 863 ZHT32TRIS 35f 864 ZHT16DBCS 360 865 ZHT16BIG5 361 866 ZHT16CCDC 362 867 ZHT16MSWIN950 363 868 ZHT16HKSCS 364 870 AL24UTFFSS 366 871 UTF8 367 872 UTFE 368 .................................. |
在很多时候,当我们进行导入操作的时候,已经离开了源数据库,这时如果目标数据库的字符集和导出文件不一致,很多时候就需要进行特殊处理,
以下介绍几种方法,主要以US7ASCII和ZHS16GBK为例
1. 源数据库字符集为US7ASCII,导出文件字符集为US7ASCII或ZHS16GBK,目标数据库字符集为ZHS16GBK
在Oracle92中,我们发现对于这种情况,不论怎样处理,这个导出文件都无法正确导入到Oracle9i数据库中,这可能是因为Oracle9i的编码方案发生了较大改变。
以下是我们所做的简单测试,其中导出文件命名规则为:
S-Server ,后跟Server字符集
C-client , 后跟导出操作时客户端字符集
导入时客户端字符集设置在命令行完成,限于篇幅,我们省略了部分测试过程。
对于Oracle9iR2,我们的测试结果是US7ASCII字符集,不管怎样转换,都无法正确导入ZHS16GBK字符集的数据库中。
在进行导入操作时,如果字符不能正常转换,Oracle数据库会自动用一个”?”代替,也就是编码63。
E:\nls2>set NLS_LANG=AMERICAN_AMERICA.US7ASCII E:\nls2>imp eygle/eygle file=Sus7ascii-Cus7ascii.dmp fromuser=eygle touser=eygle tables=test Import: Release 9.2.0.4.0 - Production on Mon Nov 3 17:14:39 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production Export file created by EXPORT:V09.02.00 via conventional path import done in US7ASCII character set and AL16UTF16 NCHAR character set import server uses ZHS16GBK character set (possible charset conversion) . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. E:\nls2>sqlplus eygle/eygle SQL*Plus: Release 9.2.0.4.0 - Production on Mon Nov 3 17:14:50 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select name,dump(name) from test; NAME DUMP(NAME) ----------------------------- ???? Typ=1 Len=4: 63,63,63,63 test Typ=1 Len=4: 116,101,115,116 2 rows selected. SQL> exit Disconnected from Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production E:\nls2>set NLS_LANG=AMERICAN_AMERICA.ZHS16GBK E:\nls2>imp eygle/eygle file=Sus7ascii-Cus7ascii.dmp fromuser=eygle touser=eygle tables=test ignore=y Import: Release 9.2.0.4.0 - Production on Mon Nov 3 17:15:28 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production Export file created by EXPORT:V09.02.00 via conventional path import done in ZHS16GBK character set and AL16UTF16 NCHAR character set export client uses US7ASCII character set (possible charset conversion) . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. E:\nls2>sqlplus eygle/eygle SQL*Plus: Release 9.2.0.4.0 - Production on Mon Nov 3 17:15:34 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select name,dump(name) from test; NAME DUMP(NAME) -------------------------------------------------------------------------------- ???? Typ=1 Len=4: 63,63,63,63 test Typ=1 Len=4: 116,101,115,116 ???? Typ=1 Len=4: 63,63,63,63 test Typ=1 Len=4: 116,101,115,116 4 rows selected. SQL> drop table test; Table dropped. SQL> exit Disconnected from Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production E:\nls2>set NLS_LANG=AMERICAN_AMERICA.ZHS16GBK E:\nls2>imp eygle/eygle file=Sus7ascii-Czhs16gbk.dmp fromuser=eygle touser=eygle tables=test ignore=y Import: Release 9.2.0.4.0 - Production on Mon Nov 3 17:17:21 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production Export file created by EXPORT:V09.02.00 via conventional path import done in ZHS16GBK character set and AL16UTF16 NCHAR character set . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. E:\nls2>sqlplus eygle/eygle SQL*Plus: Release 9.2.0.4.0 - Production on Mon Nov 3 17:17:30 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select name,dump(name) from test; NAME DUMP(NAME) ---------------------------------------------- ???? Typ=1 Len=4: 63,63,63,63 test Typ=1 Len=4: 116,101,115,116 2 rows selected. SQL> exit Disconnected from Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production E:\nls2>set NLS_LANG=AMERICAN_AMERICA.US7ASCII E:\nls2>imp eygle/eygle file=Sus7ascii-Czhs16gbk.dmp fromuser=eygle touser=eygle tables=test ignore=y Import: Release 9.2.0.4.0 - Production on Mon Nov 3 17:18:00 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production Export file created by EXPORT:V09.02.00 via conventional path import done in US7ASCII character set and AL16UTF16 NCHAR character set import server uses ZHS16GBK character set (possible charset conversion) export client uses ZHS16GBK character set (possible charset conversion) . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. E:\nls2>sqlplus eygle/eygle SQL*Plus: Release 9.2.0.4.0 - Production on Mon Nov 3 17:18:08 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select name,dump(name) from test; NAME DUMP(NAME) ---------------------------------------- ???? Typ=1 Len=4: 63,63,63,63 test Typ=1 Len=4: 116,101,115,116 ???? Typ=1 Len=4: 63,63,63,63 test Typ=1 Len=4: 116,101,115,116 4 rows selected. SQL> |
对于这种情况,我们可以通过使用Oracle8i的导出工具,设置导出字符集为US7ASCII,导出后修改第二、三字符,修改 0001 为
0354,这样就可以将US7ASCII字符集的数据正确导入到ZHS16GBK的数据库中。
修改导出文件:
导入修改后的导出文件:
E:\nls2>set NLS_LANG=AMERICAN_AMERICA.ZHS16GBK E:\nls2>imp eygle/eygle file=Sus7ascii-Cus7ascii-exp817.dmp fromuser=eygle touser=eygle tables=test Import: Release 9.2.0.4.0 - Production on Mon Nov 3 17:37:17 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production Export file created by EXPORT:V08.01.07 via conventional path import done in ZHS16GBK character set and AL16UTF16 NCHAR character set export server uses UTF8 NCHAR character set (possible ncharset conversion) . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. E:\nls2>sqlplus eygle/eygle SQL*Plus: Release 9.2.0.4.0 - Production on Mon Nov 3 17:37:23 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select name,dump(name) from test; NAME DUMP(NAME) -------------------------------------------------------------------------------- 测试 Typ=1 Len=4: 178,226,202,212 Test Typ=1 Len=4: 116,101,115,116 2 rows selected. SQL> |
2. 使用create database的方法
如果导出文件使用的字符集是US7ASCII,目标数据库的字符集是ZHS16GBK,我们可以使用create database的方法来修改,具体如下:
SQL> col parameter for a30 SQL> col value for a30 SQL> select * from v$nls_parameters; PARAMETER VALUE ------------------------------ ------------------------------ NLS_LANGUAGE AMERICAN NLS_TERRITORY AMERICA NLS_CURRENCY $ NLS_ISO_CURRENCY AMERICA NLS_NUMERIC_CHARACTERS ., NLS_CALENDAR GREGORIAN NLS_DATE_FORMAT DD-MON-RR NLS_DATE_LANGUAGE AMERICAN NLS_CHARACTERSET ZHS16GBK NLS_SORT BINARY ………………. 19 rows selected. SQL> create database character set us7ascii; create database character set us7ascii * ERROR at line 1: ORA-01031: insufficient privileges SQL> select * from v$nls_parameters; PARAMETER VALUE ------------------------------ ------------------------------ NLS_LANGUAGE AMERICAN NLS_TERRITORY AMERICA NLS_CURRENCY $ NLS_ISO_CURRENCY AMERICA NLS_NUMERIC_CHARACTERS ., NLS_CALENDAR GREGORIAN NLS_DATE_FORMAT DD-MON-RR NLS_DATE_LANGUAGE AMERICAN NLS_CHARACTERSET US7ASCII NLS_SORT BINARY ………….. 19 rows selected. SQL> exit Disconnected from Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production E:\nls2>set nls_lang=AMERICAN_AMERICA.US7ASCII E:\nls2>imp eygle/eygle file=Sus7ascii-Cus7ascii.dmp fromuser=eygle touser=eygle Import: Release 9.2.0.4.0 - Production on Sun Nov 2 14:53:26 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production Export file created by EXPORT:V09.02.00 via conventional path import done in US7ASCII character set and AL16UTF16 NCHAR character set import server uses ZHS16GBK character set (possible charset conversion) . . importing table "TEST" 2 rows imported Import terminated successfully without warnings. E:\nls2>sqlplus eygle/eygle SQL*Plus: Release 9.2.0.4.0 - Production on Sun Nov 2 14:53:35 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select * from test; NAME ---------- 测试 test 2 rows selected. |
我们看到,当发出create database character set us7ascii;命令时,数据库v$nls_parameters中的字符集设置随之更改,该参数影响导入进程,
更改后可以正确导入数据,重起数据库后,该设置恢复。
提示:v$nls_paraemters来源于x$nls_parameters,该动态性能视图影响导入操作;而nls_database_parameters来源于props$数据表,影响数据存储。
3. Oracle提供的字符扫描工具csscan
我们说以上的方法只是应该在不得已的情况下使用,其本质是欺骗数据库,强制导入数据,可能损失元数据。
如果要确保数据的完整性,应该使用csscan扫描数据库,找出所有不兼容的字符,然后通过编写相应的脚本及代码,在转换之后进行更新,确保数据的正确性。
我们简单看一下csscan的使用。
要使用csscan之前,需要以sys用户身份创建相应数据字典对象:
E:\nls2>sqlplus "/ as sysdba" SQL*Plus: Release 9.2.0.4.0 - Production on Sun Nov 2 19:42:07 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production SQL> select instance_name from v$intance; select instance_name from v$intance * ERROR at line 1: ORA-00942: table or view does not exist SQL> select instance_name from v$instance; INSTANCE_NAME ---------------- penny 1 row selected. SQL> @?/rdbms/admin/csminst.sql User created. Grant succeeded. ……….. |
这个脚本创建相应用户(csmig)及数据字典对象,扫描信息会记录在相应的数据字典表里。
我们可以在命令行调用这个工具对数据库进行扫描:
E:\nls2>csscan FULL=Y FROMCHAR=ZHS16GBK TOCHAR=US7ASCII LOG=US7check.log CAPTURE=Y ARRAY=1000000 PROCESS=2 Character Set Scanner v1.1 : Release 9.2.0.1.0 - Production on Sun Nov 2 20:24:45 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Username: eygle/eygle Connected to: Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP and Oracle Data Mining options JServer Release 9.2.0.4.0 - Production Enumerating tables to scan... . process 1 scanning SYS.SOURCE$[AAAABHAABAAAAIRAAA] . process 2 scanning SYS.ATTRIBUTE$[AAAAEoAABAAAAhZAAA] . process 2 scanning SYS.PARAMETER$[AAAAEoAABAAAAhZAAA] . process 2 scanning SYS.METHOD$[AAAAEoAABAAAAhZAAA] …….. . process 2 scanning SYSTEM.DEF$_AQERROR[AAAA8fAABAAACWJAAA] . process 1 scanning WMSYS.WM$ENV_VARS[AAABeWAABAAAFMZAAA] …………………. . process 2 scanning SYS.UGROUP$[AAAAA5AABAAAAGpAAA] . process 2 scanning SYS.CON$[AAAAAcAABAAAACpAAA] . process 1 scanning SYS.FILE$[AAAAARAABAAAABxAAA] Creating Database Scan Summary Report... Creating Individual Exception Report... Scanner terminated successfully. |
然后我们可以检查输出的日志来查看数据库扫描情况:
Database Scan Individual Exception Report [Database Scan Parameters] Parameter Value ------------------------------ ------------------------------------------------ Scan type Full database Scan CHAR data? YES Current database character set ZHS16GBK New database character set US7ASCII Scan NCHAR data? NO Array fetch buffer size 1000000 Number of processes 2 Capture convertible data? YES ------------------------------ ------------------------------------------------ [Data Dictionary individual exceptions] [Application data individual exceptions] User : EYGLE Table : TEST Column: NAME Type : VARCHAR2(10) Number of Exceptions : 1 Max Post Conversion Data Size: 4 ROWID Exception Type Size Cell Data(first 30 bytes) ------------------ ------------------ ----- ------------------------------ AAABpIAADAAAAAMAAA lossy conversion 测试 ------------------ ------------------ ----- ------------------------------ |
不能转换的数据将会被记录下来,我们可以根据这些信息在转换之后,对数据进行相应的更新,确保转换无误。
由于不同语言及字符集的共同存储存在设置上具有一定的复杂性,字符集一度成为普遍困扰大家的一个主要问题。
本文试图就一些常见问题进行探讨,希望可以把一些实际经验共享给大家!
1. 字符集的基本知识
Oracle数据库最早支持的编码方案也就是US7ASCII.
但是我们知道,英文字符一般是以一个字节来存储的,7位的编码方案最多只能代表128个字符;经过扩展的8位的编码方案也只能代表256个字符,这远远不能满足计算机发展的需要,对于亚洲国家复杂的字符存储需要更多的码位,于是各种编码方案随之而生。
为了容纳全世界各种语言的所有字符和符号,解决不同编码之间的兼容和转换问题,1991年元月,10多家公司共同出资,组建Unicode协会,随后Unicode编码产生了。
Unicode协会的口号是: 给每个字符提供了一个唯一的数字,不论是什么平台,不论是什么程序,不论什么语言。
最初Unicode编码使用2-Byte(16bit)来进行编码,但是最多只能容纳65536个字符,仍然不够使用,后来进行了扩充,也就是Unicode3.1标准,增加了额外的补充字符定义,现在Unicode4.0标准已经发布,具体可以参考Unicode官方站点:
www.unicode.org
Unicode编码方案主要有三个实施标准:
UTF-8
USC-2
UTF-16
Oracle从7.2开始支持UTF-8编码,提供Unicode编码支持。
2. 数据库的字符集
在创建数据库时,我们可以指定字符集(CHARACTER SET)和国家字符集(NATIONAL CHARACTER SET)。
字符集用来存储:
CHAR, VARCHAR2, CLOB, LONG等类型数据
用来标示诸如表名、列名以及PL/SQL变量等
SQL和PL/SQL程序单元等
国家字符集用以存储:
NCHAR, NVARCHAR2, NCLOB等类型数据
这些设置在数据库创建时指定,我们可以看一下数据库的创建脚本:
connect SYS/change_on_install as SYSDBAset echo onspool E:\oracle\ora92\assistants\dbca\logs\CreateDB.logstartup nomount pfile="E:\oracle\admin\eygle\scripts\init.ora";CREATE DATABASE eygleMAXINSTANCES 1MAXLOGHISTORY 1MAXLOGFILES 5MAXLOGMEMBERS 3MAXDATAFILES 100DATAFILE 'E:\oracle\oradata\eygle\system01.dbf' SIZE 250M REUSE AUTOEXTEND ON NEXT 10240K MAXSIZE UNLIMITEDEXTENT MANAGEMENT LOCALDEFAULT TEMPORARY TABLESPACE TEMP TEMPFILE 'E:\oracle\oradata\eygle\temp01.dbf' SIZE 40M REUSE AUTOEXTEND ON NEXT 640K MAXSIZE UNLIMITEDUNDO TABLESPACE "UNDOTBS1" DATAFILE 'E:\oracle\oradata\eygle\undotbs01.dbf' SIZE 50M REUSE AUTOEXTEND ON NEXT 5120K MAXSIZE UNLIMITEDCHARACTER SET ZHS16GBKNATIONAL CHARACTER SET AL16UTF16LOGFILE GROUP 1 ('E:\oracle\oradata\eygle\redo01.log') SIZE 10M,GROUP 2 ('E:\oracle\oradata\eygle\redo02.log') SIZE 10M,GROUP 3 ('E:\oracle\oradata\eygle\redo03.log') SIZE 10M;spool offexit;
|
以上用粗体显示的就是对我们至关重要的字符集设置。
在创建数据库的过程中,在以下界面选择你的字符集,对于简体中文平台,缺省的字符集是:ZHS16GBK
一旦你的字符集选定了,数据库中能够存储的字符就受到了限制,所以你选择的字符集的应该可以容纳所有你将用到字符。
常见的中文字符集有:
ZHS16CGB231280 CGB2312-80 16-bit Simplified Chinese MB, ASCII ZHS16GBK GBK 16-bit Simplified Chinese MB, ASCII, UDC |
其中GB2312码是中华人民共和国国家汉字信息交换用编码,全称《信息交换用汉字编码字符集--基本集》,由国家标准总局发布,
1981年5月1日实施,通行于大陆。新加坡等地也使用此编码。
GBK编码是1995年12月颁布的指导性规范。
GBK与国家标准 GB 2312-80 信息处理交换码所对应的、事实上的内码标准兼容;同时,在字汇一级支持 ISO/IEC 10646-1 和
GB 13000-1 的全部中日韩 (CJK) 汉字(20902字)。包含了更多的编码。
但是我们说,ZHS16GBK 并非是ZHS16CGB231280的严格超集(虽然后者的汉字在前者中都存在,但是同样的编码在不同两个
字符集中可能表达不同的汉字),所以在做数据库字符转换时仍然需要特别注意。
Oracle的字符集命名遵循以下命名规则:
|