Issue with LPAD/RPAD when using with Japanese data

Question

Hi,
I am trying to apply LPAD/RPAD on Japanese data, but its giving different results than expected.
LPAD/RPAD is returning less data than length passed.
Database: Oracle 10g
Could some one help me on this.
SQL Query: select length('アップリカ'),lpad('アップリカ',5,' '), length(lpad('アップリカ',5,' ')) from dual;
Output:
LENGTH('アップリカ') --> *5*
LPAD('アップリカ',5,'') --> アッ
LENGTH(LPAD('アップリカ',5,'')) --> *3*
Thanks in advance !!
Edited by: 871132 on Jul 7, 2011 8:18 PM
Edited by: 871132 on Jul 7, 2011 8:20 PM
Edited by: 871132 on Jul 7, 2011 8:21 PM

Answer 1

I have made a little research and it seems that your problem is connected with way RPAD function works.
Actually it counts 'display units' instead of 'real' characters count. And in japanesse one 'real' character may be formed from few display units.
Oracle gives some info about it in its official documentation:
http://download.oracle.com/docs/cd/B19306_01/olap.102/b14346/dml_x_reserved010.htm
Some more info is given here:
http://www.rhinocerus.net/forum/databases-oracle-tools/426832-oracle-bug-rpad-japanese-kanji-character-oracle-10gr2-utf8database.html
And also sugestion is given:
+"If you really want the number of characters then you can use+
+something like:+
+RPAD ( str , n - LENGTHC(str),'c')+
+Use LENGTHB, if the requested width is in bytes, LENGTHC, if in+
+characters+
+Or+
+SUBSTR( str || RPAD( 'c', n, 'c' ), 1, n )+
+Use SUBSTRB, if the requested width is in bytes, SUBSTR, if in+
+characters+
+In above the+
+* str is the string to be padded.+
+*'c' is the fill character (usually blank) -- we a assume single-+
+byte char+
+* n is the requested width in bytes or characters+
+"+
Edited by: chudapet on 8.7.2011 1:52