Replacing non-ASCII characters with HTML charcter references

Question

Hi All,
In Oracle 10g or greater is there a built-in function that will convert a string with non-ASCII characters like this
a b č 뮼
into an ASCII string with HTML character references like this?
a b & # x 0 1 0 D ; & # x B B B C ;
(note I had to include spaces between each character in the sample code for message to prevent the forum software from converting my text)
I tried using
utl_i18n.escape_reference( val, 'us7ascii' )
but for some reason it returns
a b c & # x B B B C ;
Note how it converted the Western European character "č" to its unaccented counterpart "c", not "& # x 0 1 0 D ;" (is this a bug?).
I also tried a custom solution using regexp_replace and asciistr (which I can't include here because the forum software chokes on it) but it only returns the correct result for values <=4000 characters long. Unfortunately asciistr doesn't appear to accept CLOB values larger than 4000 characters. It returns an error message like
(ORA-22835: Buffer too small for CLOB to CHAR or BLOB to RAW conversion (actual: 30251, maximum: 4000) ).
I'm looking for a solution that works on CLOB data of any size.
Thanks in advance for any insight you can provide.
Joe Fuda

Answer 1

So with that (UTF8) in mind, let's take another look.....
As shown below, I used a AL32UTF8 database.
Note: I did not use a unicode capable tool for querying. So I set console mode code page to 1250 just to have č displayed properly (instead of posing as an è).
Also, as a result of using windows-1250 for client character set, in the val column and in the second select's ncr column (iso8859-1), è (00e8) has been replaced with e through character set conversion going from server back to client.
Running the same code on a database with a db character set such as we8mswin1252, that doesn't define the č (latin small c with caron) character, would yield results with a c in the ncr column.
C:\>chcp 1250
Aktuell teckentabell: 1250
C:\>set nls_lang=.ee8mswin1250
C:\>sqlplus test/test
SQL*Plus: Release 11.1.0.6.0 - Production on Fri May 23 21:25:29 2008
Copyright (c) 1982, 2007, Oracle.  All rights reserved.
Connected to:
Oracle Database 11g Enterprise Edition Release 11.1.0.6.0 - Production
With the OLAP option
SQL> select * from nls_database_parameters where parameter like '%CHARACTERSET';
PARAMETER              VALUE
NLS_CHARACTERSET       AL32UTF8
NLS_NCHAR_CHARACTERSET AL16UTF16
SQL> select unistr('\010d \00e8') val, utl_i18n.escape_reference(unistr('\010d \00e8'),'us7ascii') NCR from dual;
VAL  NCR
č e  c e
SQL> select unistr('\010d \00e8') val, utl_i18n.escape_reference(unistr('\010d \00e8'),'we8iso8859p1') NCR from dual;
VAL  NCR
č e  &# x10d; e     <- "è"
SQL> select unistr('\010d \00e8') val, utl_i18n.escape_reference(unistr('\010d \00e8'),'ee8iso8859p2') NCR from dual;
VAL  NCR
č e  č &# xe8;
SQL> select unistr('\010d \00e8') val, utl_i18n.escape_reference(unistr('\010d \00e8'),'cl8iso8859p5') NCR from dual;
VAL  NCR
č e  &# x10d; &# xe8;In the US7ASCII case, where it should be possible for all non-ascii characters to be escaped, it seems as if the actual escape step is skipped over.
Hope this helps to understand whether utl_i8n is usable or not in your case.
Message was edited by:
orafad
Fixed replaced character references :)