Character references are mainly used in HTML and XML documents to represent characters independently of the encoding of the document. Character references may appear in two forms, numeric character references and character entity references. Numeric character references specify the Unicode code point value of a character, while character entity references use symbolic names to refer to the same character. For example, å is the numeric character reference for the small letter "a" with a ring above, whereas å is the character entity reference for the same character. Character entity references are also used to escape special characters, as an example, < represents the < (less than) sign. This is to avoid possible confusion with the beginning of a tag in Markup languages.

Syntax

UTL_I18N.ESCAPE_REFERENCE(
    str            IN VARCHAR2 CHARACTER SET ANY_CS,
    page_cs_name   IN VARCHAR2 DEFAULT NULL)
 RETURN VARCHAR2 CHARACTER SET str%CHARSET;

Parameters

Usage Notes

If the user specifies an invalid character set or a NULL string, then the function returns a NULL string.

Examples

UTL_I18N.ESCAPE_REFERENCE('hello < '||chr(229),'us7ascii')

This returns 'hello < å'.

Syntax

UTL_I18N.GET_COMMON_TIME_ZONES
 RETURN STRING_ARRAY;

Examples

Returns the list of the most commonly used time zones.

DECLARE
   retval UTL_I18N.STRING_ARRAY;
BEGIN
   retval := UTL_I18N.GET_COMMON_TIME_ZONES;
END;
/

Syntax

UTL_I18N.GET_DEFAULT_CHARSET( 
   language  IN VARCHAR2,
   context   IN PLS_INTEGER DEFAULT GENERIC_CONTEXT,
   iswindows IN BOOLEAN DEFAULT FALSE)
 RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid language name or an invalid flag, then the function returns a NULL string.

Examples

GENERIC_CONTEXT, iswindows=FALSE

UTL_I18N.GET_DEFAULT_CHARSET('French', UTL_I18N.GENERIC_CONTEXT, FALSE)

This returns 'WE8ISO8859P1'.

MAIL_CONTEXT, iswindows=TRUE

UTL_I18N.GET_DEFAULT_CHARSET('French', UTL_I18N.MAIL_CONTEXT, TRUE)

This returns 'WE8MSWIN1252'.

MAIL_CONTEXT, iswindows=FALSE

UTL_I18N.GET_DEFAULT_CHARSET('French', UTL_I18N.MAIL_CONTEXT, FALSE)

This returns 'WE8ISO8859P1'.

Syntax

UTL_I18N.GET_DEFAULT_ISO_CURRENCY (
   territory    IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid territory name, then the function returns a NULL string.

Examples

Displays the default ISO currency code for China.

DECLARE
   retval VARCHAR2(50);
BEGIN
   retval := UTL_I18N.GET_DEFAULT_ISO_CURRENCY('CHINA');
   DBMS_OUTPUT.PUT_LINE(retval);
END;
/

Syntax

UTL_I18N.GET_DEFAULT_LINGUISTIC_SORT (
   language  IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid language name, then the function returns a NULL string.

Examples

Displays the name of the most appropriate linguistic sort name for the language used in the current SQL session.

DECLARE
   retval VARCHAR2(50);
BEGIN
  SELECT value INTO retval FROM nls_database_parameters
   WHERE parameter = 'NLS_LANGUAGE';
   retval := UTL_I18N.GET_DEFAULT_LINGUISTIC_SORT(retval);
   DBMS_OUTPUT.PUT_LINE(retval);
END;
/

Syntax

UTL_I18N.GET_LOCAL_LANGUAGES (
   territory    IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN STRING_ARRAY;

Parameters

Usage Notes

If the user specifies an invalid territory name, then the function returns a NULL string.

Examples

Returns the list of local languages used in Belgium.

DECLARE
   retval UTL_I18N.STRING_ARRAY;
   cnt    INTEGER;
BEGIN
   retval   :=  UTL_I18N.GET_LOCAL_LANGUAGES('BELGIUM');
   DBMS_OUTPUT.PUT('Count = ');
   DBMS_OUTPUT.PUT_LINE(retval.LAST);
   cnt := retval.FIRST;
   WHILE cnt IS NOT NULL LOOP
      DBMS_OUTPUT.PUT_LINE(retval(cnt));
      cnt := retval.NEXT(cnt);
   END LOOP;
END;
/
...
Count = 2
DUTCH
FRENCH

Syntax

UTL_I18N.GET_LOCAL_LINGUISTIC_SORTS (
   language  IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN STRING_ARRAY;

Parameters

Usage Notes

If the user specifies an invalid language name, then the function returns a NULL string.

Examples

Displays the local linguistic sort names for JAPANESE.

DECLARE
   retval UTL_I18N.STRING_ARRAY;
   cnt INTEGER;
BEGIN
  retval := UTL_I18N.GET_LOCAL_LINGUISTIC_SORTS('Japanese');
  DBMS_OUTPUT.PUT('Count = ');
  DBMS_OUTPUT.PUT_LINE(retval.COUNT);
  cnt := retval.FIRST;
   WHILE cnt IS NOT NULL LOOP
      DBMS_OUTPUT.PUT_LINE(retval(cnt));
      cnt := retval.NEXT(cnt);
    END LOOP;
END;
/

...
Count = 2
JAPANESE_M
BINARY

Syntax

UTL_I18N.GET_LOCAL_TERRITORIES (
   language  IN VARCHAR2 CHARACTER SET ANY_CS)
 RETURN STRING_ARRAY;

Parameters

Usage Notes

If the user specifies an invalid language name, then the function returns a NULL string.

Examples

Returns the list of Oracle territories that use German as one of their local languages.

DECLARE
   retval  UTL_I18N.STRING_ARRAY;
   cnt     INTEGER;
BEGIN
   retval  := UTL_I18N.GET_LCOAL_TERRITORIIES('GERMAN');
   DBMS_OUTPUT.PUT('Count = ');
   DBMS_OUTPUT.PUT_LINE(retval.LAST);
  cnt := retval.FIRST;
   WHILE cnt IS NOT NULL LOOP
      DBMS_OUTPUT.PUT_LINE(retval(cnt));
      cnt := retval.NEXT(cnt));
   END LOOP;
END;
/
...
Count = 4
GERMANY
AUSTRIA
LUXEMBOURG
SWITZERLAND

Syntax

UTL_I18N.GET_LOCAL_TIME_ZONES ( 
   territory      IN VARCHAR2 CHARACTER SET ANY_CS DEFAULT NULL)
RETURN STRING_ARRAY;

Parameters

Usage Notes

If the user specifies an invalid territory name, then the function returns a NULL string.

Examples

Creates a function that returns the list of time zones locally used in the territory AZERBAIJAN followed by the general common time zones. This is useful for when the user's territory is known and the application still allows the user to choose other time zones as a user's preference.

CREATE OR REPLACE FUNCTION get_time_zones
(territory IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN utl_i18n.string_array
IS
  retval  utl_i18n.string_array;
  retval2 utl_i18n.string_array;
  stpos   INTEGER;
BEGIN
  retval  := utl_i18n.get_local_time_zones(
    territory);
  retval2 := utl_i18n.get_common_time_zones;
  stpos := retval.LAST + 1;
  retval(stpos) := '-----'; -- a separator
  FOR i IN retval2.FIRST..retval2.LAST LOOP
    stpos := stpos + 1;
    retval(stpos) := retval2(i);
  END LOOP;
  RETURN retval;
END;
/

Returns the list of local time zones for AZERBAIJAN followed by the common time zones with a separator string of five dashes (-----).

DECLARE
   retval UTL_I18N.STRING_ARRAY;
   cnt INTEGER;
BEGIN
   DBMS_OUTPUT.ENABLE(100000);
   retval UTL_I18N.GET_TIME_ZONES('AZERBAIJAN');
   cnt := retval.FIRST;
  WHILE cnt IS NOT NULL LOOP
    DBMS_OUTPUT.PUT_LINE(retval(cnt));
    cnt := retval.NEXT(cnt);
  END LOOP;
END;
/

Asia/Baku
-----
Pacific/Pago_Pago
Pacific/Honolulu
America/Anchorage
America/Vancouver
America/Los_Angeles
America/Tijuana
America/Edmonton
America/Denver
America/Phoenix
America/Mazatlan
America/Winnipeg
America/Regina
America/Chicago
America/Mexico_City
America/Guatemala
America/El_Salvador
America/Managua
America/Costa_Rica
America/Montreal
...

Syntax

UTL_I18N.GET_MAX_CHARACTER_SIZE(
    charset_name       IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN PLS_INTEGER;

Parameters

Usage Notes

For shift-sensitive character sets, the returned maximum character size will include the possible extra shift characters.

Examples

UTL_I18N.GET_MAX_CHARACTER_SIZE('AL32UTF8');

This returns 4.

Syntax

UTL_I18N.GET_TRANSLATION ( 
   parameter        IN VARCHAR2 CHARACTER SET ANY_CS,
   trans_language   IN VARCHAR2 'AMERICAN',
   flag             IN PLS_INTEGER DEFAULT LANGUAGE_TRANS)
RETURN VARCHAR2 CHARACTER SET parameter%CHARSET;

Parameters

Usage Notes

If VARCHAR2 is used as a parameter type, the returned translation text can be corrupted due to the conversion to the database character set. Using NVARCHAR2 as the parameter type will preserve the translation text because Unicode can encode all translated languages.

If the specified translation language is not available or an invalid name is provided, the default "American English" translations are returned. For example, Oracle does not provide GUJARATI translations, so the returned translation would be in American English.

Examples

The following returns the names of all the Oracle-supported languages in Italian.

DECLARE
   CURSOR c1 IS
     SELECT value FROM V$NLS_VALID_VALUES
     WHERE parameter = 'LANGUAGE'
  ORDER BY value;
  retval NVARCHAR2(100);
BEGIN
   FOR item IN c1 LOOP
     retval := UTL_I18N.GET_TRANSLATION (TO_NCHAR(item.value), 'italian');
   END LOOP;
END;

It maps the following:

• An Oracle character set name to an IANA character set name.

• An IANA character set name to an Oracle character set name.

• An Oracle character set to an e-mail safe character set name.

Syntax

UTL_I18N.MAP_CHARSET( 
   charset   IN VARCHAR2,
   context   IN PLS_INTEGER DEFAULT GENERIC_CONTEXT,
   flag      IN PLS_INTEGER DEFAULT ORACLE_TO_IANA)
RETURN VARCHAR2;

Parameters

Usage Notes

An e-mail safe character set is an Oracle character set that is commonly used by applications when they submit e-mail messages. The character set is usually used to convert contents in the database character set to e-mail safe contents. To specify the character set name in the mail header, you should use the corresponding IANA character set name obtained by calling the MAP_CHARSET function with the ORACLE_TO_IANA option, providing the e-mail safe character set name as input.

For example, no e-mail client recognizes message contents in the WE8DEC character set, whose corresponding IANA name is DEC-MCS. If WE8DEC is passed to the MAP_CHARSET function with the MAIL_CONTEXT option, then the function returns WE8ISO8859P1. Its corresponding IANA name, ISO-8859-1, is recognized by most e-mail clients.

The steps in this example are as follows:

1. Call the MAP_CHARSET function with the MAIL_CONTEXT | MAIL_GENERIC option with the database character set name, WE8DEC. The result is WE8ISO8859P1.

2. Convert the contents stored in the database to WE8ISO8859P1.

3. Call the MAP_CHARSET function with the ORACLE_TO_IANA | GENERIC_CONTEXT option with the e-mail safe character set, WE8ISO8859P1. The result is ISO-8859-1.

4. Specify ISO-8859-1 in the mail header when the e-mail message is submitted.

The function returns a character set name if a match is found. If no match is found or if the flag is invalid, then it returns NULL.

Examples

Generic Context

UTL_I18N.MAP_CHARSET('iso-8859-1',UTL_I18N.GENERIC_CONTEXT,UTL_I18N.IANA_TO_ORACLE)

This returns 'WE8ISO8859P1'.

Context

UTL_I18N.MAP_CHARSET('WE8DEC', utl_i18n.mail_context,  utl_i18n.mail_generic)

This returns 'WE8ISO8859P1'.

Syntax

UTL_I18N.MAP_FROM_SHORT_LANGUAGE (
   language           IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid language name, then the function returns a NULL string.

Examples

Returns the default linguistic sort name for the customer with the ID of 9000. Note that the table customers is from the oe user in the Common Schema. Because the customer's language preference is stored using a short language name, you need to convert to a full language name by calling the GET_DEFAULT_LINGUISTIC_SORT procedure.

DECLARE
   short_n VARCHAR2(10);
   ling_n VARCHAR2(50);
BEGIN
   SELECT nls_language INTO short 
   FROM customers WHERE customer_id = 9000;
   ling_n := UTL_I18N.GET_DEFAULT_LINGUISTIC_SORT (
   UTL_I18N.MAP_FROM_SHORT_LANGUAGE(short_n));
   DBMS_OUTPUT.PUT_LINE(ling_n);
END;
/

Syntax

UTL_I18N.MAP_LANGUAGE_FROM_ISO( 
  isolocale IN VARCHAR2)
RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid locale string, then the function returns a NULL string.

If the user specifies a locale string that includes only the language (for example, en_ instead of en_US), then the function returns the default language name for the specified language (for example, American).

Examples

UTL_I18N.MAP_LANGUAGE_FROM_ISO('en_US')

This returns 'American'.

A valid string must include at least one of the following: a valid Oracle language name or a valid Oracle territory name.

Syntax

UTL_I18N.MAP_LOCALE_TO_ISO ( 
   ora_language   IN VARCHAR2,
   ora_territory  IN VARCHAR2)
 RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid string, then the function returns a NULL string.

Examples

UTL_I18N.MAP_LOCALE_TO_ISO('American','America')

This returns 'en_US'.

Syntax

UTL_I18N.MAP_TERRITORY_FROM_ISO (
  isolocale IN VARCHAR2)
 RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid locale string, then the function returns a NULL string.

If the user specifies a locale string that includes only the territory (for example, _fr instead of fr_fr), then the function returns the default territory name for the specified territory (for example, France).

Examples

UTL_I18N.MAP_TERRITORY_FROM_ISO('en_US')

This returns 'America'.

Syntax

UTL_I18N.MAP_TO_SHORT_LANGUAGE (
   language    IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid language name, then the function returns a NULL string.

Examples

Returns the short language name for the language.

DECLARE  retval VARCHAR2(100);BEGIN  retval := UTL_I18N.MAP_TO_SHORT_LANGUAGE('american');  DBMS_OUTPUT.PUT_LINE(retval);END;/US

The function is overloaded. The different forms of functionality are described along with the syntax declarations.

Syntax

Buffer Conversion:

UTL_I18N.RAW_TO_CHAR(
   data          IN RAW,
   src_charset   IN VARCHAR2 DEFAULT NULL)
 RETURN VARCHAR2;

Piecewise conversion converts raw data into character data piece by piece:

UTL_I18N.RAW_TO_CHAR (
   data            IN RAW,
   src_charset     IN VARCHAR2 DEFAULT NULL,
   scanned_length  OUT PLS_INTEGER,
   shift_status    IN OUT PLS_INTEGER)
RETURN VARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid character set, NULL data, or data whose length is 0, then the function returns a NULL string.

Examples

Buffer Conversion

UTL_I18N.RAW_TO_CHAR(hextoraw('616263646566C2AA'), 'utf8')

This returns the following string in the database character set:

'abcde'||chr(170)

Piecewise Conversion

UTL_I18N.RAW_TO_CHAR(hextoraw('616263646566C2AA'),'utf8',shf,slen)

This expression returns the following string in the database character set:

'abcde'||chr(170)

It also sets shf to SHIFT_IN and slen to 8.

The following example converts data from the Internet piece by piece to the database character set.

rvalue RAW(1050); 
  nvalue VARCHAR2(1024); 
  conversion_state  PLS_INTEGER = 0; 
  converted_len   PLS_INTEGER; 
  rtemp  RAW(10) = ''; 
  conn   utl_tcp.connection; 
  tlen PLS_INTEGER;

  ... 
  conn := utl_tcp.open_connection ( remote_host => 'localhost', 
                                    remote_port => 2000); 
  LOOP 
      tlen := utl_tcp.read_raw(conn, rvalue, 1024); 
      rvalue := utl_raw.concat(rtemp, rvalue); 
      nvalue := utl_i18n.raw_to_char(rvalue, 'JA16SJIS', converted_len, conversion_stat); 
      if (converted_len < utl_raw.length(rvalue) ) 
      then 
        rtemp := utl_raw.substr(rvalue, converted_len+1); 
      else 
        rtemp := ''; 
      end if; 
      /* do anything you want with nvalue */ 
      /* e.g htp.prn(nvalue); */ 
    END LOOP; 
    utl_tcp.close_connection(conn); 
  EXCEPTION 
    WHEN utl_tcp.end_of_input THEN 
      utl_tcp.close_connection(conn);
END;

The function is overloaded. The different forms of functionality are described along with the syntax declarations.

Syntax

Buffer Conversion:

UTL_I18N.RAW_TO_NCHAR ( 
   data         IN RAW,
   src_charset  IN VARCHAR2 DEFAULT NULL)
 RETURN NVARCHAR2;

Piecewise conversion converts raw data into character data piece by piece:

UTL_I18N.RAW_TO_NCHAR ( 
   data            IN RAW,
   src_charset      IN VARCHAR2 DEFAULT NULL,
   scanned_length   OUT PLS_INTEGER,
   shift_status     IN OUT PLS_INTEGER)
 RETURN NVARCHAR2;

Parameters

Usage Notes

If the user specifies an invalid character set, NULL data, or data whose length is 0, then the function returns a NULL string.

Examples

Buffer Conversion

UTL_I18N.RAW_TO_NCHAR(hextoraw('616263646566C2AA'),'utf8')

This returns the following string in the national character set:

'abcde'||chr(170)

Piecewise Conversion

UTL_I18N.RAW_TO_NCHAR(hextoraw('616263646566C2AA'),'utf8', shf, slen)

This expression returns the following string in the national character set:

'abcde'||chr(170)

It also sets shf to SHIFT_IN and slen to 8.

The following example converts data from the Internet piece by piece to the national character set.

rvalue RAW(1050); 
  nvalue NVARCHAR2(1024); 
  converstion_state  PLS_INTEGER = 0; 
  converted_len   PLS_INTEGER; 
  rtemp  RAW(10) = ''; 
  conn   utl_tcp.connection; 
  tlen PLS_INTEGER;

  ... 
  conn := utl_tcp.open_connection ( remote_host => 'localhost', 
                                    remote_port => 2000); 
  LOOP 
      tlen := utl_tcp.read_raw(conn, rvalue, 1024); 
      rvalue := utl_raw.concat(rtemp, rvalue); 
      nvalue := utl_i18n.raw_to_nchar(rvalue, 'JA16SJIS', converted_len, conversion_stat); 
      if (converted_len < utl_raw.length(rvalue) ) 
      then 
        rtemp := utl_raw.substr(rvalue, converted_len+1); 
      else 
        rtemp := ''; 
      end if; 
      /* do anything you want with nvalue */ 
      /* e.g htp.prn(nvalue); */ 
    END LOOP; 
    utl_tcp.close_connection(conn); 
  EXCEPTION 
    WHEN utl_tcp.end_of_input THEN 
      utl_tcp.close_connection(conn); 
  END; 

Syntax

UTL_I18N.STRING_TO_RAW( 
   data          IN VARCHAR2 CHARACTER SET ANY_CS,
   dst_charset   IN VARCHAR2 DEFAULT NULL)
RETURN RAW;

Parameters

Usage Notes

If the user specifies an invalid character set, a NULL string, or a string whose length is 0, then the function returns a NULL string.

Examples

DECLARE 
    r raw(50); 
    s varchar2(20); 
  BEGIN 
    s:='abcdef'||chr(170); 
    r:=utl_i18n.string_to_raw(s,'utf8'); 
    dbms_output.put_line(rawtohex(r)); 
  end; 
/ 

This returns a hex value of '616263646566C2AA'.

Syntax

UTL_I18N.TRANSLITERATE (
  data IN VARCHAR2 CHARACTER SET ANY_CS,
  name IN VARCHAR2)
RETURN VARCHAR2 CHARACTER SET data%CHARSET;

Parameters

Constants

These options specify Japanese Kana conversions.

Usage Notes

The function returns the converted string.

Examples

Given a table japanese_emp, containing an NVARCHAR2 column ename, the following statement can be used to normalize all the kana names in ename to hiragana:

UPDATE japanese_emp
   SET ename = UTL_I18N.TRANSLITERATE (ename, 'kana_hiragana');

The following figure shows how this output might look.

Figure 263-1 Loading Locale-Specific Data to the Database

Description of "Figure 263-1 Loading Locale-Specific Data to the Database"

The following statement normalizes one kana name to hiragana:

DECLARE
   Name  japanese_emp.ename%TYPE;
   Eno   CONSTANT  NUMBER(4) := 1;
BEGIN
   SELECT ename INTO name FROM japanese_emp WHERE enumber = eno;
   name := UTL_I18N.TRANSLITERATE(name, UTL_I18N.KANA_HIRAGANA);
   UPDATE japanese_emp SET ename = name WHERE enumber = eno;
EXCEPTION
   WHEN  UTL_I18N.UNSUPPORTED_TRANSLITERATION THEN
   DBMS_OUTPUT.PUT_LINE('transliteration not supported');
END;
/

Syntax

UTL_I18N.UNESCAPE_REFERENCE ( 
   str IN VARCHAR2 CHARACTER SET ANY_CS)
 RETURN VARCHAR2 CHARACTER SET str%CHARSET;

Parameters

Usage Notes

If the user specifies a NULL string or a string whose length is 0, then the function returns a NULL string. If the function fails, then it returns the original string.

Examples

UTL_I18N.UNESCAPE_REFERENCE('hello < å')

This returns 'hello <'||chr(229).

For Unicode character sets, such as AL32UTF8, AL16UTF16, AL16UTF16LE, UTF8, and UTFE, any byte sequences mapped to the following Unicode code points are considered invalid:

• Unpaired surrogate code point

• Non-character code point

In addition, any irregular or illegal UTF-8 byte sequence is considered invalid for AL32UTF8 and UTF8 character sets.

The VALIDATE_CHARACTER_ENCODING function is overloaded. One function is for validating VARCHAR2 and NVARCHAR2 data, while the other function is for validating CLOB and NCLOB data.

• Validating VARCHAR2 and NVARCHAR2 data

  A VARCHAR2 or NVARCHAR2 byte or its byte sequence is considered invalid for a character set, if it does not map to any of the characters defined in the character set.

• Validating CLOB and NCLOB data

  A LOB character is considered invalid for a character set if a byte (in case of a single-byte database character set) or a byte pair (in case of UTF-16 encoding used with a multibyte database character set) corresponding to the encoding of the LOB character does not map to any of the characters defined in the character set.

Syntax

This function validates VARCHAR2 and NVARCHAR2 data:

UTL_I18N.VALIDATE_CHARACTER_ENCODING (
   data IN VARCHAR2 CHARACTER SET ANY_CS)
RETURN PLS_INTEGER;

This function validates CLOB and NCLOB data:

UTL_I18N.VALIDATE_CHARACTER_ENCODING (
   lob_loc IN CLOB CHARACTER SET ANY_CS)
RETURN PLS_INTEGER;

Parameters

Usage Notes

This function returns the offset of the first invalid byte for the VARCHAR2 or NVARCHAR2 data. It returns the offset of the first invalid character for the CLOB or NCLOB data. It returns 0, if all the bytes in the character data are valid. It returns NULL, if the value of the parameter data or lob_loc is NULL.

Examples

This example validates the character encoding of NVARCHAR2 and CLOB data where the database character set is AL32UTF8 while the national character set is AL16UTF16.

CREATE TABLE temp(col1 NVARCHAR2(20), col2 CLOB);
INSERT INTO temp VALUES(UNISTR('foo\D800bar'), UNISTR('foo\D800bar'));
COMMIT;
SELECT UTL_I18N.VALIDATE_CHARACTER_ENCODING(col1) invalid_offset_column1,
       UTL_I18N.VALIDATE_CHARACTER_ENCODING(col2) invalid_offset_column2
FROM temp;

The query returns:

INVALID_OFFSET_COLUMN1 INVALID_OFFSET_COLUMN2
---------------------- ----------------------
                     7                      4

Here, the surrogate code point U+D800 is invalid. The number 7 is returned as INVALID_OFFSET_COLUMN1, because for col1, ‘foo’ is encoded in 6 bytes in NVARCHAR2 and the invalid code point U+D800 starts at offset 7. The number 4 is returned as INVALID_OFFSET_COLUMN2, because for col2, ‘foo’ is encoded in 3 UTF-16 code points in CLOB and the invalid code point U+D800 starts at offset 4.