Network Working Group
Request for Comments: 4042
Category: Informational

M. Crispin
Panda Programming
1 April 2005

UTF-9 and UTF-18

Efficient Transformation Formats of Unicode

Status of This Memo

This memo provides information for the Internet community. It does not specify an Internet standard of any kind. Distribution of this memo is unlimited.

Copyright Notice

Abstract

ISO-10646 defines a large character set called the Universal Character Set (UCS), which encompasses most of the world's writing systems. The same set of codepoints is defined by Unicode, which further defines additional character properties and other implementation details. By policy of the relevant standardization committees, changes to Unicode and amendments and additions to ISO/IEC 646 track each other, so that the character repertoires and code point assignments remain in synchronization.

The current representation formats for Unicode (UTF-7, UTF-8, UTF-16) are not storage and computation efficient on platforms that utilize the 9 bit nonet as a natural storage unit instead of the 8 bit octet.

This document describes a transformation format of Unicode that takes advantage of the nonet so that the format will be storage and computation efficient.

1. Introduction

A number of Internet sites utilize platforms that are not based upon the traditional 8-bit byte or octet. One such platform is the PDP- 10, which is based upon a 36-bit word. On these platforms, it is wasteful to represent data in octets, since 4 bits are left unused in each word. The 9-bit nonet is a much more sensible representation.

Although these platforms support IETF standards, many of these platforms still utilize a text representation based upon the septet, which is only suitable for [US-ASCII] (although it has been used for various ISO 10646 national variants).

To maximize international and multi-lingual interoperability, the IAB has recommended ([IAB-CHARACTER]) that [ISO-10646] be the default coded character set.

Although other transformation formats of [UNICODE] exist, and conceivably can be used on nonet-oriented machines (most notably [UTF-8]), they suffer significant disadvantages:

[UTF-8]: requires one to three octets to represent codepoints in the Basic Multilingual Plane (BMP), four octets to represent [UNICODE] codepoints outside the BMP, and six octets to represent non-[UNICODE] codepoints. When stored in nonets, this results in as many as four wasted bits per [UNICODE] character.

[UTF-16]: requires a hexadecet to represent codepoints in the BMP, and two hexadecets to represent [UNICODE] codepoints outside the BMP. When stored in nonet pairs, this results in as many as four wasted bits per [UNICODE] character. This transformation format requires complex surrogates to represent codepoints outside the BMP, and can not represent non-[UNICODE] codepoints at all.

[UTF-7]: requires one to five septets to represent codepoints in the BMP, and as many as eight septets to represent codepoints outside the BMP. When stored in nonets, this results in as many as sixteen wasted bits per character. This transformation format requires very complex and computationally expensive shifting and "modified BASE64" processing, and can not represent non-[UNICODE] codepoints at all.

By comparison, UTF-9 uses one to two nonets to represent codepoints in the BMP, three nonets to represent [UNICODE] codepoints outside the BMP, and three or four nonets to represent non-[UNICODE] codepoints. There are no wasted bits, and as the examples in this document demonstrate, the computational processing is minimal.

Transformation between [UTF-8] and UTF-9 is straightforward, with most of the complexity in the handling of [UTF-8]. It is hoped that future extensions to protocols such as SMTP will permit the use of UTF-9 in these protocols between nonet platforms without the use of [UTF-8] as an "on the wire" format.

Similarly, transformation between [UNICODE] codepoints and UTF-18 is also quite simple. Although (like UCS-2) UTF-18 only represents a subset of the available [UNICODE] codepoints, it encompasses the non-private codepoints that are currently assigned in [UNICODE].

1.1. Conventions Used in This Document

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14, RFC 2119 [KEYWORDS].

2. Overview

UTF-9 encodes [UNICODE] codepoints in the low order 8 bits of a nonet, using the high order bit to indicate continuation. Surrogates are not used.

[UNICODE] codepoints in the range U+0000 - U+00FF ([US-ASCII] and Latin 1) are represented by a single nonet; codepoints in the range U+0100 - U+FFFF (the remainder of the BMP) are represented by two nonets; and codepoints in the range U+1000 - U+10FFFF (remainder of [UNICODE]) are represented by three nonets.

Non-[UNICODE] codepoints in [ISO-10646] (that is, codepoints in the range 0x110000 - 0x7fffffff) can also be represented in UTF-9 by obvious extension, but this is not discussed further as these codepoints have been removed from [ISO-10646] by ISO.

UTF-18 encodes [UNICODE] codepoints in the Basic Multilingual Plane (BMP, plane 0), Supplementary Multilingual Plane (SMP, plane 1), Supplementary Ideographic Plane (SIP, plane 2), and Supplementary Special-purpose Plane (SSP, plane 14) in a single 18-bit value. It does not encode planes 3 though 13, which are currently unused; nor planes 15 or 16, which are private spaces.

Normally, UTF-9 and UTF-18 should only be used in the context of 9 bit storage and transport. Although some protocols, e.g., [FTP], support transport of nonets, the current IETF protocol suite is quite deficient in this area. The IETF is urged to take action to improve IETF protocol support for nonets.

3. UTF-9 Definition

A UTF-9 stream represents [ISO-10646] codepoints using 9 bit nonets. The low order 8-bits of a nonet is an octet, and the high order bit indicates continuation.

UTF-9 does not use surrogates; consequently a UTF-16 value must be transformed into the UCS-4 equivalent, and U+D800 - U+DBFF are never transmitted in UTF-9.

Octets of the [UNICODE] codepoint value are then copied into successive UTF-9 nonets, starting with the most-significant non-zero octet. All but the least significant octet have the continuation bit set in the associated nonet.

Examples:

   Character  Name                                UTF-9 (in octal)
   ---------  ----                                ----------------
    U+0041    LATIN CAPITAL LETTER A              101
    U+00C0    LATIN CAPITAL LETTER A WITH GRAVE   300
    U+0391    GREEK CAPITAL LETTER ALPHA          403 221
    U+611B    <CJK ideograph meaning "love">      541 33
    U+10330   GOTHIC LETTER AHSA                  401 403 60
    U+E0041   TAG LATIN CAPITAL LETTER A          416 400 101
    U+10FFFD  <Plane 16 Private Use, Last>        420 777 375
   0x345ecf1b (UCS-4 value not in [UNICODE])      464 536 717 33

4. UTF-18 Definition

A UTF-18 stream represents [ISO-10646] codepoints using a pair of 9 bit nonets to form an 18-bit value.

UTF-18 does not use surrogates; consequently a UTF-16 value must be transformed into the UCS-4 equivalent, and U+D800 - U+DBFF are never transmitted in UTF-18.

[UNICODE] codepoint values in the range U+0000 - U+2FFFF are copied as the same value into a UTF-18 value. [UNICODE] codepoint values in the range U+E0000 - U+EFFFF are copied as values 0x30000 - 0x3ffff; that is, these values are shifted by 0x70000. Other codepoint values can not be represented in UTF-18.

Examples:

   Character  Name                                UTF-18 (in octal)
   ---------  ----                                ----------------
    U+0041    LATIN CAPITAL LETTER A              000101
    U+00C0    LATIN CAPITAL LETTER A WITH GRAVE   000300
    U+0391    GREEK CAPITAL LETTER ALPHA          001621
    U+611B    <CJK ideograph meaning "love">      060433
    U+10330   GOTHIC LETTER AHSA                  201460
    U+E0041   TAG LATIN CAPITAL LETTER A          600101

5. Sample Routines

5.1. [UNICODE] Codepoint to UTF-9 Conversion

The following routines demonstrate conversion from UCS-4 to UTF-9. For simplicity, these routines do not do any validity checking. Routines used in applications SHOULD reject invalid UTF-9 sequences; that is, the first nonet with a value of 400 octal (0x100), or sequences that result in an overflow (exceeding 0x10ffff for [UNICODE]), or codepoints used for UTF-16 surrogates.

   ; Return UCS-4 value from UTF-9 string (PDP-10 assembly version)
   ; Accepts: P1/ 9-bit byte pointer to UTF-9 string
   ; Returns +1: Always, T1/ UCS-4 value, P1/ updated byte pointer
   ; Clobbers T2
   
   UT92U4: TDZA T1,T1              ; start with zero
   U92U41:  XOR T1,T2              ; insert octet into UCS-4 value
           LSH T1,^D8              ; shift UCS-4 value
           ILDB T2,P1              ; get next nonet
           TRZE T2,400             ; extract octet, any continuation?
            JRST U92U41            ; yes, continue
           XOR T1,T2               ; insert final octet
           POPJ P,
   
   /* Return UCS-4 value from UTF-9 string (C version)
    * Accepts: pointer to pointer to UTF-9 string
    * Returns: UCS-4 character, nonet pointer updated
    */
   
   UINT31 UTF9_to_UCS4 (UINT9 **utf9PP)
   {
     UINT9 nonet;
     UINT31 ucs4;
     for (ucs4 = (nonet = *(*utf9PP)++) & 0xff;
          nonet & 0x100;
          ucs4 |= (nonet = *(*utf9PP)++) & 0xff)
       ucs4 <<= 8;
     return ucs4;
   }

5.2. UTF-9 to UCS-4 Conversion

The following routines demonstrate conversion from UTF-9 to UCS-4. For simplicity, these routines do not do any validity checking. Routines used in applications SHOULD reject invalid UCS-4 codepoints; that is, codepoints used for UTF-16 surrogates or codepoints with values exceeding 0x10ffff for [UNICODE].

   ; Write UCS-4 character to UTF-9 string (PDP-10 assembly version)
   ; Accepts: P1/ 9-bit byte pointer to UTF-9 string
   ;          T1/ UCS-4 character to write
   ; Returns +1: Always, P1/ updated byte pointer
   ; Clobbers T1, T2; (T1, T2) must be an accumulator pair
   
   U42UT9: SETO T2,            ; we'll need some of these 1-bits later
           ASHC T1,-^D8        ; low octet becomes nonet with high 0-bit
   U32U91: JUMPE T1,U42U9X     ; done if no more octets
           LSHC T1,-^D8        ; shift next octet into T2
           ROT T2,-1           ; turn it into nonet with high 1 bit
           PUSHJ P,U42U91      ; recurse for remainder
   U42U9X: LSHC T1,^D9         ; get next nonet back from T2
           IDPB T1,P1          ; write nonet
           POPJ P,
   
   /* Write UCS-4 character to UTF-9 string (C version)
    * Accepts: pointer to nonet string
    *          UCS-4 character to write
    * Returns: updated pointer
    */
   
   UINT9 *UCS4_to_UTF9 (UINT9 *utf9P,UINT31 ucs4)
   {
     if (ucs4 > 0x100) {
       if (ucs4 > 0x10000) {
         if (ucs4 > 0x1000000)
           *utf9P++ = 0x100 | ((ucs4 >> 24) & 0xff);
         *utf9P++ = 0x100 | ((ucs4 >> 16) & 0xff);
       }
       *utf9P++ = 0x100 | ((ucs4 >> 8) & 0xff);
     }
     *utf9P++ = ucs4 & 0xff;
     return utf9P;
   }

6. Implementation Experience

   As the sample routines demonstrate, it is quite simple to implement
   UTF-9 and UTF-18 on a nonet-based architecture.  More sophisticated
   routines can be found in ftp://panda.com/tops-20/utools.mac.txt or
   from lingling.panda.com via the file <UTF9>UTOOLS.MAC via ANONYMOUS
   [FTP].

We are now in the process of implementing support for nonet-based text files and automated transformation between septet, octet, and nonet textual data.

7. References

7.1. Normative References

   [FTP]           Postel, J. and J. Reynolds, "File Transfer Protocol",
                   STD 9, RFC 959, October 1985.
   
   [IAB-CHARACTER] Weider, C., Preston, C., Simonsen, K., Alvestrand,
                   H., Atkinson, R., Crispin, M., and P. Svanberg, "The
                   Report of the IAB Character Set Workshop held 29
                   February - 1 March, 1996", RFC 2130, April 1997.
   
   [ISO-10646]     International Organization for Standardization,
                   "Information Technology - Universal Multiple-octet
                   coded Character Set (UCS)", ISO/IEC Standard 10646,
                   comprised of ISO/IEC 10646-1:2000, "Information
                   technology - Universal Multiple-Octet Coded Character
                   Set (UCS) - Part 1: Architecture and Basic
                   Multilingual Plane", ISO/IEC 10646-2:2001,
                   "Information technology - Universal Multiple-Octet
                   Coded Character Set (UCS) - Part 2:  Supplementary
                   Planes" and ISO/IEC 10646-1:2000/Amd 1:2002,
                   "Mathematical symbols and other characters".
   
   [KEYWORDS]      Bradner, S., "Key words for use in RFCs to Indicate
                   Requirement Levels", BCP 14, RFC 2119, March 1997.
   
   [UNICODE]       The Unicode Consortium, "The Unicode Standard -
                   Version 3.2", defined by The Unicode Standard,
                   Version 3.0 (Reading, MA, Addison-Wesley, 2000.  ISBN
                   0-201-61633-5), as amended by the Unicode Standard
                   Annex #27: Unicode 3.1 and by the Unicode Standard
                   Annex #28: Unicode 3.2, March 2002.

7.2. Informative References

   [US-ASCII]      American National Standards Institute, "Coded
                   Character Set - 7-bit American Standard Code for
                   Information Interchange", ANSI X3.4, 1986.
   
   [UTF-16]        Hoffman, P. and F. Yergeau, "UTF-16, an encoding of
                   ISO 10646", RFC 2781, February 2000.
   
   [UTF-7]         Goldsmith, D. and M. Davis, "UTF-7 A Mail-Safe
                   Transformation Format of Unicode", RFC 2152, May
                   1997.
   
   [UTF-8]         Sollins, K., "Architectural Principles of Uniform
                   Resource Name Resolution", RFC 2276, January 1998.

8. Security Considerations

As with UTF-8, UTF-9 can represent codepoints that are not in [UNICODE]. Applications should validate UTF-9 strings to ensure that all codepoints do not exceed the [UNICODE] maximum of U+10FFFF.

The sample routines in this document are for example purposes, and make no attempt to validate their arguments, e.g., test for overflow ([UNICODE] values great than 0x10ffff) or codepoints used for surrogates. Besides resulting in invalid data, this can also create covert channels.

9. IANA Considerations

The IANA shall reserve the charset names "UTF-9" and "UTF-18" for future assignment.

Author's Address

Mark R. Crispin
Panda Programming
6158 NE Lariat Loop
Bainbridge Island, WA 98110-2098

   Phone: (206) 842-2385
   EMail: UTF9@Lingling.Panda.COM

Full Copyright Statement

This document is subject to the rights, licenses and restrictions contained in BCP 78 and at www.rfc-editor.org/copyright.html, and except as set forth therein, the authors retain all their rights.

This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79.

Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr.

The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf- ipr@ietf.org.

Acknowledgement

Funding for the RFC Editor function is currently provided by the Internet Society.