path: root/man7/utf-8.7

.\" Hey Emacs! This file is -*- nroff -*- source.
.\"
.\" Copyright (C) Markus Kuhn, 1996, 2001
.\"
.\" This is free documentation; you can redistribute it and/or
.\" modify it under the terms of the GNU General Public License as
.\" published by the Free Software Foundation; either version 2 of
.\" the License, or (at your option) any later version.
.\"
.\" The GNU General Public License's references to "object code"
.\" and "executables" are to be interpreted as the output of any
.\" document formatting or typesetting system, including
.\" intermediate and printed output.
.\"
.\" This manual is distributed in the hope that it will be useful,
.\" but WITHOUT ANY WARRANTY; without even the implied warranty of
.\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
.\" GNU General Public License for more details.
.\"
.\" You should have received a copy of the GNU General Public
.\" License along with this manual; if not, write to the Free
.\" Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111,
.\" USA.
.\"
.\" 1995-11-26  Markus Kuhn <mskuhn@cip.informatik.uni-erlangen.de>
.\"      First version written
.\" 2001-05-11  Markus Kuhn <mgk25@cl.cam.ac.uk>
.\"      Update
.\"
.TH UTF-8 7 2001-05-11 "GNU" "Linux Programmer's Manual"
.SH NAME
UTF-8 \- an ASCII compatible multi-byte Unicode encoding
.SH DESCRIPTION
The
.B Unicode 3.0
character set occupies a 16-bit code space. The most obvious
Unicode encoding (known as
.BR UCS-2 )
consists of a sequence of 16-bit words. Such strings can contain as
parts of many 16-bit characters bytes like '\\0' or '/' which have a
special meaning in filenames and other C library function parameters.
In addition, the majority of UNIX tools expects ASCII files and can't
read 16-bit words as characters without major modifications. For these
reasons,
.B UCS-2
is not a suitable external encoding of
.B Unicode
in filenames, text files, environment variables, etc. The
.BR "ISO 10646 Universal Character Set (UCS)" ,
a superset of Unicode, occupies even a 31-bit code space and the obvious
.B UCS-4
encoding  for it (a sequence of 32-bit words) has the same problems.

The
.B UTF-8
encoding of
.B Unicode
and
.B UCS
does not have these problems and is the common way in which
.B Unicode
is used on Unix-style operating systems.
.SH PROPERTIES
The 
.B UTF-8 
encoding has the following nice properties:
.TP 0.2i
*
.B UCS
characters 0x00000000 to 0x0000007f (the classic
.B US-ASCII
characters) are encoded simply as bytes 0x00 to 0x7f (ASCII
compatibility). This means that files and strings which contain only
7-bit ASCII characters have the same encoding under both 
.B ASCII
and
.BR UTF-8 .
.TP
*
All
.B UCS
characters > 0x7f are encoded as a multi-byte sequence
consisting only of bytes in the range 0x80 to 0xfd, so no ASCII
byte can appear as part of another character and there are no
problems with e.g. '\\0' or '/'.
.TP
*
The lexicographic sorting order of
.B UCS-4
strings is preserved.
.TP
*
All possible 2^31 UCS codes can be encoded using 
.BR UTF-8 .
.TP
*
The bytes 0xfe and 0xff are never used in the
.B UTF-8
encoding.
.TP
* 
The first byte of a multi-byte sequence which represents a single non-ASCII
.B UCS
character is always in the range 0xc0 to 0xfd and indicates how long
this multi-byte sequence is. All further bytes in a multi-byte sequence
are in the range 0x80 to 0xbf. This allows easy resynchronization and
makes the encoding stateless and robust against missing bytes.
.TP
*
.B UTF-8
encoded
.B UCS
characters may be up to six bytes long, however the
.B Unicode
standard specifies no characters above 0x10ffff, so Unicode characters
can only be up to four bytes long in
.BR UTF-8 .
.SH ENCODING
The following byte sequences are used to represent a character. The
sequence to be used depends on the UCS code number of the character:
.TP 0.4i
0x00000000 - 0x0000007F:
.RI 0 xxxxxxx
.TP
0x00000080 - 0x000007FF:
.RI 110 xxxxx 
.RI 10 xxxxxx
.TP
0x00000800 - 0x0000FFFF:
.RI 1110 xxxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.TP
0x00010000 - 0x001FFFFF:
.RI 11110 xxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.TP
0x00200000 - 0x03FFFFFF:
.RI 111110 xx
.RI 10 xxxxxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.TP
0x04000000 - 0x7FFFFFFF:
.RI 1111110 x
.RI 10 xxxxxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.RI 10 xxxxxx
.PP
The
.I xxx
bit positions are filled with the bits of the character code number in
binary representation. Only the shortest possible multi-byte sequence
which can represent the code number of the character can be used.
.PP
The
.B UCS
code values 0xd800\(en0xdfff (UTF-16 surrogates) as well as 0xfffe and
0xffff (UCS non-characters) should not appear in conforming
.B UTF-8
streams.
.SH EXAMPLES
The
.B Unicode
character 0xa9 = 1010 1001 (the copyright sign) is encoded
in UTF-8 as
.PP
.RS
11000010 10101001 = 0xc2 0xa9
.RE
.PP
and character 0x2260 = 0010 0010 0110 0000 (the "not equal" symbol) is
encoded as:
.PP
.RS
11100010 10001001 10100000 = 0xe2 0x89 0xa0
.RE
.SH "APPLICATION NOTES"
Users have to select a 
.B UTF-8
locale, for example with
.PP
.RS
export LANG=en_GB.UTF-8
.RE
.PP
in order to activate the 
.B UTF-8
support in applications.
.PP
Application software that has to be aware of the used character
encoding should always set the locale with for example
.PP
.RS
setlocale(LC_CTYPE, "")
.RE
.PP
and programmers can then test the expression
.PP
.RS
strcmp(nl_langinfo(CODESET), "UTF-8") == 0
.RE
.PP
to determine whether a 
.B UTF-8
locale has been selected and whether
therefore all plaintext standard input and output, terminal
communication, plaintext file content, filenames and environment
variables are encoded in 
.BR UTF-8 .
.PP
Programmers accustomed to single-byte encodings such as
.B US-ASCII
or
.B ISO 8859
have to be aware that two assumptions made so far are no longer valid
in
.B UTF-8
locales. Firstly, a single byte does not necessarily correspond any
more to a single character. Secondly, since modern terminal emulators
in 
.B UTF-8
mode also support Chinese, Japanese, and Korean
.B double-width characters
as well as non-spacing
.BR "combining characters"  ,
outputting a single character does not necessarily advance the cursor
by one position as it did in 
.BR ASCII .
Library functions such as
.BR mbsrtowcs (3)
and
.BR wcswidth (3)
should be used today to count characters and cursor positions.
.PP
The official ESC sequence to switch from an
.B ISO 2022
encoding scheme (as used for instance by VT100 terminals) to
.B UTF-8
is ESC % G
("\\x1b%G"). The corresponding return sequence from
.B UTF-8
to ISO 2022 is ESC % @ ("\\x1b%@"). Other ISO 2022 sequences (such as
for switching the G0 and G1 sets) are not applicable in UTF-8 mode.
.PP
It can be hoped that in the foreseeable future,
.B UTF-8
will replace
.B ASCII
and
.B ISO 8859
at all levels as the common character encoding on POSIX systems,
leading to a significantly richer environment for handling plain text.
.SH SECURITY
The
.BR Unicode " and " UCS
standards require that producers of 
.B UTF-8
shall use the shortest form possible, e.g., producing a two-byte
sequence with first byte 0xc0 is non-conforming.
.B Unicode 3.1
has added the requirement that conforming programs must not accept
non-shortest forms in their input. This is for security reasons: if
user input is checked for possible security violations, a program
might check only for the
.B ASCII
version of "/../" or ";" or NUL and overlook that there are many
.RB non- ASCII
ways to represent these things in a non-shortest
.B UTF-8
encoding.
.SH STANDARDS
ISO/IEC 10646-1:2000, Unicode 3.1, RFC 2279, Plan 9.
.SH AUTHOR
Markus Kuhn <mgk25@cl.cam.ac.uk>
.SH "SEE ALSO"
.BR nl_langinfo (3),
.BR setlocale (3),
.BR charsets (7),
.BR unicode (7)