9 our @ISA = qw(Exporter DynaLoader);
11 # Public, encouraged API is exported by default
37 # Documentation moved after __END__ for speed - NI-S
41 # Make a %encoding package variable to allow a certain amount of cheating
43 my @alias; # ordered matching list
44 my %alias; # cached known aliases
49 return keys %encoding;
56 unless (exists $alias{$_})
58 for (my $i=0; $i < @alias; $i += 2)
60 my $alias = $alias[$i];
61 my $val = $alias[$i+1];
63 if (ref($alias) eq 'Regexp' && $_ =~ $alias)
67 elsif (ref($alias) eq 'CODE')
69 $new = &{$alias}($val)
71 elsif (lc($_) eq $alias)
77 next if $new eq $_; # avoid (direct) recursion on bugs
78 my $enc = (ref($new)) ? $new : find_encoding($new);
94 my ($alias,$name) = splice(@_,0,2);
95 push(@alias, $alias => $name);
99 define_alias( qr/^iso(\d+-\d+)$/i => '"iso-$1"' );
100 define_alias( qr/^(\S+)\s+(.*)$/i => '"$1-$2"' );
101 #define_alias( sub { return /^iso-(\d+-\d+)$/i ? "iso$1" : '' } );
102 define_alias( 'ascii' => 'US-ascii');
103 define_alias( 'ibm-1047' => 'cp1047');
109 $encoding{$name} = $obj;
111 define_alias($lc => $obj) unless $lc eq $name;
115 define_alias($alias,$obj);
122 my ($class,$name) = @_;
124 if (exists $encoding{$name})
126 return $encoding{$name};
130 return $class->findAlias($name);
137 return __PACKAGE__->getEncoding($name);
142 my ($name,$string,$check) = @_;
143 my $enc = find_encoding($name);
144 croak("Unknown encoding '$name'") unless defined $enc;
145 my $octets = $enc->encode($string,$check);
146 return undef if ($check && length($string));
152 my ($name,$octets,$check) = @_;
153 my $enc = find_encoding($name);
154 croak("Unknown encoding '$name'") unless defined $enc;
155 my $string = $enc->decode($octets,$check);
156 return undef if ($check && length($octets));
162 my ($string,$from,$to,$check) = @_;
163 my $f = find_encoding($from);
164 croak("Unknown encoding '$from'") unless defined $f;
165 my $t = find_encoding($to);
166 croak("Unknown encoding '$to'") unless defined $t;
167 my $uni = $f->decode($string,$check);
168 return undef if ($check && length($string));
169 $string = $t->encode($uni,$check);
170 return undef if ($check && length($uni));
171 return length($_[0] = $string);
184 return undef unless utf8_decode($str);
188 package Encode::Encoding;
189 # Base class for classes which implement encodings
194 my $canonical = shift;
195 $obj = bless { Name => $canonical },$obj unless ref $obj;
196 # warn "$canonical => $obj\n";
197 Encode::define_encoding($obj, $canonical, @_);
200 sub name { shift->{'Name'} }
202 # Temporary legacy methods
203 sub toUnicode { shift->decode(@_) }
204 sub fromUnicode { shift->encode(@_) }
206 sub new_sequence { return $_[0] }
209 use base 'Encode::Encoding';
211 package Encode::Unicode;
212 use base 'Encode::Encoding';
214 # Dummy package that provides the encode interface but leaves data
215 # as UTF-8 encoded. It is here so that from_to() works.
217 __PACKAGE__->Define('Unicode');
221 my ($obj,$str,$chk) = @_;
222 Encode::utf8_upgrade($str);
229 package Encode::utf8;
230 use base 'Encode::Encoding';
231 # package to allow long-hand
232 # $octets = encode( utf8 => $string );
235 __PACKAGE__->Define(qw(UTF-8 utf8));
239 my ($obj,$octets,$chk) = @_;
240 my $str = Encode::decode_utf8($octets);
251 my ($obj,$string,$chk) = @_;
252 my $octets = Encode::encode_utf8($string);
257 package Encode::iso10646_1;
258 use base 'Encode::Encoding';
259 # Encoding is 16-bit network order Unicode (no surogates)
260 # Used for X font encodings
262 __PACKAGE__->Define(qw(UCS-2 iso10646-1));
266 my ($obj,$str,$chk) = @_;
270 my $code = unpack('n',substr($str,0,2,'')) & 0xffff;
273 $_[1] = $str if $chk;
274 Encode::utf8_upgrade($uni);
280 my ($obj,$uni,$chk) = @_;
284 my $ch = substr($uni,0,1,'');
291 $str .= pack('n',$x);
293 $_[1] = $uni if $chk;
297 # switch back to Encode package in case we ever add AutoLoader
306 Encode - character encodings
314 The C<Encode> module provides the interfaces between perl's strings
315 and the rest of the system. Perl strings are sequences of B<characters>.
317 The repertoire of characters that Perl can represent is at least that
318 defined by the Unicode Consortium. On most platforms the ordinal values
319 of the characters (as returned by C<ord(ch)>) is the "Unicode codepoint" for
320 the character (the exceptions are those platforms where the legacy
321 encoding is some variant of EBCDIC rather than a super-set of ASCII
322 - see L<perlebcdic>).
324 Traditionaly computer data has been moved around in 8-bit chunks
325 often called "bytes". These chunks are also known as "octets" in
326 networking standards. Perl is widely used to manipulate data of
327 many types - not only strings of characters representing human or
328 computer languages but also "binary" data being the machines representation
329 of numbers, pixels in an image - or just about anything.
331 When perl is processing "binary data" the programmer wants perl to process
332 "sequences of bytes". This is not a problem for perl - as a byte has 256
333 possible values it easily fits in perl's much larger "logical character".
341 I<character>: a character in the range 0..(2**32-1) (or more).
342 (What perl's strings are made of.)
346 I<byte>: a character in the range 0..255
347 (A special case of a perl character.)
351 I<octet>: 8 bits of data, with ordinal values 0..255
352 (Term for bytes passed to or from a non-perl context, e.g. disk file.)
356 The marker [INTERNAL] marks Internal Implementation Details, in
357 general meant only for those who think they know what they are doing,
358 and such details may change in future releases.
362 =head2 Characteristics of an Encoding
364 An encoding has a "repertoire" of characters that it can represent,
365 and for each representable character there is at least one sequence of
366 octets that represents it.
368 =head2 Types of Encodings
370 Encodings can be divided into the following types:
374 =item * Fixed length 8-bit (or less) encodings.
376 Each character is a single octet so may have a repertoire of up to
377 256 characters. ASCII and iso-8859-* are typical examples.
379 =item * Fixed length 16-bit encodings
381 Each character is two octets so may have a repertoire of up to
382 65,536 characters. Unicode's UCS-2 is an example. Also used for
383 encodings for East Asian languages.
385 =item * Fixed length 32-bit encodings.
387 Not really very "encoded" encodings. The Unicode code points
388 are just represented as 4-octet integers. None the less because
389 different architectures use different representations of integers
390 (so called "endian") there at least two disctinct encodings.
392 =item * Multi-byte encodings
394 The number of octets needed to represent a character varies.
395 UTF-8 is a particularly complex but regular case of a multi-byte
396 encoding. Several East Asian countries use a multi-byte encoding
397 where 1-octet is used to cover western roman characters and Asian
398 characters get 2-octets.
399 (UTF-16 is strictly a multi-byte encoding taking either 2 or 4 octets
400 to represent a Unicode code point.)
402 =item * "Escape" encodings.
404 These encodings embed "escape sequences" into the octet sequence
405 which describe how the following octets are to be interpreted.
406 The iso-2022-* family is typical. Following the escape sequence
407 octets are encoded by an "embedded" encoding (which will be one
408 of the above types) until another escape sequence switches to
409 a different "embedded" encoding.
411 These schemes are very flexible and can handle mixed languages but are
412 very complex to process (and have state).
413 No escape encodings are implemented for perl yet.
417 =head2 Specifying Encodings
419 Encodings can be specified to the API described below in two ways:
425 Encoding names are strings with characters taken from a restricted repertoire.
426 See L</"Encoding Names">.
428 =item 2. As an object
430 Encoding objects are returned by C<find_encoding($name)>.
434 =head2 Encoding Names
436 Encoding names are case insensitive. White space in names is ignored.
437 In addition an encoding may have aliases. Each encoding has one "canonical" name.
438 The "canonical" name is chosen from the names of the encoding by picking
439 the first in the following sequence:
443 =item * The MIME name as defined in IETF RFC-XXXX.
445 =item * The name in the IANA registry.
447 =item * The name used by the the organization that defined it.
451 Because of all the alias issues, and because in the general case
452 encodings have state C<Encode> uses the encoding object internally
453 once an operation is in progress.
455 =head1 PERL ENCODING API
457 =head2 Generic Encoding Interface
463 $bytes = encode(ENCODING, $string[, CHECK])
465 Encodes string from perl's internal form into I<ENCODING> and returns a
467 See L</"Handling Malformed Data">.
471 $string = decode(ENCODING, $bytes[, CHECK])
473 Decode sequence of octets assumed to be in I<ENCODING> into perls internal
474 form and returns the resuting string.
475 See L</"Handling Malformed Data">.
479 =head2 Handling Malformed Data
481 If CHECK is not set, C<undef> is returned. If the data is supposed to
482 be UTF-8, an optional lexical warning (category utf8) is given.
483 If CHECK is true but not a code reference, dies.
485 It would desirable to have a way to indicate that transform should use the
486 encodings "replacement character" - no such mechanism is defined yet.
488 It is also planned to allow I<CHECK> to be a code reference.
490 This is not yet implemented as there are design issues with what its arguments
491 should be and how it returns its results.
497 Passed remaining fragment of string being processed.
498 Modifies it in place to remove bytes/characters it can understand
499 and returns a string used to represent them.
503 my $ch = substr($_[0],0,1,'');
504 return sprintf("\x{%02X}",ord($ch);
507 This scheme is close to how underlying C code for Encode works, but gives
508 the fixup routine very little context.
512 Passed original string, and an index into it of the problem area,
513 and output string so far.
514 Appends what it will to output string and returns new index into
519 # my ($s,$i,$d) = @_;
520 my $ch = substr($_[0],$_[1],1);
521 $_[2] .= sprintf("\x{%02X}",ord($ch);
525 This scheme gives maximal control to the fixup routine but is more complicated
526 to code, and may need internals of Encode to be tweaked to keep original
533 Multiple return values rather than in-place modifications.
535 Index into the string could be pos($str) allowing s/\G...//.
541 The Unicode consortium defines the UTF-8 standard as a way of encoding
542 the entire Unicode repertiore as sequences of octets. This encoding
543 is expected to become very widespread. Perl can use this form internaly
544 to represent strings, so conversions to and from this form are particularly
545 efficient (as octets in memory do not have to change, just the meta-data
546 that tells perl how to treat them).
552 $bytes = encode_utf8($string);
554 The characters that comprise string are encoded in perl's superset of UTF-8
555 and the resulting octets returned as a sequence of bytes. All possible
556 characters have a UTF-8 representation so this function cannot fail.
560 $string = decode_utf8($bytes [,CHECK]);
562 The sequence of octets represented by $bytes is decoded from UTF-8 into
563 a sequence of logical characters. Not all sequences of octets form valid
564 UTF-8 encodings, so it is possible for this call to fail.
565 See L</"Handling Malformed Data">.
569 =head2 Other Encodings of Unicode
571 UTF-16 is similar to UCS-2, 16 bit or 2-byte chunks.
572 UCS-2 can only represent 0..0xFFFF, while UTF-16 has a "surogate pair"
573 scheme which allows it to cover the whole Unicode range.
575 Encode implements big-endian UCS-2 aliased to "iso10646-1" as that
576 happens to be the name used by that representation when used with X11 fonts.
578 UTF-32 or UCS-4 is 32-bit or 4-byte chunks. Perl's logical characters
579 can be considered as being in this form without encoding. An encoding
580 to transfer strings in this form (e.g. to write them to a file) would need to
582 pack('L',map(chr($_),split(//,$string))); # native
584 pack('V',map(chr($_),split(//,$string))); # little-endian
586 pack('N',map(chr($_),split(//,$string))); # big-endian
588 depending on the endian required.
590 No UTF-32 encodings are implemented yet.
592 Both UCS-2 and UCS-4 style encodings can have "byte order marks" by representing
593 the code point 0xFFFE as the very first thing in a file.
595 =head2 Listing available encodings
597 use Encode qw(encodings);
600 Returns a list of the canonical names of the available encodings.
602 =head2 Defining Aliases
604 use Encode qw(define_alias);
605 define_alias( newName => ENCODING);
607 Allows newName to be used as am alias for ENCODING. ENCODING may be either the
608 name of an encoding or and encoding object (as above).
610 Currently I<newName> can be specified in the following ways:
614 =item As a simple string.
616 =item As a qr// compiled regular expression, e.g.:
618 define_alias( qr/^iso8859-(\d+)$/i => '"iso-8859-$1"' );
620 In this case if I<ENCODING> is not a reference it is C<eval>-ed to allow
621 C<$1> etc. to be subsituted.
622 The example is one way to names as used in X11 font names to alias the MIME names for the
625 =item As a code reference, e.g.:
627 define_alias( sub { return /^iso8859-(\d+)$/i ? "iso-8859-$1" : undef } , '');
629 In this case C<$_> will be set to the name that is being looked up and
630 I<ENCODING> is passed to the sub as its first argument.
631 The example is another way to names as used in X11 font names to alias the MIME names for
632 the iso-8859-* family.
636 =head2 Defining Encodings
638 use Encode qw(define_alias);
639 define_encoding( $object, 'canonicalName' [,alias...]);
641 Causes I<canonicalName> to be associated with I<$object>.
642 The object should provide the interface described in L</"IMPLEMENTATION CLASSES"> below.
643 If more than two arguments are provided then additional arguments are taken
644 as aliases for I<$object> as for C<define_alias>.
646 =head1 Encoding and IO
648 It is very common to want to do encoding transformations when
649 reading or writing files, network connections, pipes etc.
650 If perl is configured to use the new 'perlio' IO system then
651 C<Encode> provides a "layer" (See L<perliol>) which can transform
652 data as it is read or written.
654 open(my $ilyad,'>:encoding(iso-8859-7)','ilyad.greek');
657 In addition the new IO system can also be configured to read/write
658 UTF-8 encoded characters (as noted above this is efficient):
660 open(my $fh,'>:utf8','anything');
661 print $fh "Any \x{0021} string \N{SMILEY FACE}\n";
663 Either of the above forms of "layer" specifications can be made the default
664 for a lexical scope with the C<use open ...> pragma. See L<open>.
666 Once a handle is open is layers can be altered using C<binmode>.
668 Without any such configuration, or if perl itself is built using
669 system's own IO, then write operations assume that file handle accepts
670 only I<bytes> and will C<die> if a character larger than 255 is
671 written to the handle. When reading, each octet from the handle
672 becomes a byte-in-a-character. Note that this default is the same
673 behaviour as bytes-only languages (including perl before v5.6) would have,
674 and is sufficient to handle native 8-bit encodings e.g. iso-8859-1,
675 EBCDIC etc. and any legacy mechanisms for handling other encodings
678 In other cases it is the programs responsibility
679 to transform characters into bytes using the API above before
680 doing writes, and to transform the bytes read from a handle into characters
681 before doing "character operations" (e.g. C<lc>, C</\W+/>, ...).
683 =head1 Encoding How to ...
689 =item * IO with mixed content (faking iso-2020-*)
691 =item * MIME's Content-Length:
693 =item * UTF-8 strings in binary data.
695 =item * perl/Encode wrappers on non-Unicode XS modules.
699 =head1 Messing with Perl's Internals
701 The following API uses parts of perl's internals in the current implementation.
702 As such they are efficient, but may change.
708 $num_octets = utf8_upgrade($string);
710 Converts internal representation of string to the UTF-8 form.
711 Returns the number of octets necessary to represent the string as UTF-8.
713 =item * utf8_downgrade($string[, CHECK])
715 Converts internal representation of string to be un-encoded bytes.
717 =item * is_utf8(STRING [, CHECK])
719 [INTERNAL] Test whether the UTF-8 flag is turned on in the STRING.
720 If CHECK is true, also checks the data in STRING for being
721 well-formed UTF-8. Returns true if successful, false otherwise.
723 =item * valid_utf8(STRING)
725 [INTERNAL] Test whether STRING is in a consistent state.
726 Will return true if string is held as bytes, or is well-formed UTF-8
727 and has the UTF-8 flag on.
728 Main reason for this routine is to allow perl's testsuite to check
729 that operations have left strings in a consistent state.
735 [INTERNAL] Turn on the UTF-8 flag in STRING. The data in STRING is
736 B<not> checked for being well-formed UTF-8. Do not use unless you
737 B<know> that the STRING is well-formed UTF-8. Returns the previous
738 state of the UTF-8 flag (so please don't test the return value as
739 I<not> success or failure), or C<undef> if STRING is not a string.
745 [INTERNAL] Turn off the UTF-8 flag in STRING. Do not use frivolously.
746 Returns the previous state of the UTF-8 flag (so please don't test the
747 return value as I<not> success or failure), or C<undef> if STRING is
752 =head1 IMPLEMENTATION CLASSES
754 As mentioned above encodings are (in the current implementation at least)
755 defined by objects. The mapping of encoding name to object is via the
758 The values of the hash can currently be either strings or objects.
759 The string form may go away in the future. The string form occurs
760 when C<encodings()> has scanned C<@INC> for loadable encodings but has
761 not actually loaded the encoding in question. This is because the
762 current "loading" process is all perl and a bit slow.
764 Once an encoding is loaded then value of the hash is object which implements
765 the encoding. The object should provide the following interface:
771 Should return the string representing the canonical name of the encoding.
773 =item -E<gt>new_sequence
775 This is a placeholder for encodings with state. It should return an object
776 which implements this interface, all current implementations return the
779 =item -E<gt>encode($string,$check)
781 Should return the octet sequence representing I<$string>. If I<$check> is true
782 it should modify I<$string> in place to remove the converted part (i.e.
783 the whole string unless there is an error).
784 If an error occurs it should return the octet sequence for the
785 fragment of string that has been converted, and modify $string in-place
786 to remove the converted part leaving it starting with the problem fragment.
788 If check is is false then C<encode> should make a "best effort" to convert
789 the string - for example by using a replacement character.
791 =item -E<gt>decode($octets,$check)
793 Should return the string that I<$octets> represents. If I<$check> is true
794 it should modify I<$octets> in place to remove the converted part (i.e.
795 the whole sequence unless there is an error).
796 If an error occurs it should return the fragment of string
797 that has been converted, and modify $octets in-place to remove the converted part
798 leaving it starting with the problem fragment.
800 If check is is false then C<decode> should make a "best effort" to convert
801 the string - for example by using Unicode's "\x{FFFD}" as a replacement character.
805 It should be noted that the check behaviour is different from the outer
806 public API. The logic is that the "unchecked" case is useful when
807 encoding is part of a stream which may be reporting errors (e.g. STDERR).
808 In such cases it is desirable to get everything through somehow without
809 causing additional errors which obscure the original one. Also the encoding
810 is best placed to know what the correct replacement character is, so if that
811 is the desired behaviour then letting low level code do it is the most efficient.
813 In contrast if check is true, the scheme above allows the encoding to do as
814 much as it can and tell layer above how much that was. What is lacking
815 at present is a mechanism to report what went wrong. The most likely interface
816 will be an additional method call to the object, or perhaps
817 (to avoid forcing per-stream objects on otherwise stateless encodings)
818 and additional parameter.
820 It is also highly desirable that encoding classes inherit from C<Encode::Encoding>
821 as a base class. This allows that class to define additional behaviour for
822 all encoding objects. For example built in Unicode, UCS-2 and UTF-8 classes
825 package Encode::MyEncoding;
826 use base qw(Encode::Encoding);
828 __PACKAGE__->Define(qw(myCanonical myAlias));
830 To create an object with bless {Name => ...},$class, and call define_encoding.
831 They inherit their C<name> method from C<Encode::Encoding>.
833 =head2 Compiled Encodings
835 F<Encode.xs> provides a class C<Encode::XS> which provides the interface described
836 above. It calls a generic octet-sequence to octet-sequence "engine" that is
837 driven by tables (defined in F<encengine.c>). The same engine is used for both
838 encode and decode. C<Encode:XS>'s C<encode> forces perl's characters to their UTF-8 form
839 and then treats them as just another multibyte encoding. C<Encode:XS>'s C<decode> transforms
840 the sequence and then turns the UTF-8-ness flag as that is the form that the tables
841 are defined to produce. For details of the engine see the comments in F<encengine.c>.
843 The tables are produced by the perl script F<compile> (the name needs to change so
844 we can eventually install it somewhere). F<compile> can currently read two formats:
850 This is a coined format used by Tcl. It is documented in Encode/EncodeFormat.pod.
854 This is the semi-standard format used by IBM's ICU package.
858 F<compile> can write the following forms:
864 See above - the F<Encode/*.ucm> files provided with the distribution have
865 been created from the original Tcl .enc files using this approach.
869 Produces tables as C data structures - this is used to build in encodings
870 into F<Encode.so>/F<Encode.dll>.
874 In theory this allows encodings to be stand-alone loadable perl extensions.
875 The process has not yet been tested. The plan is to use this approach
876 for large East Asian encodings.
880 The set of encodings built-in to F<Encode.so>/F<Encode.dll> is determined by
881 F<Makefile.PL>. The current set is as follows:
885 =item ascii and iso-8859-*
887 That is all the common 8-bit "western" encodings.
889 =item IBM-1047 and two other variants of EBCDIC.
891 These are the same variants that are supported by EBCDIC perl as "native" encodings.
892 They are included to prove "reversibility" of some constructs in EBCDIC perl.
894 =item symbol and dingbats as used by Tk on X11.
896 (The reason Encode got started was to support perl/Tk.)
900 That set is rather ad. hoc. and has been driven by the needs of the tests rather
901 than the needs of typical applications. It is likely to be rationalized.
905 L<perlunicode>, L<perlebcdic>, L<perlfunc/open>