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 => $name);
113 define_alias( qr/^iso[-_]?(\d+)[-_](\d+)$/i => '"iso-$1-$2"' );
114 define_alias( qr/^(?:iso[-_]?)?latin[-_]?(\d+)$/i =>
115 '"iso-8859-$isolatin2num{$1}"' );
116 define_alias( qr/^(\S+)[\s_]+(.*)$/i => '"$1-$2"' );
117 #define_alias( sub { return /^iso-(\d+-\d+)$/i ? "iso$1" : '' } );
118 define_alias( 'ascii' => 'US-ascii');
119 define_alias( 'ibm-1047' => 'cp1047');
125 $encoding{$name} = $obj;
127 define_alias($lc => $obj) unless $lc eq $name;
131 define_alias($alias,$obj);
138 my ($class,$name) = @_;
140 if (exists $encoding{$name})
142 return $encoding{$name};
146 return $class->findAlias($name);
153 return __PACKAGE__->getEncoding($name);
158 my ($name,$string,$check) = @_;
159 my $enc = find_encoding($name);
160 croak("Unknown encoding '$name'") unless defined $enc;
161 my $octets = $enc->encode($string,$check);
162 return undef if ($check && length($string));
168 my ($name,$octets,$check) = @_;
169 my $enc = find_encoding($name);
170 croak("Unknown encoding '$name'") unless defined $enc;
171 my $string = $enc->decode($octets,$check);
172 return undef if ($check && length($octets));
178 my ($string,$from,$to,$check) = @_;
179 my $f = find_encoding($from);
180 croak("Unknown encoding '$from'") unless defined $f;
181 my $t = find_encoding($to);
182 croak("Unknown encoding '$to'") unless defined $t;
183 my $uni = $f->decode($string,$check);
184 return undef if ($check && length($string));
185 $string = $t->encode($uni,$check);
186 return undef if ($check && length($uni));
187 return length($_[0] = $string);
200 return undef unless utf8_decode($str);
204 package Encode::Encoding;
205 # Base class for classes which implement encodings
210 my $canonical = shift;
211 $obj = bless { Name => $canonical },$obj unless ref $obj;
212 # warn "$canonical => $obj\n";
213 Encode::define_encoding($obj, $canonical, @_);
216 sub name { shift->{'Name'} }
218 # Temporary legacy methods
219 sub toUnicode { shift->decode(@_) }
220 sub fromUnicode { shift->encode(@_) }
222 sub new_sequence { return $_[0] }
225 use base 'Encode::Encoding';
227 package Encode::Unicode;
228 use base 'Encode::Encoding';
230 # Dummy package that provides the encode interface but leaves data
231 # as UTF-8 encoded. It is here so that from_to() works.
233 __PACKAGE__->Define('Unicode');
237 my ($obj,$str,$chk) = @_;
238 Encode::utf8_upgrade($str);
245 package Encode::utf8;
246 use base 'Encode::Encoding';
247 # package to allow long-hand
248 # $octets = encode( utf8 => $string );
251 __PACKAGE__->Define(qw(UTF-8 utf8));
255 my ($obj,$octets,$chk) = @_;
256 my $str = Encode::decode_utf8($octets);
267 my ($obj,$string,$chk) = @_;
268 my $octets = Encode::encode_utf8($string);
273 package Encode::iso10646_1;
274 use base 'Encode::Encoding';
275 # Encoding is 16-bit network order Unicode (no surogates)
276 # Used for X font encodings
278 __PACKAGE__->Define(qw(UCS-2 iso10646-1));
282 my ($obj,$str,$chk) = @_;
286 my $code = unpack('n',substr($str,0,2,'')) & 0xffff;
289 $_[1] = $str if $chk;
290 Encode::utf8_upgrade($uni);
296 my ($obj,$uni,$chk) = @_;
300 my $ch = substr($uni,0,1,'');
307 $str .= pack('n',$x);
309 $_[1] = $uni if $chk;
313 # switch back to Encode package in case we ever add AutoLoader
322 Encode - character encodings
330 The C<Encode> module provides the interfaces between perl's strings
331 and the rest of the system. Perl strings are sequences of B<characters>.
333 The repertoire of characters that Perl can represent is at least that
334 defined by the Unicode Consortium. On most platforms the ordinal values
335 of the characters (as returned by C<ord(ch)>) is the "Unicode codepoint" for
336 the character (the exceptions are those platforms where the legacy
337 encoding is some variant of EBCDIC rather than a super-set of ASCII
338 - see L<perlebcdic>).
340 Traditionaly computer data has been moved around in 8-bit chunks
341 often called "bytes". These chunks are also known as "octets" in
342 networking standards. Perl is widely used to manipulate data of
343 many types - not only strings of characters representing human or
344 computer languages but also "binary" data being the machines representation
345 of numbers, pixels in an image - or just about anything.
347 When perl is processing "binary data" the programmer wants perl to process
348 "sequences of bytes". This is not a problem for perl - as a byte has 256
349 possible values it easily fits in perl's much larger "logical character".
357 I<character>: a character in the range 0..(2**32-1) (or more).
358 (What perl's strings are made of.)
362 I<byte>: a character in the range 0..255
363 (A special case of a perl character.)
367 I<octet>: 8 bits of data, with ordinal values 0..255
368 (Term for bytes passed to or from a non-perl context, e.g. disk file.)
372 The marker [INTERNAL] marks Internal Implementation Details, in
373 general meant only for those who think they know what they are doing,
374 and such details may change in future releases.
378 =head2 Characteristics of an Encoding
380 An encoding has a "repertoire" of characters that it can represent,
381 and for each representable character there is at least one sequence of
382 octets that represents it.
384 =head2 Types of Encodings
386 Encodings can be divided into the following types:
390 =item * Fixed length 8-bit (or less) encodings.
392 Each character is a single octet so may have a repertoire of up to
393 256 characters. ASCII and iso-8859-* are typical examples.
395 =item * Fixed length 16-bit encodings
397 Each character is two octets so may have a repertoire of up to
398 65,536 characters. Unicode's UCS-2 is an example. Also used for
399 encodings for East Asian languages.
401 =item * Fixed length 32-bit encodings.
403 Not really very "encoded" encodings. The Unicode code points
404 are just represented as 4-octet integers. None the less because
405 different architectures use different representations of integers
406 (so called "endian") there at least two disctinct encodings.
408 =item * Multi-byte encodings
410 The number of octets needed to represent a character varies.
411 UTF-8 is a particularly complex but regular case of a multi-byte
412 encoding. Several East Asian countries use a multi-byte encoding
413 where 1-octet is used to cover western roman characters and Asian
414 characters get 2-octets.
415 (UTF-16 is strictly a multi-byte encoding taking either 2 or 4 octets
416 to represent a Unicode code point.)
418 =item * "Escape" encodings.
420 These encodings embed "escape sequences" into the octet sequence
421 which describe how the following octets are to be interpreted.
422 The iso-2022-* family is typical. Following the escape sequence
423 octets are encoded by an "embedded" encoding (which will be one
424 of the above types) until another escape sequence switches to
425 a different "embedded" encoding.
427 These schemes are very flexible and can handle mixed languages but are
428 very complex to process (and have state).
429 No escape encodings are implemented for perl yet.
433 =head2 Specifying Encodings
435 Encodings can be specified to the API described below in two ways:
441 Encoding names are strings with characters taken from a restricted repertoire.
442 See L</"Encoding Names">.
444 =item 2. As an object
446 Encoding objects are returned by C<find_encoding($name)>.
450 =head2 Encoding Names
452 Encoding names are case insensitive. White space in names is ignored.
453 In addition an encoding may have aliases. Each encoding has one "canonical" name.
454 The "canonical" name is chosen from the names of the encoding by picking
455 the first in the following sequence:
459 =item * The MIME name as defined in IETF RFC-XXXX.
461 =item * The name in the IANA registry.
463 =item * The name used by the the organization that defined it.
467 Because of all the alias issues, and because in the general case
468 encodings have state C<Encode> uses the encoding object internally
469 once an operation is in progress.
471 =head1 PERL ENCODING API
473 =head2 Generic Encoding Interface
479 $bytes = encode(ENCODING, $string[, CHECK])
481 Encodes string from perl's internal form into I<ENCODING> and returns a
483 See L</"Handling Malformed Data">.
487 $string = decode(ENCODING, $bytes[, CHECK])
489 Decode sequence of octets assumed to be in I<ENCODING> into perls internal
490 form and returns the resuting string.
491 See L</"Handling Malformed Data">.
495 =head2 Handling Malformed Data
497 If CHECK is not set, C<undef> is returned. If the data is supposed to
498 be UTF-8, an optional lexical warning (category utf8) is given.
499 If CHECK is true but not a code reference, dies.
501 It would desirable to have a way to indicate that transform should use the
502 encodings "replacement character" - no such mechanism is defined yet.
504 It is also planned to allow I<CHECK> to be a code reference.
506 This is not yet implemented as there are design issues with what its arguments
507 should be and how it returns its results.
513 Passed remaining fragment of string being processed.
514 Modifies it in place to remove bytes/characters it can understand
515 and returns a string used to represent them.
519 my $ch = substr($_[0],0,1,'');
520 return sprintf("\x{%02X}",ord($ch);
523 This scheme is close to how underlying C code for Encode works, but gives
524 the fixup routine very little context.
528 Passed original string, and an index into it of the problem area,
529 and output string so far.
530 Appends what it will to output string and returns new index into
535 # my ($s,$i,$d) = @_;
536 my $ch = substr($_[0],$_[1],1);
537 $_[2] .= sprintf("\x{%02X}",ord($ch);
541 This scheme gives maximal control to the fixup routine but is more complicated
542 to code, and may need internals of Encode to be tweaked to keep original
549 Multiple return values rather than in-place modifications.
551 Index into the string could be pos($str) allowing s/\G...//.
557 The Unicode consortium defines the UTF-8 standard as a way of encoding
558 the entire Unicode repertiore as sequences of octets. This encoding
559 is expected to become very widespread. Perl can use this form internaly
560 to represent strings, so conversions to and from this form are particularly
561 efficient (as octets in memory do not have to change, just the meta-data
562 that tells perl how to treat them).
568 $bytes = encode_utf8($string);
570 The characters that comprise string are encoded in perl's superset of UTF-8
571 and the resulting octets returned as a sequence of bytes. All possible
572 characters have a UTF-8 representation so this function cannot fail.
576 $string = decode_utf8($bytes [,CHECK]);
578 The sequence of octets represented by $bytes is decoded from UTF-8 into
579 a sequence of logical characters. Not all sequences of octets form valid
580 UTF-8 encodings, so it is possible for this call to fail.
581 See L</"Handling Malformed Data">.
585 =head2 Other Encodings of Unicode
587 UTF-16 is similar to UCS-2, 16 bit or 2-byte chunks.
588 UCS-2 can only represent 0..0xFFFF, while UTF-16 has a "surogate pair"
589 scheme which allows it to cover the whole Unicode range.
591 Encode implements big-endian UCS-2 aliased to "iso10646-1" as that
592 happens to be the name used by that representation when used with X11 fonts.
594 UTF-32 or UCS-4 is 32-bit or 4-byte chunks. Perl's logical characters
595 can be considered as being in this form without encoding. An encoding
596 to transfer strings in this form (e.g. to write them to a file) would need to
598 pack('L',map(chr($_),split(//,$string))); # native
600 pack('V',map(chr($_),split(//,$string))); # little-endian
602 pack('N',map(chr($_),split(//,$string))); # big-endian
604 depending on the endian required.
606 No UTF-32 encodings are implemented yet.
608 Both UCS-2 and UCS-4 style encodings can have "byte order marks" by representing
609 the code point 0xFFFE as the very first thing in a file.
611 =head2 Listing available encodings
613 use Encode qw(encodings);
616 Returns a list of the canonical names of the available encodings.
618 =head2 Defining Aliases
620 use Encode qw(define_alias);
621 define_alias( newName => ENCODING);
623 Allows newName to be used as am alias for ENCODING. ENCODING may be either the
624 name of an encoding or and encoding object (as above).
626 Currently I<newName> can be specified in the following ways:
630 =item As a simple string.
632 =item As a qr// compiled regular expression, e.g.:
634 define_alias( qr/^iso8859-(\d+)$/i => '"iso-8859-$1"' );
636 In this case if I<ENCODING> is not a reference it is C<eval>-ed to allow
637 C<$1> etc. to be subsituted.
638 The example is one way to names as used in X11 font names to alias the MIME names for the
641 =item As a code reference, e.g.:
643 define_alias( sub { return /^iso8859-(\d+)$/i ? "iso-8859-$1" : undef } , '');
645 In this case C<$_> will be set to the name that is being looked up and
646 I<ENCODING> is passed to the sub as its first argument.
647 The example is another way to names as used in X11 font names to alias the MIME names for
648 the iso-8859-* family.
652 =head2 Defining Encodings
654 use Encode qw(define_alias);
655 define_encoding( $object, 'canonicalName' [,alias...]);
657 Causes I<canonicalName> to be associated with I<$object>.
658 The object should provide the interface described in L</"IMPLEMENTATION CLASSES"> below.
659 If more than two arguments are provided then additional arguments are taken
660 as aliases for I<$object> as for C<define_alias>.
662 =head1 Encoding and IO
664 It is very common to want to do encoding transformations when
665 reading or writing files, network connections, pipes etc.
666 If perl is configured to use the new 'perlio' IO system then
667 C<Encode> provides a "layer" (See L<perliol>) which can transform
668 data as it is read or written.
670 open(my $ilyad,'>:encoding(iso-8859-7)','ilyad.greek');
673 In addition the new IO system can also be configured to read/write
674 UTF-8 encoded characters (as noted above this is efficient):
676 open(my $fh,'>:utf8','anything');
677 print $fh "Any \x{0021} string \N{SMILEY FACE}\n";
679 Either of the above forms of "layer" specifications can be made the default
680 for a lexical scope with the C<use open ...> pragma. See L<open>.
682 Once a handle is open is layers can be altered using C<binmode>.
684 Without any such configuration, or if perl itself is built using
685 system's own IO, then write operations assume that file handle accepts
686 only I<bytes> and will C<die> if a character larger than 255 is
687 written to the handle. When reading, each octet from the handle
688 becomes a byte-in-a-character. Note that this default is the same
689 behaviour as bytes-only languages (including perl before v5.6) would have,
690 and is sufficient to handle native 8-bit encodings e.g. iso-8859-1,
691 EBCDIC etc. and any legacy mechanisms for handling other encodings
694 In other cases it is the programs responsibility
695 to transform characters into bytes using the API above before
696 doing writes, and to transform the bytes read from a handle into characters
697 before doing "character operations" (e.g. C<lc>, C</\W+/>, ...).
699 =head1 Encoding How to ...
705 =item * IO with mixed content (faking iso-2020-*)
707 =item * MIME's Content-Length:
709 =item * UTF-8 strings in binary data.
711 =item * perl/Encode wrappers on non-Unicode XS modules.
715 =head1 Messing with Perl's Internals
717 The following API uses parts of perl's internals in the current implementation.
718 As such they are efficient, but may change.
724 $num_octets = utf8_upgrade($string);
726 Converts internal representation of string to the UTF-8 form.
727 Returns the number of octets necessary to represent the string as UTF-8.
729 =item * utf8_downgrade($string[, CHECK])
731 Converts internal representation of string to be un-encoded bytes.
733 =item * is_utf8(STRING [, CHECK])
735 [INTERNAL] Test whether the UTF-8 flag is turned on in the STRING.
736 If CHECK is true, also checks the data in STRING for being
737 well-formed UTF-8. Returns true if successful, false otherwise.
739 =item * valid_utf8(STRING)
741 [INTERNAL] Test whether STRING is in a consistent state.
742 Will return true if string is held as bytes, or is well-formed UTF-8
743 and has the UTF-8 flag on.
744 Main reason for this routine is to allow perl's testsuite to check
745 that operations have left strings in a consistent state.
751 [INTERNAL] Turn on the UTF-8 flag in STRING. The data in STRING is
752 B<not> checked for being well-formed UTF-8. Do not use unless you
753 B<know> that the STRING is well-formed UTF-8. Returns the previous
754 state of the UTF-8 flag (so please don't test the return value as
755 I<not> success or failure), or C<undef> if STRING is not a string.
761 [INTERNAL] Turn off the UTF-8 flag in STRING. Do not use frivolously.
762 Returns the previous state of the UTF-8 flag (so please don't test the
763 return value as I<not> success or failure), or C<undef> if STRING is
768 =head1 IMPLEMENTATION CLASSES
770 As mentioned above encodings are (in the current implementation at least)
771 defined by objects. The mapping of encoding name to object is via the
774 The values of the hash can currently be either strings or objects.
775 The string form may go away in the future. The string form occurs
776 when C<encodings()> has scanned C<@INC> for loadable encodings but has
777 not actually loaded the encoding in question. This is because the
778 current "loading" process is all perl and a bit slow.
780 Once an encoding is loaded then value of the hash is object which implements
781 the encoding. The object should provide the following interface:
787 Should return the string representing the canonical name of the encoding.
789 =item -E<gt>new_sequence
791 This is a placeholder for encodings with state. It should return an object
792 which implements this interface, all current implementations return the
795 =item -E<gt>encode($string,$check)
797 Should return the octet sequence representing I<$string>. If I<$check> is true
798 it should modify I<$string> in place to remove the converted part (i.e.
799 the whole string unless there is an error).
800 If an error occurs it should return the octet sequence for the
801 fragment of string that has been converted, and modify $string in-place
802 to remove the converted part leaving it starting with the problem fragment.
804 If check is is false then C<encode> should make a "best effort" to convert
805 the string - for example by using a replacement character.
807 =item -E<gt>decode($octets,$check)
809 Should return the string that I<$octets> represents. If I<$check> is true
810 it should modify I<$octets> in place to remove the converted part (i.e.
811 the whole sequence unless there is an error).
812 If an error occurs it should return the fragment of string
813 that has been converted, and modify $octets in-place to remove the converted part
814 leaving it starting with the problem fragment.
816 If check is is false then C<decode> should make a "best effort" to convert
817 the string - for example by using Unicode's "\x{FFFD}" as a replacement character.
821 It should be noted that the check behaviour is different from the outer
822 public API. The logic is that the "unchecked" case is useful when
823 encoding is part of a stream which may be reporting errors (e.g. STDERR).
824 In such cases it is desirable to get everything through somehow without
825 causing additional errors which obscure the original one. Also the encoding
826 is best placed to know what the correct replacement character is, so if that
827 is the desired behaviour then letting low level code do it is the most efficient.
829 In contrast if check is true, the scheme above allows the encoding to do as
830 much as it can and tell layer above how much that was. What is lacking
831 at present is a mechanism to report what went wrong. The most likely interface
832 will be an additional method call to the object, or perhaps
833 (to avoid forcing per-stream objects on otherwise stateless encodings)
834 and additional parameter.
836 It is also highly desirable that encoding classes inherit from C<Encode::Encoding>
837 as a base class. This allows that class to define additional behaviour for
838 all encoding objects. For example built in Unicode, UCS-2 and UTF-8 classes
841 package Encode::MyEncoding;
842 use base qw(Encode::Encoding);
844 __PACKAGE__->Define(qw(myCanonical myAlias));
846 To create an object with bless {Name => ...},$class, and call define_encoding.
847 They inherit their C<name> method from C<Encode::Encoding>.
849 =head2 Compiled Encodings
851 F<Encode.xs> provides a class C<Encode::XS> which provides the interface described
852 above. It calls a generic octet-sequence to octet-sequence "engine" that is
853 driven by tables (defined in F<encengine.c>). The same engine is used for both
854 encode and decode. C<Encode:XS>'s C<encode> forces perl's characters to their UTF-8 form
855 and then treats them as just another multibyte encoding. C<Encode:XS>'s C<decode> transforms
856 the sequence and then turns the UTF-8-ness flag as that is the form that the tables
857 are defined to produce. For details of the engine see the comments in F<encengine.c>.
859 The tables are produced by the perl script F<compile> (the name needs to change so
860 we can eventually install it somewhere). F<compile> can currently read two formats:
866 This is a coined format used by Tcl. It is documented in Encode/EncodeFormat.pod.
870 This is the semi-standard format used by IBM's ICU package.
874 F<compile> can write the following forms:
880 See above - the F<Encode/*.ucm> files provided with the distribution have
881 been created from the original Tcl .enc files using this approach.
885 Produces tables as C data structures - this is used to build in encodings
886 into F<Encode.so>/F<Encode.dll>.
890 In theory this allows encodings to be stand-alone loadable perl extensions.
891 The process has not yet been tested. The plan is to use this approach
892 for large East Asian encodings.
896 The set of encodings built-in to F<Encode.so>/F<Encode.dll> is determined by
897 F<Makefile.PL>. The current set is as follows:
901 =item ascii and iso-8859-*
903 That is all the common 8-bit "western" encodings.
905 =item IBM-1047 and two other variants of EBCDIC.
907 These are the same variants that are supported by EBCDIC perl as "native" encodings.
908 They are included to prove "reversibility" of some constructs in EBCDIC perl.
910 =item symbol and dingbats as used by Tk on X11.
912 (The reason Encode got started was to support perl/Tk.)
916 That set is rather ad. hoc. and has been driven by the needs of the tests rather
917 than the needs of typical applications. It is likely to be rationalized.
921 L<perlunicode>, L<perlebcdic>, L<perlfunc/open>