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 => $name);
99 define_alias( qr/^iso\W?(\d+)\W(\d+)$/i => '"iso-$1-$2"' );
100 define_alias( qr/^(?:iso\W?)?(?:8859|latin)\W(\d+)$/i => '"iso-8859-$1"' );
101 define_alias( qr/^(\S+)\s+(.*)$/i => '"$1-$2"' );
102 #define_alias( sub { return /^iso-(\d+-\d+)$/i ? "iso$1" : '' } );
103 define_alias( 'ascii' => 'US-ascii');
104 define_alias( 'ibm-1047' => 'cp1047');
110 $encoding{$name} = $obj;
112 define_alias($lc => $obj) unless $lc eq $name;
116 define_alias($alias,$obj);
123 my ($class,$name) = @_;
125 if (exists $encoding{$name})
127 return $encoding{$name};
131 return $class->findAlias($name);
138 return __PACKAGE__->getEncoding($name);
143 my ($name,$string,$check) = @_;
144 my $enc = find_encoding($name);
145 croak("Unknown encoding '$name'") unless defined $enc;
146 my $octets = $enc->encode($string,$check);
147 return undef if ($check && length($string));
153 my ($name,$octets,$check) = @_;
154 my $enc = find_encoding($name);
155 croak("Unknown encoding '$name'") unless defined $enc;
156 my $string = $enc->decode($octets,$check);
157 return undef if ($check && length($octets));
163 my ($string,$from,$to,$check) = @_;
164 my $f = find_encoding($from);
165 croak("Unknown encoding '$from'") unless defined $f;
166 my $t = find_encoding($to);
167 croak("Unknown encoding '$to'") unless defined $t;
168 my $uni = $f->decode($string,$check);
169 return undef if ($check && length($string));
170 $string = $t->encode($uni,$check);
171 return undef if ($check && length($uni));
172 return length($_[0] = $string);
185 return undef unless utf8_decode($str);
189 package Encode::Encoding;
190 # Base class for classes which implement encodings
195 my $canonical = shift;
196 $obj = bless { Name => $canonical },$obj unless ref $obj;
197 # warn "$canonical => $obj\n";
198 Encode::define_encoding($obj, $canonical, @_);
201 sub name { shift->{'Name'} }
203 # Temporary legacy methods
204 sub toUnicode { shift->decode(@_) }
205 sub fromUnicode { shift->encode(@_) }
207 sub new_sequence { return $_[0] }
210 use base 'Encode::Encoding';
212 package Encode::Unicode;
213 use base 'Encode::Encoding';
215 # Dummy package that provides the encode interface but leaves data
216 # as UTF-8 encoded. It is here so that from_to() works.
218 __PACKAGE__->Define('Unicode');
222 my ($obj,$str,$chk) = @_;
223 Encode::utf8_upgrade($str);
230 package Encode::utf8;
231 use base 'Encode::Encoding';
232 # package to allow long-hand
233 # $octets = encode( utf8 => $string );
236 __PACKAGE__->Define(qw(UTF-8 utf8));
240 my ($obj,$octets,$chk) = @_;
241 my $str = Encode::decode_utf8($octets);
252 my ($obj,$string,$chk) = @_;
253 my $octets = Encode::encode_utf8($string);
258 package Encode::iso10646_1;
259 use base 'Encode::Encoding';
260 # Encoding is 16-bit network order Unicode (no surogates)
261 # Used for X font encodings
263 __PACKAGE__->Define(qw(UCS-2 iso10646-1));
267 my ($obj,$str,$chk) = @_;
271 my $code = unpack('n',substr($str,0,2,'')) & 0xffff;
274 $_[1] = $str if $chk;
275 Encode::utf8_upgrade($uni);
281 my ($obj,$uni,$chk) = @_;
285 my $ch = substr($uni,0,1,'');
292 $str .= pack('n',$x);
294 $_[1] = $uni if $chk;
298 # switch back to Encode package in case we ever add AutoLoader
307 Encode - character encodings
315 The C<Encode> module provides the interfaces between perl's strings
316 and the rest of the system. Perl strings are sequences of B<characters>.
318 The repertoire of characters that Perl can represent is at least that
319 defined by the Unicode Consortium. On most platforms the ordinal values
320 of the characters (as returned by C<ord(ch)>) is the "Unicode codepoint" for
321 the character (the exceptions are those platforms where the legacy
322 encoding is some variant of EBCDIC rather than a super-set of ASCII
323 - see L<perlebcdic>).
325 Traditionaly computer data has been moved around in 8-bit chunks
326 often called "bytes". These chunks are also known as "octets" in
327 networking standards. Perl is widely used to manipulate data of
328 many types - not only strings of characters representing human or
329 computer languages but also "binary" data being the machines representation
330 of numbers, pixels in an image - or just about anything.
332 When perl is processing "binary data" the programmer wants perl to process
333 "sequences of bytes". This is not a problem for perl - as a byte has 256
334 possible values it easily fits in perl's much larger "logical character".
342 I<character>: a character in the range 0..(2**32-1) (or more).
343 (What perl's strings are made of.)
347 I<byte>: a character in the range 0..255
348 (A special case of a perl character.)
352 I<octet>: 8 bits of data, with ordinal values 0..255
353 (Term for bytes passed to or from a non-perl context, e.g. disk file.)
357 The marker [INTERNAL] marks Internal Implementation Details, in
358 general meant only for those who think they know what they are doing,
359 and such details may change in future releases.
363 =head2 Characteristics of an Encoding
365 An encoding has a "repertoire" of characters that it can represent,
366 and for each representable character there is at least one sequence of
367 octets that represents it.
369 =head2 Types of Encodings
371 Encodings can be divided into the following types:
375 =item * Fixed length 8-bit (or less) encodings.
377 Each character is a single octet so may have a repertoire of up to
378 256 characters. ASCII and iso-8859-* are typical examples.
380 =item * Fixed length 16-bit encodings
382 Each character is two octets so may have a repertoire of up to
383 65,536 characters. Unicode's UCS-2 is an example. Also used for
384 encodings for East Asian languages.
386 =item * Fixed length 32-bit encodings.
388 Not really very "encoded" encodings. The Unicode code points
389 are just represented as 4-octet integers. None the less because
390 different architectures use different representations of integers
391 (so called "endian") there at least two disctinct encodings.
393 =item * Multi-byte encodings
395 The number of octets needed to represent a character varies.
396 UTF-8 is a particularly complex but regular case of a multi-byte
397 encoding. Several East Asian countries use a multi-byte encoding
398 where 1-octet is used to cover western roman characters and Asian
399 characters get 2-octets.
400 (UTF-16 is strictly a multi-byte encoding taking either 2 or 4 octets
401 to represent a Unicode code point.)
403 =item * "Escape" encodings.
405 These encodings embed "escape sequences" into the octet sequence
406 which describe how the following octets are to be interpreted.
407 The iso-2022-* family is typical. Following the escape sequence
408 octets are encoded by an "embedded" encoding (which will be one
409 of the above types) until another escape sequence switches to
410 a different "embedded" encoding.
412 These schemes are very flexible and can handle mixed languages but are
413 very complex to process (and have state).
414 No escape encodings are implemented for perl yet.
418 =head2 Specifying Encodings
420 Encodings can be specified to the API described below in two ways:
426 Encoding names are strings with characters taken from a restricted repertoire.
427 See L</"Encoding Names">.
429 =item 2. As an object
431 Encoding objects are returned by C<find_encoding($name)>.
435 =head2 Encoding Names
437 Encoding names are case insensitive. White space in names is ignored.
438 In addition an encoding may have aliases. Each encoding has one "canonical" name.
439 The "canonical" name is chosen from the names of the encoding by picking
440 the first in the following sequence:
444 =item * The MIME name as defined in IETF RFC-XXXX.
446 =item * The name in the IANA registry.
448 =item * The name used by the the organization that defined it.
452 Because of all the alias issues, and because in the general case
453 encodings have state C<Encode> uses the encoding object internally
454 once an operation is in progress.
456 =head1 PERL ENCODING API
458 =head2 Generic Encoding Interface
464 $bytes = encode(ENCODING, $string[, CHECK])
466 Encodes string from perl's internal form into I<ENCODING> and returns a
468 See L</"Handling Malformed Data">.
472 $string = decode(ENCODING, $bytes[, CHECK])
474 Decode sequence of octets assumed to be in I<ENCODING> into perls internal
475 form and returns the resuting string.
476 See L</"Handling Malformed Data">.
480 =head2 Handling Malformed Data
482 If CHECK is not set, C<undef> is returned. If the data is supposed to
483 be UTF-8, an optional lexical warning (category utf8) is given.
484 If CHECK is true but not a code reference, dies.
486 It would desirable to have a way to indicate that transform should use the
487 encodings "replacement character" - no such mechanism is defined yet.
489 It is also planned to allow I<CHECK> to be a code reference.
491 This is not yet implemented as there are design issues with what its arguments
492 should be and how it returns its results.
498 Passed remaining fragment of string being processed.
499 Modifies it in place to remove bytes/characters it can understand
500 and returns a string used to represent them.
504 my $ch = substr($_[0],0,1,'');
505 return sprintf("\x{%02X}",ord($ch);
508 This scheme is close to how underlying C code for Encode works, but gives
509 the fixup routine very little context.
513 Passed original string, and an index into it of the problem area,
514 and output string so far.
515 Appends what it will to output string and returns new index into
520 # my ($s,$i,$d) = @_;
521 my $ch = substr($_[0],$_[1],1);
522 $_[2] .= sprintf("\x{%02X}",ord($ch);
526 This scheme gives maximal control to the fixup routine but is more complicated
527 to code, and may need internals of Encode to be tweaked to keep original
534 Multiple return values rather than in-place modifications.
536 Index into the string could be pos($str) allowing s/\G...//.
542 The Unicode consortium defines the UTF-8 standard as a way of encoding
543 the entire Unicode repertiore as sequences of octets. This encoding
544 is expected to become very widespread. Perl can use this form internaly
545 to represent strings, so conversions to and from this form are particularly
546 efficient (as octets in memory do not have to change, just the meta-data
547 that tells perl how to treat them).
553 $bytes = encode_utf8($string);
555 The characters that comprise string are encoded in perl's superset of UTF-8
556 and the resulting octets returned as a sequence of bytes. All possible
557 characters have a UTF-8 representation so this function cannot fail.
561 $string = decode_utf8($bytes [,CHECK]);
563 The sequence of octets represented by $bytes is decoded from UTF-8 into
564 a sequence of logical characters. Not all sequences of octets form valid
565 UTF-8 encodings, so it is possible for this call to fail.
566 See L</"Handling Malformed Data">.
570 =head2 Other Encodings of Unicode
572 UTF-16 is similar to UCS-2, 16 bit or 2-byte chunks.
573 UCS-2 can only represent 0..0xFFFF, while UTF-16 has a "surogate pair"
574 scheme which allows it to cover the whole Unicode range.
576 Encode implements big-endian UCS-2 aliased to "iso10646-1" as that
577 happens to be the name used by that representation when used with X11 fonts.
579 UTF-32 or UCS-4 is 32-bit or 4-byte chunks. Perl's logical characters
580 can be considered as being in this form without encoding. An encoding
581 to transfer strings in this form (e.g. to write them to a file) would need to
583 pack('L',map(chr($_),split(//,$string))); # native
585 pack('V',map(chr($_),split(//,$string))); # little-endian
587 pack('N',map(chr($_),split(//,$string))); # big-endian
589 depending on the endian required.
591 No UTF-32 encodings are implemented yet.
593 Both UCS-2 and UCS-4 style encodings can have "byte order marks" by representing
594 the code point 0xFFFE as the very first thing in a file.
596 =head2 Listing available encodings
598 use Encode qw(encodings);
601 Returns a list of the canonical names of the available encodings.
603 =head2 Defining Aliases
605 use Encode qw(define_alias);
606 define_alias( newName => ENCODING);
608 Allows newName to be used as am alias for ENCODING. ENCODING may be either the
609 name of an encoding or and encoding object (as above).
611 Currently I<newName> can be specified in the following ways:
615 =item As a simple string.
617 =item As a qr// compiled regular expression, e.g.:
619 define_alias( qr/^iso8859-(\d+)$/i => '"iso-8859-$1"' );
621 In this case if I<ENCODING> is not a reference it is C<eval>-ed to allow
622 C<$1> etc. to be subsituted.
623 The example is one way to names as used in X11 font names to alias the MIME names for the
626 =item As a code reference, e.g.:
628 define_alias( sub { return /^iso8859-(\d+)$/i ? "iso-8859-$1" : undef } , '');
630 In this case C<$_> will be set to the name that is being looked up and
631 I<ENCODING> is passed to the sub as its first argument.
632 The example is another way to names as used in X11 font names to alias the MIME names for
633 the iso-8859-* family.
637 =head2 Defining Encodings
639 use Encode qw(define_alias);
640 define_encoding( $object, 'canonicalName' [,alias...]);
642 Causes I<canonicalName> to be associated with I<$object>.
643 The object should provide the interface described in L</"IMPLEMENTATION CLASSES"> below.
644 If more than two arguments are provided then additional arguments are taken
645 as aliases for I<$object> as for C<define_alias>.
647 =head1 Encoding and IO
649 It is very common to want to do encoding transformations when
650 reading or writing files, network connections, pipes etc.
651 If perl is configured to use the new 'perlio' IO system then
652 C<Encode> provides a "layer" (See L<perliol>) which can transform
653 data as it is read or written.
655 open(my $ilyad,'>:encoding(iso-8859-7)','ilyad.greek');
658 In addition the new IO system can also be configured to read/write
659 UTF-8 encoded characters (as noted above this is efficient):
661 open(my $fh,'>:utf8','anything');
662 print $fh "Any \x{0021} string \N{SMILEY FACE}\n";
664 Either of the above forms of "layer" specifications can be made the default
665 for a lexical scope with the C<use open ...> pragma. See L<open>.
667 Once a handle is open is layers can be altered using C<binmode>.
669 Without any such configuration, or if perl itself is built using
670 system's own IO, then write operations assume that file handle accepts
671 only I<bytes> and will C<die> if a character larger than 255 is
672 written to the handle. When reading, each octet from the handle
673 becomes a byte-in-a-character. Note that this default is the same
674 behaviour as bytes-only languages (including perl before v5.6) would have,
675 and is sufficient to handle native 8-bit encodings e.g. iso-8859-1,
676 EBCDIC etc. and any legacy mechanisms for handling other encodings
679 In other cases it is the programs responsibility
680 to transform characters into bytes using the API above before
681 doing writes, and to transform the bytes read from a handle into characters
682 before doing "character operations" (e.g. C<lc>, C</\W+/>, ...).
684 =head1 Encoding How to ...
690 =item * IO with mixed content (faking iso-2020-*)
692 =item * MIME's Content-Length:
694 =item * UTF-8 strings in binary data.
696 =item * perl/Encode wrappers on non-Unicode XS modules.
700 =head1 Messing with Perl's Internals
702 The following API uses parts of perl's internals in the current implementation.
703 As such they are efficient, but may change.
709 $num_octets = utf8_upgrade($string);
711 Converts internal representation of string to the UTF-8 form.
712 Returns the number of octets necessary to represent the string as UTF-8.
714 =item * utf8_downgrade($string[, CHECK])
716 Converts internal representation of string to be un-encoded bytes.
718 =item * is_utf8(STRING [, CHECK])
720 [INTERNAL] Test whether the UTF-8 flag is turned on in the STRING.
721 If CHECK is true, also checks the data in STRING for being
722 well-formed UTF-8. Returns true if successful, false otherwise.
724 =item * valid_utf8(STRING)
726 [INTERNAL] Test whether STRING is in a consistent state.
727 Will return true if string is held as bytes, or is well-formed UTF-8
728 and has the UTF-8 flag on.
729 Main reason for this routine is to allow perl's testsuite to check
730 that operations have left strings in a consistent state.
736 [INTERNAL] Turn on the UTF-8 flag in STRING. The data in STRING is
737 B<not> checked for being well-formed UTF-8. Do not use unless you
738 B<know> that the STRING is well-formed UTF-8. Returns the previous
739 state of the UTF-8 flag (so please don't test the return value as
740 I<not> success or failure), or C<undef> if STRING is not a string.
746 [INTERNAL] Turn off the UTF-8 flag in STRING. Do not use frivolously.
747 Returns the previous state of the UTF-8 flag (so please don't test the
748 return value as I<not> success or failure), or C<undef> if STRING is
753 =head1 IMPLEMENTATION CLASSES
755 As mentioned above encodings are (in the current implementation at least)
756 defined by objects. The mapping of encoding name to object is via the
759 The values of the hash can currently be either strings or objects.
760 The string form may go away in the future. The string form occurs
761 when C<encodings()> has scanned C<@INC> for loadable encodings but has
762 not actually loaded the encoding in question. This is because the
763 current "loading" process is all perl and a bit slow.
765 Once an encoding is loaded then value of the hash is object which implements
766 the encoding. The object should provide the following interface:
772 Should return the string representing the canonical name of the encoding.
774 =item -E<gt>new_sequence
776 This is a placeholder for encodings with state. It should return an object
777 which implements this interface, all current implementations return the
780 =item -E<gt>encode($string,$check)
782 Should return the octet sequence representing I<$string>. If I<$check> is true
783 it should modify I<$string> in place to remove the converted part (i.e.
784 the whole string unless there is an error).
785 If an error occurs it should return the octet sequence for the
786 fragment of string that has been converted, and modify $string in-place
787 to remove the converted part leaving it starting with the problem fragment.
789 If check is is false then C<encode> should make a "best effort" to convert
790 the string - for example by using a replacement character.
792 =item -E<gt>decode($octets,$check)
794 Should return the string that I<$octets> represents. If I<$check> is true
795 it should modify I<$octets> in place to remove the converted part (i.e.
796 the whole sequence unless there is an error).
797 If an error occurs it should return the fragment of string
798 that has been converted, and modify $octets in-place to remove the converted part
799 leaving it starting with the problem fragment.
801 If check is is false then C<decode> should make a "best effort" to convert
802 the string - for example by using Unicode's "\x{FFFD}" as a replacement character.
806 It should be noted that the check behaviour is different from the outer
807 public API. The logic is that the "unchecked" case is useful when
808 encoding is part of a stream which may be reporting errors (e.g. STDERR).
809 In such cases it is desirable to get everything through somehow without
810 causing additional errors which obscure the original one. Also the encoding
811 is best placed to know what the correct replacement character is, so if that
812 is the desired behaviour then letting low level code do it is the most efficient.
814 In contrast if check is true, the scheme above allows the encoding to do as
815 much as it can and tell layer above how much that was. What is lacking
816 at present is a mechanism to report what went wrong. The most likely interface
817 will be an additional method call to the object, or perhaps
818 (to avoid forcing per-stream objects on otherwise stateless encodings)
819 and additional parameter.
821 It is also highly desirable that encoding classes inherit from C<Encode::Encoding>
822 as a base class. This allows that class to define additional behaviour for
823 all encoding objects. For example built in Unicode, UCS-2 and UTF-8 classes
826 package Encode::MyEncoding;
827 use base qw(Encode::Encoding);
829 __PACKAGE__->Define(qw(myCanonical myAlias));
831 To create an object with bless {Name => ...},$class, and call define_encoding.
832 They inherit their C<name> method from C<Encode::Encoding>.
834 =head2 Compiled Encodings
836 F<Encode.xs> provides a class C<Encode::XS> which provides the interface described
837 above. It calls a generic octet-sequence to octet-sequence "engine" that is
838 driven by tables (defined in F<encengine.c>). The same engine is used for both
839 encode and decode. C<Encode:XS>'s C<encode> forces perl's characters to their UTF-8 form
840 and then treats them as just another multibyte encoding. C<Encode:XS>'s C<decode> transforms
841 the sequence and then turns the UTF-8-ness flag as that is the form that the tables
842 are defined to produce. For details of the engine see the comments in F<encengine.c>.
844 The tables are produced by the perl script F<compile> (the name needs to change so
845 we can eventually install it somewhere). F<compile> can currently read two formats:
851 This is a coined format used by Tcl. It is documented in Encode/EncodeFormat.pod.
855 This is the semi-standard format used by IBM's ICU package.
859 F<compile> can write the following forms:
865 See above - the F<Encode/*.ucm> files provided with the distribution have
866 been created from the original Tcl .enc files using this approach.
870 Produces tables as C data structures - this is used to build in encodings
871 into F<Encode.so>/F<Encode.dll>.
875 In theory this allows encodings to be stand-alone loadable perl extensions.
876 The process has not yet been tested. The plan is to use this approach
877 for large East Asian encodings.
881 The set of encodings built-in to F<Encode.so>/F<Encode.dll> is determined by
882 F<Makefile.PL>. The current set is as follows:
886 =item ascii and iso-8859-*
888 That is all the common 8-bit "western" encodings.
890 =item IBM-1047 and two other variants of EBCDIC.
892 These are the same variants that are supported by EBCDIC perl as "native" encodings.
893 They are included to prove "reversibility" of some constructs in EBCDIC perl.
895 =item symbol and dingbats as used by Tk on X11.
897 (The reason Encode got started was to support perl/Tk.)
901 That set is rather ad. hoc. and has been driven by the needs of the tests rather
902 than the needs of typical applications. It is likely to be rationalized.
906 L<perlunicode>, L<perlebcdic>, L<perlfunc/open>