Commit | Line | Data |
b786ee6f |
1 | # $Id: encoding.pm,v 1.47 2003/08/20 11:15:31 dankogai Exp dankogai $ |
3ef515df |
2 | package encoding; |
b786ee6f |
3 | our $VERSION = do { my @r = (q$Revision: 1.47 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r }; |
3ef515df |
4 | |
5 | use Encode; |
046f36bf |
6 | use strict; |
8f139f4c |
7 | sub DEBUG () { 0 } |
3ef515df |
8 | |
9 | BEGIN { |
10 | if (ord("A") == 193) { |
11 | require Carp; |
10c5ecbb |
12 | Carp::croak("encoding pragma does not support EBCDIC platforms"); |
3ef515df |
13 | } |
14 | } |
15 | |
0ab8f81e |
16 | our $HAS_PERLIO = 0; |
17 | eval { require PerlIO::encoding }; |
18 | unless ($@){ |
19 | $HAS_PERLIO = (PerlIO::encoding->VERSION >= 0.02); |
20 | } |
b2704119 |
21 | |
151b5d36 |
22 | sub _exception{ |
23 | my $name = shift; |
b5ab1f6f |
24 | $] > 5.008 and return 0; # 5.8.1 or higher then no |
151b5d36 |
25 | my %utfs = map {$_=>1} |
26 | qw(utf8 UCS-2BE UCS-2LE UTF-16 UTF-16BE UTF-16LE |
27 | UTF-32 UTF-32BE UTF-32LE); |
b5ab1f6f |
28 | $utfs{$name} or return 0; # UTFs or no |
151b5d36 |
29 | require Config; Config->import(); our %Config; |
2b6a28d4 |
30 | return $Config{perl_patchlevel} ? 0 : 1 # maintperl then no |
151b5d36 |
31 | } |
fa6f41cf |
32 | |
3ef515df |
33 | sub import { |
34 | my $class = shift; |
35 | my $name = shift; |
36 | my %arg = @_; |
37 | $name ||= $ENV{PERL_ENCODING}; |
3ef515df |
38 | my $enc = find_encoding($name); |
39 | unless (defined $enc) { |
40 | require Carp; |
10c5ecbb |
41 | Carp::croak("Unknown encoding '$name'"); |
3ef515df |
42 | } |
fa6f41cf |
43 | $name = $enc->name; # canonize |
0f7c507f |
44 | unless ($arg{Filter}) { |
8f139f4c |
45 | DEBUG and warn "_exception($name) = ", _exception($name); |
151b5d36 |
46 | _exception($name) or ${^ENCODING} = $enc; |
85982a32 |
47 | $HAS_PERLIO or return 1; |
aae85ceb |
48 | }else{ |
49 | defined(${^ENCODING}) and undef ${^ENCODING}; |
151b5d36 |
50 | # implicitly 'use utf8' |
51 | require utf8; # to fetch $utf8::hint_bits; |
52 | $^H |= $utf8::hint_bits; |
aae85ceb |
53 | eval { |
54 | require Filter::Util::Call ; |
55 | Filter::Util::Call->import ; |
aae85ceb |
56 | filter_add(sub{ |
151b5d36 |
57 | my $status = filter_read(); |
58 | if ($status > 0){ |
aae85ceb |
59 | $_ = $enc->decode($_, 1); |
8f139f4c |
60 | DEBUG and warn $_; |
aae85ceb |
61 | } |
62 | $status ; |
63 | }); |
64 | }; |
8f139f4c |
65 | } DEBUG and warn "Filter installed"; |
05ef2f67 |
66 | defined ${^UNICODE} and ${^UNICODE} != 0 and return 1; |
151b5d36 |
67 | for my $h (qw(STDIN STDOUT)){ |
68 | if ($arg{$h}){ |
69 | unless (defined find_encoding($arg{$h})) { |
70 | require Carp; |
71 | Carp::croak("Unknown encoding for $h, '$arg{$h}'"); |
72 | } |
6be7c101 |
73 | eval { binmode($h, ":raw :encoding($arg{$h})") }; |
151b5d36 |
74 | }else{ |
75 | unless (exists $arg{$h}){ |
76 | eval { |
77 | no warnings 'uninitialized'; |
6be7c101 |
78 | binmode($h, ":raw :encoding($name)"); |
151b5d36 |
79 | }; |
80 | } |
81 | } |
82 | if ($@){ |
83 | require Carp; |
84 | Carp::croak($@); |
85 | } |
3ef515df |
86 | } |
87 | return 1; # I doubt if we need it, though |
88 | } |
89 | |
90 | sub unimport{ |
91 | no warnings; |
92 | undef ${^ENCODING}; |
621b0f8d |
93 | if ($HAS_PERLIO){ |
94 | binmode(STDIN, ":raw"); |
95 | binmode(STDOUT, ":raw"); |
96 | }else{ |
6be7c101 |
97 | binmode(STDIN); |
98 | binmode(STDOUT); |
621b0f8d |
99 | } |
aae85ceb |
100 | if ($INC{"Filter/Util/Call.pm"}){ |
101 | eval { filter_del() }; |
102 | } |
3ef515df |
103 | } |
104 | |
105 | 1; |
106 | __END__ |
85982a32 |
107 | |
3ef515df |
108 | =pod |
109 | |
110 | =head1 NAME |
111 | |
0ab8f81e |
112 | encoding - allows you to write your script in non-ascii or non-utf8 |
3ef515df |
113 | |
114 | =head1 SYNOPSIS |
115 | |
962111ca |
116 | use encoding "greek"; # Perl like Greek to you? |
3ef515df |
117 | use encoding "euc-jp"; # Jperl! |
118 | |
962111ca |
119 | # or you can even do this if your shell supports your native encoding |
3ef515df |
120 | |
962111ca |
121 | perl -Mencoding=latin2 -e '...' # Feeling centrally European? |
0ab8f81e |
122 | perl -Mencoding=euc-kr -e '...' # Or Korean? |
3ef515df |
123 | |
3ef515df |
124 | # more control |
125 | |
962111ca |
126 | # A simple euc-cn => utf-8 converter |
6d1c0808 |
127 | use encoding "euc-cn", STDOUT => "utf8"; while(<>){print}; |
3ef515df |
128 | |
129 | # "no encoding;" supported (but not scoped!) |
130 | no encoding; |
131 | |
aae85ceb |
132 | # an alternate way, Filter |
133 | use encoding "euc-jp", Filter=>1; |
aae85ceb |
134 | # now you can use kanji identifiers -- in euc-jp! |
135 | |
3ef515df |
136 | =head1 ABSTRACT |
137 | |
962111ca |
138 | Let's start with a bit of history: Perl 5.6.0 introduced Unicode |
139 | support. You could apply C<substr()> and regexes even to complex CJK |
140 | characters -- so long as the script was written in UTF-8. But back |
0ab8f81e |
141 | then, text editors that supported UTF-8 were still rare and many users |
142 | instead chose to write scripts in legacy encodings, giving up a whole |
143 | new feature of Perl 5.6. |
3ef515df |
144 | |
0ab8f81e |
145 | Rewind to the future: starting from perl 5.8.0 with the B<encoding> |
962111ca |
146 | pragma, you can write your script in any encoding you like (so long |
147 | as the C<Encode> module supports it) and still enjoy Unicode support. |
0f29a567 |
148 | This pragma achieves that by doing the following: |
05ef2f67 |
149 | |
150 | =over |
151 | |
152 | =item * |
153 | |
154 | Internally converts all literals (C<q//,qq//,qr//,qw///, qx//>) from |
155 | the encoding specified to utf8. In Perl 5.8.1 and later, literals in |
156 | C<tr///> and C<DATA> pseudo-filehandle are also converted. |
157 | |
158 | =item * |
159 | |
160 | Changing PerlIO layers of C<STDIN> and C<STDOUT> to the encoding |
161 | specified. |
162 | |
163 | =back |
164 | |
165 | =head2 Literal Conversions |
166 | |
0ab8f81e |
167 | You can write code in EUC-JP as follows: |
3ef515df |
168 | |
169 | my $Rakuda = "\xF1\xD1\xF1\xCC"; # Camel in Kanji |
170 | #<-char-><-char-> # 4 octets |
171 | s/\bCamel\b/$Rakuda/; |
172 | |
173 | And with C<use encoding "euc-jp"> in effect, it is the same thing as |
962111ca |
174 | the code in UTF-8: |
3ef515df |
175 | |
32b9ed1f |
176 | my $Rakuda = "\x{99F1}\x{99DD}"; # two Unicode Characters |
3ef515df |
177 | s/\bCamel\b/$Rakuda/; |
178 | |
05ef2f67 |
179 | =head2 PerlIO layers for C<STD(IN|OUT)> |
180 | |
181 | The B<encoding> pragma also modifies the filehandle layers of |
4b291ae6 |
182 | STDIN and STDOUT to the specified encoding. Therefore, |
3ef515df |
183 | |
184 | use encoding "euc-jp"; |
185 | my $message = "Camel is the symbol of perl.\n"; |
186 | my $Rakuda = "\xF1\xD1\xF1\xCC"; # Camel in Kanji |
187 | $message =~ s/\bCamel\b/$Rakuda/; |
188 | print $message; |
189 | |
962111ca |
190 | Will print "\xF1\xD1\xF1\xCC is the symbol of perl.\n", |
191 | not "\x{99F1}\x{99DD} is the symbol of perl.\n". |
3ef515df |
192 | |
0ab8f81e |
193 | You can override this by giving extra arguments; see below. |
3ef515df |
194 | |
990e18f7 |
195 | =head2 Implicit upgrading for byte strings |
196 | |
197 | By default, if strings operating under byte semantics and strings |
198 | with Unicode character data are concatenated, the new string will |
199 | be created by decoding the byte strings as I<ISO 8859-1 (Latin-1)>. |
200 | |
201 | The B<encoding> pragma changes this to use the specified encoding |
202 | instead. For example: |
203 | |
204 | use encoding 'utf8'; |
205 | my $string = chr(20000); # a Unicode string |
206 | utf8::encode($string); # now it's a UTF-8 encoded byte string |
207 | # concatenate with another Unicode string |
208 | print length($string . chr(20000)); |
209 | |
210 | Will print C<2>, because C<$string> is upgraded as UTF-8. Without |
211 | C<use encoding 'utf8';>, it will print C<4> instead, since C<$string> |
212 | is three octets when interpreted as Latin-1. |
213 | |
05ef2f67 |
214 | =head1 FEATURES THAT REQUIRE 5.8.1 |
215 | |
216 | Some of the features offered by this pragma requires perl 5.8.1. Most |
0f29a567 |
217 | of these are done by Inaba Hiroto. Any other features and changes |
05ef2f67 |
218 | are good for 5.8.0. |
219 | |
220 | =over |
221 | |
222 | =item "NON-EUC" doublebyte encodings |
223 | |
0f29a567 |
224 | Because perl needs to parse script before applying this pragma, such |
05ef2f67 |
225 | encodings as Shift_JIS and Big-5 that may contain '\' (BACKSLASH; |
226 | \x5c) in the second byte fails because the second byte may |
0f29a567 |
227 | accidentally escape the quoting character that follows. Perl 5.8.1 |
05ef2f67 |
228 | or later fixes this problem. |
229 | |
230 | =item tr// |
231 | |
232 | C<tr//> was overlooked by Perl 5 porters when they released perl 5.8.0 |
233 | See the section below for details. |
234 | |
235 | =item DATA pseudo-filehandle |
236 | |
237 | Another feature that was overlooked was C<DATA>. |
238 | |
239 | =back |
240 | |
3ef515df |
241 | =head1 USAGE |
242 | |
243 | =over 4 |
244 | |
245 | =item use encoding [I<ENCNAME>] ; |
246 | |
05ef2f67 |
247 | Sets the script encoding to I<ENCNAME>. And unless ${^UNICODE} |
248 | exists and non-zero, PerlIO layers of STDIN and STDOUT are set to |
249 | ":encoding(I<ENCNAME>)". |
250 | |
251 | Note that STDERR WILL NOT be changed. |
252 | |
253 | Also note that non-STD file handles remain unaffected. Use C<use |
254 | open> or C<binmode> to change layers of those. |
3ef515df |
255 | |
256 | If no encoding is specified, the environment variable L<PERL_ENCODING> |
962111ca |
257 | is consulted. If no encoding can be found, the error C<Unknown encoding |
258 | 'I<ENCNAME>'> will be thrown. |
3ef515df |
259 | |
aae85ceb |
260 | =item use encoding I<ENCNAME> [ STDIN =E<gt> I<ENCNAME_IN> ...] ; |
3ef515df |
261 | |
0ab8f81e |
262 | You can also individually set encodings of STDIN and STDOUT via the |
32b9ed1f |
263 | C<< STDIN => I<ENCNAME> >> form. In this case, you cannot omit the |
264 | first I<ENCNAME>. C<< STDIN => undef >> turns the IO transcoding |
aae85ceb |
265 | completely off. |
3ef515df |
266 | |
05ef2f67 |
267 | When ${^UNICODE} exists and non-zero, these options will completely |
268 | ignored. ${^UNICODE} is a variable introduced in perl 5.8.1. See |
269 | L<perlrun> see L<perlvar/"${^UNICODE}"> and L<perlrun/"-C"> for |
270 | details (perl 5.8.1 and later). |
271 | |
151b5d36 |
272 | =item use encoding I<ENCNAME> Filter=E<gt>1; |
273 | |
274 | This turns the encoding pragma into a source filter. While the |
275 | default approach just decodes interpolated literals (in qq() and |
276 | qr()), this will apply a source filter to the entire source code. See |
05ef2f67 |
277 | L</"The Filter Option"> below for details. |
151b5d36 |
278 | |
3ef515df |
279 | =item no encoding; |
280 | |
05ef2f67 |
281 | Unsets the script encoding. The layers of STDIN, STDOUT are |
962111ca |
282 | reset to ":raw" (the default unprocessed raw stream of bytes). |
3ef515df |
283 | |
284 | =back |
285 | |
151b5d36 |
286 | =head1 The Filter Option |
287 | |
288 | The magic of C<use encoding> is not applied to the names of |
289 | identifiers. In order to make C<${"\x{4eba}"}++> ($human++, where human |
290 | is a single Han ideograph) work, you still need to write your script |
291 | in UTF-8 -- or use a source filter. That's what 'Filter=>1' does. |
292 | |
151b5d36 |
293 | What does this mean? Your source code behaves as if it is written in |
294 | UTF-8 with 'use utf8' in effect. So even if your editor only supports |
295 | Shift_JIS, for example, you can still try examples in Chapter 15 of |
296 | C<Programming Perl, 3rd Ed.>. For instance, you can use UTF-8 |
297 | identifiers. |
298 | |
299 | This option is significantly slower and (as of this writing) non-ASCII |
300 | identifiers are not very stable WITHOUT this option and with the |
301 | source code written in UTF-8. |
302 | |
303 | =head2 Filter-related changes at Encode version 1.87 |
304 | |
305 | =over |
306 | |
307 | =item * |
308 | |
309 | The Filter option now sets STDIN and STDOUT like non-filter options. |
310 | And C<< STDIN=>I<ENCODING> >> and C<< STDOUT=>I<ENCODING> >> work like |
311 | non-filter version. |
312 | |
313 | =item * |
314 | |
315 | C<use utf8> is implicitly declared so you no longer have to C<use |
316 | utf8> to C<${"\x{4eba}"}++>. |
317 | |
318 | =back |
319 | |
3ef515df |
320 | =head1 CAVEATS |
321 | |
322 | =head2 NOT SCOPED |
323 | |
324 | The pragma is a per script, not a per block lexical. Only the last |
621b0f8d |
325 | C<use encoding> or C<no encoding> matters, and it affects |
326 | B<the whole script>. However, the <no encoding> pragma is supported and |
327 | B<use encoding> can appear as many times as you want in a given script. |
328 | The multiple use of this pragma is discouraged. |
329 | |
0f29a567 |
330 | By the same reason, the use this pragma inside modules is also |
331 | discouraged (though not as strongly discouranged as the case above. |
332 | See below). |
05ef2f67 |
333 | |
334 | If you still have to write a module with this pragma, be very careful |
335 | of the load order. See the codes below; |
336 | |
337 | # called module |
338 | package Module_IN_BAR; |
339 | use encoding "bar"; |
340 | # stuff in "bar" encoding here |
341 | 1; |
342 | |
343 | # caller script |
344 | use encoding "foo" |
345 | use Module_IN_BAR; |
346 | # surprise! use encoding "bar" is in effect. |
347 | |
348 | The best way to avoid this oddity is to use this pragma RIGHT AFTER |
349 | other modules are loaded. i.e. |
350 | |
351 | use Module_IN_BAR; |
352 | use encoding "foo"; |
3ef515df |
353 | |
354 | =head2 DO NOT MIX MULTIPLE ENCODINGS |
355 | |
356 | Notice that only literals (string or regular expression) having only |
357 | legacy code points are affected: if you mix data like this |
358 | |
359 | \xDF\x{100} |
360 | |
361 | the data is assumed to be in (Latin 1 and) Unicode, not in your native |
362 | encoding. In other words, this will match in "greek": |
363 | |
364 | "\xDF" =~ /\x{3af}/ |
365 | |
366 | but this will not |
367 | |
368 | "\xDF\x{100}" =~ /\x{3af}\x{100}/ |
369 | |
962111ca |
370 | since the C<\xDF> (ISO 8859-7 GREEK SMALL LETTER IOTA WITH TONOS) on |
371 | the left will B<not> be upgraded to C<\x{3af}> (Unicode GREEK SMALL |
372 | LETTER IOTA WITH TONOS) because of the C<\x{100}> on the left. You |
373 | should not be mixing your legacy data and Unicode in the same string. |
3ef515df |
374 | |
375 | This pragma also affects encoding of the 0x80..0xFF code point range: |
376 | normally characters in that range are left as eight-bit bytes (unless |
377 | they are combined with characters with code points 0x100 or larger, |
378 | in which case all characters need to become UTF-8 encoded), but if |
379 | the C<encoding> pragma is present, even the 0x80..0xFF range always |
380 | gets UTF-8 encoded. |
381 | |
382 | After all, the best thing about this pragma is that you don't have to |
0ab8f81e |
383 | resort to \x{....} just to spell your name in a native encoding. |
384 | So feel free to put your strings in your encoding in quotes and |
385 | regexes. |
3ef515df |
386 | |
151b5d36 |
387 | =head2 tr/// with ranges |
4b291ae6 |
388 | |
389 | The B<encoding> pragma works by decoding string literals in |
151b5d36 |
390 | C<q//,qq//,qr//,qw///, qx//> and so forth. In perl 5.8.0, this |
4b291ae6 |
391 | does not apply to C<tr///>. Therefore, |
392 | |
393 | use encoding 'euc-jp'; |
394 | #.... |
395 | $kana =~ tr/\xA4\xA1-\xA4\xF3/\xA5\xA1-\xA5\xF3/; |
396 | # -------- -------- -------- -------- |
397 | |
398 | Does not work as |
399 | |
400 | $kana =~ tr/\x{3041}-\x{3093}/\x{30a1}-\x{30f3}/; |
401 | |
402 | =over |
403 | |
404 | =item Legend of characters above |
405 | |
406 | utf8 euc-jp charnames::viacode() |
407 | ----------------------------------------- |
408 | \x{3041} \xA4\xA1 HIRAGANA LETTER SMALL A |
409 | \x{3093} \xA4\xF3 HIRAGANA LETTER N |
410 | \x{30a1} \xA5\xA1 KATAKANA LETTER SMALL A |
411 | \x{30f3} \xA5\xF3 KATAKANA LETTER N |
412 | |
413 | =back |
414 | |
05ef2f67 |
415 | This counterintuitive behavior has been fixed in perl 5.8.1. |
151b5d36 |
416 | |
4b291ae6 |
417 | =head3 workaround to tr///; |
418 | |
ce16148b |
419 | In perl 5.8.0, you can work around as follows; |
4b291ae6 |
420 | |
421 | use encoding 'euc-jp'; |
151b5d36 |
422 | # .... |
4b291ae6 |
423 | eval qq{ \$kana =~ tr/\xA4\xA1-\xA4\xF3/\xA5\xA1-\xA5\xF3/ }; |
424 | |
ce16148b |
425 | Note the C<tr//> expression is surrounded by C<qq{}>. The idea behind |
4b291ae6 |
426 | is the same as classic idiom that makes C<tr///> 'interpolate'. |
427 | |
428 | tr/$from/$to/; # wrong! |
429 | eval qq{ tr/$from/$to/ }; # workaround. |
430 | |
431 | Nevertheless, in case of B<encoding> pragma even C<q//> is affected so |
432 | C<tr///> not being decoded was obviously against the will of Perl5 |
05ef2f67 |
433 | Porters so it has been fixed in Perl 5.8.1 or later. |
aae85ceb |
434 | |
3ef515df |
435 | =head1 EXAMPLE - Greekperl |
436 | |
437 | use encoding "iso 8859-7"; |
438 | |
0ab8f81e |
439 | # \xDF in ISO 8859-7 (Greek) is \x{3af} in Unicode. |
3ef515df |
440 | |
441 | $a = "\xDF"; |
442 | $b = "\x{100}"; |
443 | |
444 | printf "%#x\n", ord($a); # will print 0x3af, not 0xdf |
445 | |
446 | $c = $a . $b; |
447 | |
448 | # $c will be "\x{3af}\x{100}", not "\x{df}\x{100}". |
449 | |
450 | # chr() is affected, and ... |
451 | |
452 | print "mega\n" if ord(chr(0xdf)) == 0x3af; |
453 | |
454 | # ... ord() is affected by the encoding pragma ... |
455 | |
456 | print "tera\n" if ord(pack("C", 0xdf)) == 0x3af; |
457 | |
458 | # ... as are eq and cmp ... |
459 | |
460 | print "peta\n" if "\x{3af}" eq pack("C", 0xdf); |
461 | print "exa\n" if "\x{3af}" cmp pack("C", 0xdf) == 0; |
462 | |
463 | # ... but pack/unpack C are not affected, in case you still |
0ab8f81e |
464 | # want to go back to your native encoding |
3ef515df |
465 | |
466 | print "zetta\n" if unpack("C", (pack("C", 0xdf))) == 0xdf; |
467 | |
468 | =head1 KNOWN PROBLEMS |
469 | |
151b5d36 |
470 | =over |
471 | |
0f29a567 |
472 | =item literals in regex that are longer than 127 bytes |
151b5d36 |
473 | |
0ab8f81e |
474 | For native multibyte encodings (either fixed or variable length), |
3ef515df |
475 | the current implementation of the regular expressions may introduce |
0ab8f81e |
476 | recoding errors for regular expression literals longer than 127 bytes. |
3ef515df |
477 | |
05ef2f67 |
478 | =item EBCDIC |
151b5d36 |
479 | |
3ef515df |
480 | The encoding pragma is not supported on EBCDIC platforms. |
0ab8f81e |
481 | (Porters who are willing and able to remove this limitation are |
482 | welcome.) |
3ef515df |
483 | |
05ef2f67 |
484 | =item format |
485 | |
486 | This pragma doesn't work well with format because PerlIO does not |
487 | get along very well with it. When format contains non-ascii |
488 | characters it prints funny or gets "wide character warnings". |
489 | To understand it, try the code below. |
490 | |
491 | # Save this one in utf8 |
492 | # replace *non-ascii* with a non-ascii string |
493 | my $camel; |
494 | format STDOUT = |
495 | *non-ascii*@>>>>>>> |
496 | $camel |
497 | . |
498 | $camel = "*non-ascii*"; |
499 | binmode(STDOUT=>':encoding(utf8)'); # bang! |
500 | write; # funny |
501 | print $camel, "\n"; # fine |
502 | |
503 | Without binmode this happens to work but without binmode, print() |
504 | fails instead of write(). |
505 | |
506 | At any rate, the very use of format is questionable when it comes to |
507 | unicode characters since you have to consider such things as character |
508 | width (i.e. double-width for ideographs) and directions (i.e. BIDI for |
509 | Arabic and Hebrew). |
510 | |
151b5d36 |
511 | =back |
512 | |
05ef2f67 |
513 | =head1 HISTORY |
514 | |
515 | This pragma first appeared in Perl 5.8.0. For features that require |
516 | 5.8.1 and better, see above. |
517 | |
3ef515df |
518 | =head1 SEE ALSO |
519 | |
aae85ceb |
520 | L<perlunicode>, L<Encode>, L<open>, L<Filter::Util::Call>, |
521 | |
522 | Ch. 15 of C<Programming Perl (3rd Edition)> |
523 | by Larry Wall, Tom Christiansen, Jon Orwant; |
524 | O'Reilly & Associates; ISBN 0-596-00027-8 |
3ef515df |
525 | |
526 | =cut |