encoding(3pm)

1encoding(3pm)          Perl Programmers Reference Guide          encoding(3pm)
2
3
4

NAME

6       encoding - allows you to write your script in non-ascii or non-utf8
7

SYNOPSIS

9         use encoding "greek";  # Perl like Greek to you?
10         use encoding "euc-jp"; # Jperl!
11
12         # or you can even do this if your shell supports your native encoding
13
14         perl -Mencoding=latin2 -e '...' # Feeling centrally European?
15         perl -Mencoding=euc-kr -e '...' # Or Korean?
16
17         # more control
18
19         # A simple euc-cn => utf-8 converter
20         use encoding "euc-cn", STDOUT => "utf8";  while(<>){print};
21
22         # "no encoding;" supported (but not scoped!)
23         no encoding;
24
25         # an alternate way, Filter
26         use encoding "euc-jp", Filter=>1;
27         # now you can use kanji identifiers -- in euc-jp!
28
29         # switch on locale -
30         # note that this probably means that unless you have a complete control
31         # over the environments the application is ever going to be run, you should
32         # NOT use the feature of encoding pragma allowing you to write your script
33         # in any recognized encoding because changing locale settings will wreck
34         # the script; you can of course still use the other features of the pragma.
35         use encoding ':locale';
36

ABSTRACT

38       Let's start with a bit of history: Perl 5.6.0 introduced Unicode sup‐
39       port.  You could apply "substr()" and regexes even to complex CJK char‐
40       acters -- so long as the script was written in UTF-8.  But back then,
41       text editors that supported UTF-8 were still rare and many users
42       instead chose to write scripts in legacy encodings, giving up a whole
43       new feature of Perl 5.6.
44
45       Rewind to the future: starting from perl 5.8.0 with the encoding
46       pragma, you can write your script in any encoding you like (so long as
47       the "Encode" module supports it) and still enjoy Unicode support.  This
48       pragma achieves that by doing the following:
49
50       ·   Internally converts all literals ("q//,qq//,qr//,qw///, qx//") from
51           the encoding specified to utf8.  In Perl 5.8.1 and later, literals
52           in "tr///" and "DATA" pseudo-filehandle are also converted.
53
54       ·   Changing PerlIO layers of "STDIN" and "STDOUT" to the encoding
55            specified.
56
57       Literal Conversions
58
59       You can write code in EUC-JP as follows:
60
61         my $Rakuda = "\xF1\xD1\xF1\xCC"; # Camel in Kanji
62                      #<-char-><-char->   # 4 octets
63         s/\bCamel\b/$Rakuda/;
64
65       And with "use encoding "euc-jp"" in effect, it is the same thing as the
66       code in UTF-8:
67
68         my $Rakuda = "\x{99F1}\x{99DD}"; # two Unicode Characters
69         s/\bCamel\b/$Rakuda/;
70
71       PerlIO layers for "STD(IN⎪OUT)"
72
73       The encoding pragma also modifies the filehandle layers of STDIN and
74       STDOUT to the specified encoding.  Therefore,
75
76         use encoding "euc-jp";
77         my $message = "Camel is the symbol of perl.\n";
78         my $Rakuda = "\xF1\xD1\xF1\xCC"; # Camel in Kanji
79         $message =~ s/\bCamel\b/$Rakuda/;
80         print $message;
81
82       Will print "\xF1\xD1\xF1\xCC is the symbol of perl.\n", not
83       "\x{99F1}\x{99DD} is the symbol of perl.\n".
84
85       You can override this by giving extra arguments; see below.
86
87       Implicit upgrading for byte strings
88
89       By default, if strings operating under byte semantics and strings with
90       Unicode character data are concatenated, the new string will be created
91       by decoding the byte strings as ISO 8859-1 (Latin-1).
92
93       The encoding pragma changes this to use the specified encoding instead.
94       For example:
95
96           use encoding 'utf8';
97           my $string = chr(20000); # a Unicode string
98           utf8::encode($string);   # now it's a UTF-8 encoded byte string
99           # concatenate with another Unicode string
100           print length($string . chr(20000));
101
102       Will print 2, because $string is upgraded as UTF-8.  Without "use
103       encoding 'utf8';", it will print 4 instead, since $string is three
104       octets when interpreted as Latin-1.
105

FEATURES THAT REQUIRE 5.8.1

107       Some of the features offered by this pragma requires perl 5.8.1.  Most
108       of these are done by Inaba Hiroto.  Any other features and changes are
109       good for 5.8.0.
110
111       "NON-EUC" doublebyte encodings
112           Because perl needs to parse script before applying this pragma,
113           such encodings as Shift_JIS and Big-5 that may contain '\' (BACK‐
114           SLASH; \x5c) in the second byte fails because the second byte may
115           accidentally escape the quoting character that follows.  Perl 5.8.1
116           or later fixes this problem.
117
118       tr//
119           "tr//" was overlooked by Perl 5 porters when they released perl
120           5.8.0 See the section below for details.
121
122       DATA pseudo-filehandle
123           Another feature that was overlooked was "DATA".
124

USAGE

126       use encoding [ENCNAME] ;
127           Sets the script encoding to ENCNAME.  And unless ${^UNICODE} exists
128           and non-zero, PerlIO layers of STDIN and STDOUT are set to ":encod‐
129           ing(ENCNAME)".
130
131           Note that STDERR WILL NOT be changed.
132
133           Also note that non-STD file handles remain unaffected.  Use "use
134           open" or "binmode" to change layers of those.
135
136           If no encoding is specified, the environment variable PERL_ENCODING
137           is consulted.  If no encoding can be found, the error "Unknown
138           encoding 'ENCNAME'" will be thrown.
139
140       use encoding ENCNAME [ STDIN => ENCNAME_IN ...] ;
141           You can also individually set encodings of STDIN and STDOUT via the
142           "STDIN => ENCNAME" form.  In this case, you cannot omit the first
143           ENCNAME.  "STDIN => undef" turns the IO transcoding completely off.
144
145           When ${^UNICODE} exists and non-zero, these options will completely
146           ignored.  ${^UNICODE} is a variable introduced in perl 5.8.1.  See
147           perlrun see "${^UNICODE}" in perlvar and "-C" in perlrun for
148           details (perl 5.8.1 and later).
149
150       use encoding ENCNAME Filter=>1;
151           This turns the encoding pragma into a source filter.  While the
152           default approach just decodes interpolated literals (in qq() and
153           qr()), this will apply a source filter to the entire source code.
154           See "The Filter Option" below for details.
155
156       no encoding;
157           Unsets the script encoding. The layers of STDIN, STDOUT are reset
158           to ":raw" (the default unprocessed raw stream of bytes).
159

The Filter Option

161       The magic of "use encoding" is not applied to the names of identifiers.
162       In order to make "${"\x{4eba}"}++" ($human++, where human is a single
163       Han ideograph) work, you still need to write your script in UTF-8 -- or
164       use a source filter.  That's what 'Filter=>1' does.
165
166       What does this mean?  Your source code behaves as if it is written in
167       UTF-8 with 'use utf8' in effect.  So even if your editor only supports
168       Shift_JIS, for example, you can still try examples in Chapter 15 of
169       "Programming Perl, 3rd Ed.".  For instance, you can use UTF-8 identi‐
170       fiers.
171
172       This option is significantly slower and (as of this writing) non-ASCII
173       identifiers are not very stable WITHOUT this option and with the source
174       code written in UTF-8.
175
176       Filter-related changes at Encode version 1.87
177
178       ·   The Filter option now sets STDIN and STDOUT like non-filter
179           options.  And "STDIN=>ENCODING" and "STDOUT=>ENCODING" work like
180           non-filter version.
181
182       ·   "use utf8" is implicitly declared so you no longer have to "use
183           utf8" to "${"\x{4eba}"}++".
184

CAVEATS

186       NOT SCOPED
187
188       The pragma is a per script, not a per block lexical.  Only the last
189       "use encoding" or "no encoding" matters, and it affects the whole
190       script.  However, the <no encoding> pragma is supported and use encod‐
191       ing can appear as many times as you want in a given script.  The multi‐
192       ple use of this pragma is discouraged.
193
194       By the same reason, the use this pragma inside modules is also discour‐
195       aged (though not as strongly discouraged as the case above.  See
196       below).
197
198       If you still have to write a module with this pragma, be very careful
199       of the load order.  See the codes below;
200
201         # called module
202         package Module_IN_BAR;
203         use encoding "bar";
204         # stuff in "bar" encoding here
205         1;
206
207         # caller script
208         use encoding "foo"
209         use Module_IN_BAR;
210         # surprise! use encoding "bar" is in effect.
211
212       The best way to avoid this oddity is to use this pragma RIGHT AFTER
213       other modules are loaded.  i.e.
214
215         use Module_IN_BAR;
216         use encoding "foo";
217
218       DO NOT MIX MULTIPLE ENCODINGS
219
220       Notice that only literals (string or regular expression) having only
221       legacy code points are affected: if you mix data like this
222
223               \xDF\x{100}
224
225       the data is assumed to be in (Latin 1 and) Unicode, not in your native
226       encoding.  In other words, this will match in "greek":
227
228               "\xDF" =~ /\x{3af}/
229
230       but this will not
231
232               "\xDF\x{100}" =~ /\x{3af}\x{100}/
233
234       since the "\xDF" (ISO 8859-7 GREEK SMALL LETTER IOTA WITH TONOS) on the
235       left will not be upgraded to "\x{3af}" (Unicode GREEK SMALL LETTER IOTA
236       WITH TONOS) because of the "\x{100}" on the left.  You should not be
237       mixing your legacy data and Unicode in the same string.
238
239       This pragma also affects encoding of the 0x80..0xFF code point range:
240       normally characters in that range are left as eight-bit bytes (unless
241       they are combined with characters with code points 0x100 or larger, in
242       which case all characters need to become UTF-8 encoded), but if the
243       "encoding" pragma is present, even the 0x80..0xFF range always gets
244       UTF-8 encoded.
245
246       After all, the best thing about this pragma is that you don't have to
247       resort to \x{....} just to spell your name in a native encoding.  So
248       feel free to put your strings in your encoding in quotes and regexes.
249
250       tr/// with ranges
251
252       The encoding pragma works by decoding string literals in
253       "q//,qq//,qr//,qw///, qx//" and so forth.  In perl 5.8.0, this does not
254       apply to "tr///".  Therefore,
255
256         use encoding 'euc-jp';
257         #....
258         $kana =~ tr/\xA4\xA1-\xA4\xF3/\xA5\xA1-\xA5\xF3/;
259         #           -------- -------- -------- --------
260
261       Does not work as
262
263         $kana =~ tr/\x{3041}-\x{3093}/\x{30a1}-\x{30f3}/;
264
265       Legend of characters above
266             utf8     euc-jp   charnames::viacode()
267             -----------------------------------------
268             \x{3041} \xA4\xA1 HIRAGANA LETTER SMALL A
269             \x{3093} \xA4\xF3 HIRAGANA LETTER N
270             \x{30a1} \xA5\xA1 KATAKANA LETTER SMALL A
271             \x{30f3} \xA5\xF3 KATAKANA LETTER N
272
273       This counterintuitive behavior has been fixed in perl 5.8.1.
274
275       workaround to tr///;
276
277       In perl 5.8.0, you can work around as follows;
278
279         use encoding 'euc-jp';
280         #  ....
281         eval qq{ \$kana =~ tr/\xA4\xA1-\xA4\xF3/\xA5\xA1-\xA5\xF3/ };
282
283       Note the "tr//" expression is surrounded by "qq{}".  The idea behind is
284       the same as classic idiom that makes "tr///" 'interpolate'.
285
286          tr/$from/$to/;            # wrong!
287          eval qq{ tr/$from/$to/ }; # workaround.
288
289       Nevertheless, in case of encoding pragma even "q//" is affected so
290       "tr///" not being decoded was obviously against the will of Perl5
291       Porters so it has been fixed in Perl 5.8.1 or later.
292

EXAMPLE - Greekperl

294           use encoding "iso 8859-7";
295
296           # \xDF in ISO 8859-7 (Greek) is \x{3af} in Unicode.
297
298           $a = "\xDF";
299           $b = "\x{100}";
300
301           printf "%#x\n", ord($a); # will print 0x3af, not 0xdf
302
303           $c = $a . $b;
304
305           # $c will be "\x{3af}\x{100}", not "\x{df}\x{100}".
306
307           # chr() is affected, and ...
308
309           print "mega\n"  if ord(chr(0xdf)) == 0x3af;
310
311           # ... ord() is affected by the encoding pragma ...
312
313           print "tera\n" if ord(pack("C", 0xdf)) == 0x3af;
314
315           # ... as are eq and cmp ...
316
317           print "peta\n" if "\x{3af}" eq  pack("C", 0xdf);
318           print "exa\n"  if "\x{3af}" cmp pack("C", 0xdf) == 0;
319
320           # ... but pack/unpack C are not affected, in case you still
321           # want to go back to your native encoding
322
323           print "zetta\n" if unpack("C", (pack("C", 0xdf))) == 0xdf;
324

KNOWN PROBLEMS

326       literals in regex that are longer than 127 bytes
327           For native multibyte encodings (either fixed or variable length),
328           the current implementation of the regular expressions may introduce
329           recoding errors for regular expression literals longer than 127
330           bytes.
331
332       EBCDIC
333           The encoding pragma is not supported on EBCDIC platforms.  (Porters
334           who are willing and able to remove this limitation are welcome.)
335
336       format
337           This pragma doesn't work well with format because PerlIO does not
338           get along very well with it.  When format contains non-ascii char‐
339           acters it prints funny or gets "wide character warnings".  To
340           understand it, try the code below.
341
342             # Save this one in utf8
343             # replace *non-ascii* with a non-ascii string
344             my $camel;
345             format STDOUT =
346             *non-ascii*@>>>>>>>
347             $camel
348             .
349             $camel = "*non-ascii*";
350             binmode(STDOUT=>':encoding(utf8)'); # bang!
351             write;              # funny
352             print $camel, "\n"; # fine
353
354           Without binmode this happens to work but without binmode, print()
355           fails instead of write().
356
357           At any rate, the very use of format is questionable when it comes
358           to unicode characters since you have to consider such things as
359           character width (i.e. double-width for ideographs) and directions
360           (i.e. BIDI for Arabic and Hebrew).
361
362       The Logic of :locale
363
364       The logic of ":locale" is as follows:
365
366       1.  If the platform supports the langinfo(CODESET) interface, the code‐
367           set returned is used as the default encoding for the open pragma.
368
369       2.  If 1. didn't work but we are under the locale pragma, the environ‐
370           ment variables LC_ALL and LANG (in that order) are matched for
371           encodings (the part after ".", if any), and if any found, that is
372           used as the default encoding for the open pragma.
373
374       3.  If 1. and 2. didn't work, the environment variables LC_ALL and LANG
375           (in that order) are matched for anything looking like UTF-8, and if
376           any found, ":utf8" is used as the default encoding for the open
377           pragma.
378
379       If your locale environment variables (LC_ALL, LC_CTYPE, LANG) contain
380       the strings 'UTF-8' or 'UTF8' (case-insensitive matching), the default
381       encoding of your STDIN, STDOUT, and STDERR, and of any subsequent file
382       open, is UTF-8.
383

HISTORY

385       This pragma first appeared in Perl 5.8.0.  For features that require
386       5.8.1 and better, see above.
387
388       The ":locale" subpragma was implemented in 2.01, or Perl 5.8.6.
389