3 Encode::PerlIO -- a detailed document on Encode and PerlIO
7 It is very common to want to do encoding transformations when
8 reading or writing files, network connections, pipes etc.
9 If Perl is configured to use the new 'perlio' IO system then
10 C<Encode> provides a "layer" (see L<PerlIO>) which can transform
11 data as it is read or written.
13 Here is how the blind poet would modernise the encoding:
16 open(my $iliad,'<:encoding(iso-8859-7)','iliad.greek');
17 open(my $utf8,'>:utf8','iliad.utf8');
23 In addition, the new IO system can also be configured to read/write
24 UTF-8 encoded characters (as noted above, this is efficient):
26 open(my $fh,'>:utf8','anything');
27 print $fh "Any \x{0021} string \N{SMILEY FACE}\n";
29 Either of the above forms of "layer" specifications can be made the default
30 for a lexical scope with the C<use open ...> pragma. See L<open>.
32 Once a handle is open, its layers can be altered using C<binmode>.
34 Without any such configuration, or if Perl itself is built using the
35 system's own IO, then write operations assume that the file handle
36 accepts only I<bytes> and will C<die> if a character larger than 255 is
37 written to the handle. When reading, each octet from the handle becomes
38 a byte-in-a-character. Note that this default is the same behaviour
39 as bytes-only languages (including Perl before v5.6) would have,
40 and is sufficient to handle native 8-bit encodings e.g. iso-8859-1,
41 EBCDIC etc. and any legacy mechanisms for handling other encodings
44 In other cases, it is the program's responsibility to transform
45 characters into bytes using the API above before doing writes, and to
46 transform the bytes read from a handle into characters before doing
47 "character operations" (e.g. C<lc>, C</\W+/>, ...).
49 You can also use PerlIO to convert larger amounts of data you don't
50 want to bring into memory. For example, to convert between ISO-8859-1
51 (Latin 1) and UTF-8 (or UTF-EBCDIC in EBCDIC machines):
53 open(F, "<:encoding(iso-8859-1)", "data.txt") or die $!;
54 open(G, ">:utf8", "data.utf") or die $!;
55 while (<F>) { print G }
57 # Could also do "print G <F>" but that would pull
58 # the whole file into memory just to write it out again.
62 open(my $f, "<:encoding(cp1252)")
63 open(my $g, ">:encoding(iso-8859-2)")
64 open(my $h, ">:encoding(latin9)") # iso-8859-15
66 See also L<encoding> for how to change the default encoding of the
69 =head1 How does it work?
71 Here is a crude diagram of how filehandle, PerlIO, and Encode
74 filehandle <-> PerlIO PerlIO <-> scalar (read/printed)
78 When PerlIO receives data from either direction, it fills a buffer
79 (currently with 1024 bytes) and passes the buffer to Encode.
80 Encode tries to convert the valid part and passes it back to PerlIO,
81 leaving invalid parts (usually a partial character) in the buffer.
82 PerlIO then appends more data to the buffer, calls Encode again,
83 and so on until the data stream ends.
85 To do so, PerlIO always calls (de|en)code methods with CHECK set to 1.
86 This ensures that the method stops at the right place when it
87 encounters partial character. The following is what happens when
88 PerlIO and Encode tries to encode (from utf8) more than 1024 bytes
89 and the buffer boundary happens to be in the middle of a character.
91 A B C .... ~ \x{3000} ....
92 41 42 43 .... 7E e3 80 80 ....
93 <- buffer --------------->
97 Encode converts from the beginning to \x7E, leaving \xe3 in the buffer
98 because it is invalid (partial character).
100 Unfortunately, this scheme does not work well with escape-based
101 encodings such as ISO-2022-JP. Let's see what happens in that case
106 Now let's see what happens when you try to decode from ISO-2022-JP and
107 the buffer ends in the middle of a character.
110 A B C .... ~ \e $ B |DAN | ....
111 41 42 43 .... 7E 1b 24 41 43 46 ....
112 <- buffer --------------------------->
113 << encoded >>>>>>>>>>>>>>>>>>>>>>>
115 As you see, the next buffer begins with \x43. But \x43 is 'C' in
116 ASCII, which is wrong in this case because we are now in JISX 0208
117 area so it has to convert \x43\x46, not \x43. Unlike utf8 and EUC,
118 in escape-based encodings you can't tell if a given octet is a whole
119 character or just part of it.
121 There are actually several ways to solve this problem but none of
122 them is fast enough to be practical. From Encode's point of view,
123 the easiest solution is for PerlIO to implement a line buffer instead
124 of a fixed-length buffer, but that makes PerlIO really complicated.
126 So for the time being, using escape-based encodings in the
127 ":encoding()" layer of PerlIO does not work well.
131 If you still insist, you can at least use ":encoding()" by making sure
132 the buffer never gets full. Here is an example.
135 binmode(STDOUT, ":encoding(7bit-jis)");
136 STDOUT->autoflush(1); # don't forget this!
137 for my $l (@lines){ # $l cannot be longer than 1023 bytes
141 =head2 How can I tell whether my encoding fully supports PerlIO ?
143 As of this writing, any encoding whose class belongs to Encode::XS and
144 Encode::Unicode works. The Encode module has a C<perlio_ok> method
145 which you can use before appling PerlIO encoding to the filehandle.
148 my $use_perlio = perlio_ok($enc);
149 my $layer = $use_perlio ? "<:raw" : "<:encoding($enc)";
150 open my $fh, $layer, $file or die "$file : $!";
152 $_ = decode($enc, $_) unless $use_perlio;
159 L<Encode::Supported>,
166 the Perl Unicode Mailing List E<lt>perl-unicode@perl.orgE<gt>