Commit | Line | Data |
---|---|---|
de1df517 RGS |
1 | # $File: //member/autrijus/.vimrc $ $Author: autrijus $ |
2 | # $Revision: #14 $ $Change: 4137 $ $DateTime: 2003/02/08 11:41:59 $ | |
3 | ||
4 | package encoding::warnings; | |
5 | $encoding::warnings::VERSION = '0.05'; | |
6 | ||
7 | use strict; | |
8 | ||
9 | =head1 NAME | |
10 | ||
11 | encoding::warnings - Warn on implicit encoding conversions | |
12 | ||
13 | =head1 VERSION | |
14 | ||
15 | This document describes version 0.05 of encoding::warnings, released | |
16 | July 15, 2004. | |
17 | ||
18 | =head1 SYNOPSIS | |
19 | ||
20 | use encoding::warnings; # or 'FATAL' to raise fatal exceptions | |
21 | ||
22 | utf8::encode($a = chr(20000)); # a byte-string (raw bytes) | |
23 | $b = chr(20000); # a unicode-string (wide characters) | |
24 | ||
25 | # "Bytes implicitly upgraded into wide characters as iso-8859-1" | |
26 | $c = $a . $b; | |
27 | ||
28 | =head1 DESCRIPTION | |
29 | ||
30 | =head2 Overview of the problem | |
31 | ||
32 | By default, there is a fundamental asymmetry in Perl's unicode model: | |
33 | implicit upgrading from byte-strings to unicode-strings assumes that | |
34 | they were encoded in I<ISO 8859-1 (Latin-1)>, but unicode-strings are | |
35 | downgraded with UTF-8 encoding. This happens because the first 256 | |
36 | codepoints in Unicode happens to agree with Latin-1. | |
37 | ||
38 | However, this silent upgrading can easily cause problems, if you happen | |
39 | to mix unicode strings with non-Latin1 data -- i.e. byte-strings encoded | |
40 | in UTF-8 or other encodings. The error will not manifest until the | |
41 | combined string is written to output, at which time it would be impossible | |
42 | to see where did the silent upgrading occur. | |
43 | ||
44 | =head2 Detecting the problem | |
45 | ||
46 | This module simplifies the process of diagnosing such problems. Just put | |
47 | this line on top of your main program: | |
48 | ||
49 | use encoding::warnings; | |
50 | ||
51 | Afterwards, implicit upgrading of high-bit bytes will raise a warning. | |
52 | Ex.: C<Bytes implicitly upgraded into wide characters as iso-8859-1 at | |
53 | - line 7>. | |
54 | ||
55 | However, strings composed purely of ASCII code points (C<0x00>..C<0x7F>) | |
56 | will I<not> trigger this warning. | |
57 | ||
58 | You can also make the warnings fatal by importing this module as: | |
59 | ||
60 | use encoding::warnings 'FATAL'; | |
61 | ||
62 | =head2 Solving the problem | |
63 | ||
64 | Most of the time, this warning occurs when a byte-string is concatenated | |
65 | with a unicode-string. There are a number of ways to solve it: | |
66 | ||
67 | =over 4 | |
68 | ||
69 | =item * Upgrade both sides to unicode-strings | |
70 | ||
71 | If your program does not need compatibility for Perl 5.6 and earlier, | |
72 | the recommended approach is to apply appropriate IO disciplines, so all | |
73 | data in your program become unicode-strings. See L<encoding>, L<open> and | |
74 | L<perlfunc/binmode> for how. | |
75 | ||
76 | =item * Downgrade both sides to byte-strings | |
77 | ||
78 | The other way works too, especially if you are sure that all your data | |
79 | are under the same encoding, or if compatibility with older versions | |
80 | of Perl is desired. | |
81 | ||
82 | You may downgrade strings with C<Encode::encode> and C<utf8::encode>. | |
83 | See L<Encode> and L<utf8> for details. | |
84 | ||
85 | =item * Specify the encoding for implicit byte-string upgrading | |
86 | ||
87 | If you are confident that all byte-strings will be in a specific | |
88 | encoding like UTF-8, I<and> need not support older versions of Perl, | |
89 | use the C<encoding> pragma: | |
90 | ||
91 | use encoding 'utf8'; | |
92 | ||
93 | Similarly, this will silence warnings from this module, and preserve the | |
94 | default behaviour: | |
95 | ||
96 | use encoding 'iso-8859-1'; | |
97 | ||
98 | However, note that C<use encoding> actually had three distinct effects: | |
99 | ||
100 | =over 4 | |
101 | ||
102 | =item * PerlIO layers for B<STDIN> and B<STDOUT> | |
103 | ||
104 | This is similar to what L<open> pragma does. | |
105 | ||
106 | =item * Literal conversions | |
107 | ||
108 | This turns I<all> literal string in your program into unicode-strings | |
109 | (equivalent to a C<use utf8>), by decoding them using the specified | |
110 | encoding. | |
111 | ||
112 | =item * Implicit upgrading for byte-strings | |
113 | ||
114 | This will silence warnings from this module, as shown above. | |
115 | ||
116 | =back | |
117 | ||
118 | Because literal conversions also work on empty strings, it may surprise | |
119 | some people: | |
120 | ||
121 | use encoding 'big5'; | |
122 | ||
123 | my $byte_string = pack("C*", 0xA4, 0x40); | |
124 | print length $a; # 2 here. | |
125 | $a .= ""; # concatenating with a unicode string... | |
126 | print length $a; # 1 here! | |
127 | ||
128 | In other words, do not C<use encoding> unless you are certain that the | |
129 | program will not deal with any raw, 8-bit binary data at all. | |
130 | ||
131 | However, the C<Filter =E<gt> 1> flavor of C<use encoding> will I<not> | |
132 | affect implicit upgrading for byte-strings, and is thus incapable of | |
133 | silencing warnings from this module. See L<encoding> for more details. | |
134 | ||
135 | =back | |
136 | ||
137 | =head1 CAVEATS | |
138 | ||
139 | This module currently affects the whole script, instead of inside its | |
140 | lexical block. This is expected to be addressed during Perl 5.9 development, | |
141 | where the B<encoding> module will also be made lexical. | |
142 | ||
143 | =cut | |
144 | ||
145 | # Constants. | |
146 | sub ASCII () { 0 } | |
147 | sub LATIN1 () { 1 } | |
148 | sub FATAL () { 2 } | |
149 | ||
150 | # Install a ${^ENCODING} handler if no other one are already in place. | |
151 | sub import { | |
152 | my $class = shift; | |
153 | my $fatal = shift || ''; | |
154 | ||
155 | local $@; | |
156 | return if ${^ENCODING} and ref(${^ENCODING}) ne $class; | |
157 | return unless eval { require Encode; 1 }; | |
158 | ||
159 | my $ascii = Encode::find_encoding('us-ascii') or return; | |
160 | my $latin1 = Encode::find_encoding('iso-8859-1') or return; | |
161 | ||
162 | # Have to undef explicitly here | |
163 | undef ${^ENCODING}; | |
164 | ||
165 | # Install a warning handler for decode() | |
166 | ${^ENCODING} = bless( | |
167 | [ | |
168 | $ascii, | |
169 | $latin1, | |
170 | (($fatal eq 'FATAL') ? 'Carp::croak' : 'Carp::carp'), | |
171 | ], $class, | |
172 | ); | |
173 | } | |
174 | ||
175 | # Don't worry about source code literals. | |
176 | sub cat_decode { | |
177 | my $self = shift; | |
178 | return $self->[LATIN1]->cat_decode(@_); | |
179 | } | |
180 | ||
181 | # Warn if the data is not purely US-ASCII. | |
182 | sub decode { | |
183 | my $self = shift; | |
184 | ||
185 | local $@; | |
186 | my $rv = eval { $self->[ASCII]->decode($_[0], Encode::FB_CROAK()) }; | |
187 | return $rv unless $@; | |
188 | ||
189 | require Carp; | |
190 | no strict 'refs'; | |
191 | $self->[FATAL]->( | |
192 | "Bytes implicitly upgraded into wide characters as iso-8859-1" | |
193 | ); | |
194 | return $self->[LATIN1]->decode(@_); | |
195 | } | |
196 | ||
197 | sub name { 'iso-8859-1' } | |
198 | ||
199 | 1; | |
200 | ||
201 | __END__ | |
202 | ||
203 | =head1 SEE ALSO | |
204 | ||
205 | L<perlunicode>, L<perluniintro> | |
206 | ||
207 | L<open>, L<utf8>, L<encoding>, L<Encode> | |
208 | ||
209 | =head1 AUTHORS | |
210 | ||
211 | Autrijus Tang E<lt>autrijus@autrijus.orgE<gt> | |
212 | ||
213 | =head1 COPYRIGHT | |
214 | ||
215 | Copyright 2004 by Autrijus Tang E<lt>autrijus@autrijus.orgE<gt>. | |
216 | ||
217 | This program is free software; you can redistribute it and/or modify it | |
218 | under the same terms as Perl itself. | |
219 | ||
220 | See L<http://www.perl.com/perl/misc/Artistic.html> | |
221 | ||
222 | =cut |