Commit | Line | Data |
6b48aaa4 |
1 | # |
2 | # NOTE! This module is deprecated (obsolete) after the Perl release |
3 | # 5.003_06 as the functionality has been integrated into the Perl core. |
4 | # |
5 | |
a0d0e21e |
6 | package I18N::Collate; |
7 | |
f06db76b |
8 | =head1 NAME |
9 | |
69b19ea2 |
10 | I18N::Collate - compare 8-bit scalar data according to the current locale |
f06db76b |
11 | |
12 | =head1 SYNOPSIS |
13 | |
69b19ea2 |
14 | use I18N::Collate; |
f06db76b |
15 | setlocale(LC_COLLATE, 'locale-of-your-choice'); |
69b19ea2 |
16 | $s1 = new I18N::Collate "scalar_data_1"; |
17 | $s2 = new I18N::Collate "scalar_data_2"; |
f06db76b |
18 | |
19 | =head1 DESCRIPTION |
20 | |
21 | This module provides you with objects that will collate |
69b19ea2 |
22 | according to your national character set, provided that the |
23 | POSIX setlocale() function is supported on your system. |
f06db76b |
24 | |
25 | You can compare $s1 and $s2 above with |
26 | |
27 | $s1 le $s2 |
28 | |
29 | to extract the data itself, you'll need a dereference: $$s1 |
30 | |
69b19ea2 |
31 | This uses POSIX::setlocale(). The basic collation conversion is done by |
f06db76b |
32 | strxfrm() which terminates at NUL characters being a decent C routine. |
33 | collate_xfrm() handles embedded NUL characters gracefully. Due to C<cmp> |
34 | and overload magic, C<lt>, C<le>, C<eq>, C<ge>, and C<gt> work also. The |
35 | available locales depend on your operating system; try whether C<locale |
c2960299 |
36 | -a> shows them or man pages for "locale" or "nlsinfo" or |
37 | the direct approach C<ls /usr/lib/nls/loc> or C<ls |
38 | /usr/lib/nls>. Not all the locales that your vendor supports |
39 | are necessarily installed: please consult your operating system's |
69b19ea2 |
40 | documentation and possibly your local system administration. |
c2960299 |
41 | |
42 | The locale names are probably something like |
43 | C<"xx_XX.(ISO)?8859-N"> or C<"xx_XX.(ISO)?8859N">, for example |
44 | C<"fr_CH.ISO8859-1"> is the Swiss (CH) variant of French (fr), |
45 | ISO Latin (8859) 1 (-1) which is the Western European character set. |
f06db76b |
46 | |
47 | =cut |
48 | |
69b19ea2 |
49 | # I18N::Collate.pm |
a0d0e21e |
50 | # |
51 | # Author: Jarkko Hietaniemi <Jarkko.Hietaniemi@hut.fi> |
52 | # Helsinki University of Technology, Finland |
53 | # |
54 | # Acks: Guy Decoux <decoux@moulon.inra.fr> understood |
55 | # overloading magic much deeper than I and told |
56 | # how to cut the size of this code by more than half. |
57 | # (my first version did overload all of lt gt eq le ge cmp) |
58 | # |
59 | # Purpose: compare 8-bit scalar data according to the current locale |
60 | # |
61 | # Requirements: Perl5 POSIX::setlocale() and POSIX::strxfrm() |
62 | # |
63 | # Exports: setlocale 1) |
64 | # collate_xfrm 2) |
65 | # |
66 | # Overloads: cmp # 3) |
67 | # |
69b19ea2 |
68 | # Usage: use I18N::Collate; |
c2960299 |
69 | # setlocale(LC_COLLATE, 'locale-of-your-choice'); # 4) |
69b19ea2 |
70 | # $s1 = new I18N::Collate "scalar_data_1"; |
71 | # $s2 = new I18N::Collate "scalar_data_2"; |
a0d0e21e |
72 | # |
73 | # now you can compare $s1 and $s2: $s1 le $s2 |
74 | # to extract the data itself, you need to deref: $$s1 |
75 | # |
76 | # Notes: |
77 | # 1) this uses POSIX::setlocale |
78 | # 2) the basic collation conversion is done by strxfrm() which |
79 | # terminates at NUL characters being a decent C routine. |
80 | # collate_xfrm handles embedded NUL characters gracefully. |
81 | # 3) due to cmp and overload magic, lt le eq ge gt work also |
82 | # 4) the available locales depend on your operating system; |
c2960299 |
83 | # try whether "locale -a" shows them or man pages for |
84 | # "locale" or "nlsinfo" work or the more direct |
a0d0e21e |
85 | # approach "ls /usr/lib/nls/loc" or "ls /usr/lib/nls". |
c2960299 |
86 | # Not all the locales that your vendor supports |
87 | # are necessarily installed: please consult your |
88 | # operating system's documentation. |
a0d0e21e |
89 | # The locale names are probably something like |
c2960299 |
90 | # 'xx_XX.(ISO)?8859-N' or 'xx_XX.(ISO)?8859N', |
91 | # for example 'fr_CH.ISO8859-1' is the Swiss (CH) |
92 | # variant of French (fr), ISO Latin (8859) 1 (-1) |
93 | # which is the Western European character set. |
a0d0e21e |
94 | # |
6b48aaa4 |
95 | # Updated: 19961005 |
a0d0e21e |
96 | # |
97 | # --- |
98 | |
99 | use POSIX qw(strxfrm LC_COLLATE); |
100 | |
101 | require Exporter; |
102 | |
103 | @ISA = qw(Exporter); |
104 | @EXPORT = qw(collate_xfrm setlocale LC_COLLATE); |
105 | @EXPORT_OK = qw(); |
106 | |
a5f75d66 |
107 | use overload qw( |
a0d0e21e |
108 | fallback 1 |
109 | cmp collate_cmp |
110 | ); |
111 | |
6b48aaa4 |
112 | sub new { |
113 | my $new = $_[1]; |
114 | |
115 | if ($^W && $] >= 5.003_06) { |
116 | unless ($please_use_I18N_Collate_even_if_deprecated) { |
117 | warn <<___EOD___; |
118 | *** |
119 | |
120 | WARNING: starting from the Perl version 5.003_06 the I18N::Collate |
121 | interface for comparing 8-bit scalar data according to the current locale |
122 | |
123 | HAS BEEN DEPRECATED |
124 | |
125 | (that is, please do not use it anymore for any new applications and please |
126 | migrate the old applications away from it) because its functionality |
127 | was integrated into the Perl core language in the release 5.003_06. |
128 | |
129 | All scalar data is now collated according to the current locale setting. |
130 | Also, Perl does automatically the setlocale(LC_COLLATE, "") for you. |
131 | |
132 | To convert: forget I18N::Collate completely and use scalar data in |
133 | a completely normal way. |
134 | |
135 | *** |
136 | ___EOD___ |
137 | $please_use_I18N_Collate_even_if_deprecated++; |
138 | } |
139 | } |
140 | |
141 | bless \$new; |
142 | } |
a0d0e21e |
143 | |
144 | sub setlocale { |
145 | my ($category, $locale) = @_[0,1]; |
146 | |
147 | POSIX::setlocale($category, $locale) if (defined $category); |
148 | # the current $LOCALE |
149 | $LOCALE = $locale || $ENV{'LC_COLLATE'} || $ENV{'LC_ALL'} || ''; |
150 | } |
151 | |
152 | sub C { |
153 | my $s = ${$_[0]}; |
154 | |
155 | $C->{$LOCALE}->{$s} = collate_xfrm($s) |
156 | unless (defined $C->{$LOCALE}->{$s}); # cache when met |
157 | |
158 | $C->{$LOCALE}->{$s}; |
159 | } |
160 | |
161 | sub collate_xfrm { |
162 | my $s = $_[0]; |
163 | my $x = ''; |
164 | |
165 | for (split(/(\000+)/, $s)) { |
166 | $x .= (/^\000/) ? $_ : strxfrm("$_\000"); |
167 | } |
168 | |
169 | $x; |
170 | } |
171 | |
172 | sub collate_cmp { |
173 | &C($_[0]) cmp &C($_[1]); |
174 | } |
175 | |
176 | # init $LOCALE |
177 | |
178 | &I18N::Collate::setlocale(); |
179 | |
180 | 1; # keep require happy |