| 1 |
wakaba |
1.1 |
|
| 2 |
|
|
=head1 NAME |
| 3 |
|
|
|
| 4 |
|
|
Message::Field::CSV Perl module |
| 5 |
|
|
|
| 6 |
|
|
=head1 DESCRIPTION |
| 7 |
|
|
|
| 8 |
|
|
Perl module for comma separated C<field>. |
| 9 |
|
|
|
| 10 |
|
|
This module supports a number of fields that contains |
| 11 |
|
|
(or does not contain:-)) of comma separated values, |
| 12 |
wakaba |
1.2 |
such as C<Keywords:>, C<Newsgroups:> and so on. |
| 13 |
wakaba |
1.1 |
|
| 14 |
|
|
=cut |
| 15 |
|
|
|
| 16 |
|
|
package Message::Field::CSV; |
| 17 |
|
|
require 5.6.0; |
| 18 |
|
|
use strict; |
| 19 |
|
|
use re 'eval'; |
| 20 |
|
|
use vars qw(%OPTION %REG $VERSION); |
| 21 |
wakaba |
1.6 |
$VERSION=do{my @r=(q$Revision: 1.5 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r}; |
| 22 |
wakaba |
1.5 |
use overload '@{}' => sub {shift->value}, |
| 23 |
wakaba |
1.1 |
'""' => sub {shift->stringify}; |
| 24 |
wakaba |
1.4 |
require Message::Util; |
| 25 |
wakaba |
1.1 |
$REG{comment} = qr/\x28(?:\x5C[\x00-\xFF]|[\x00-\x0C\x0E-\x27\x2A-\x5B\x5D-\xFF]+|(??{$REG{comment}}))*\x29/; |
| 26 |
|
|
$REG{quoted_string} = qr/\x22(?:\x5C[\x00-\xFF]|[\x00-\x0C\x0E-\x21\x23-\x5B\x5D-\xFF])*\x22/; |
| 27 |
|
|
$REG{domain_literal} = qr/\x5B(?:\x5C[\x00-\xFF]|[\x00-\x0C\x0E-\x5A\x5E-\xFF])*\x5D/; |
| 28 |
wakaba |
1.4 |
$REG{uri_literal} = qr/\x3C[\x09\x20\x21\x23-\x3B\x3D\x3F-\x5B\x5D\x5F\x61-\x7A\x7E]*\x3E/; |
| 29 |
wakaba |
1.1 |
|
| 30 |
|
|
$REG{WSP} = qr/[\x20\x09]+/; |
| 31 |
|
|
$REG{FWS} = qr/[\x20\x09]*/; |
| 32 |
|
|
$REG{atext} = qr/[\x21\x23-\x27\x2A\x2B\x2D\x2F\x30-\x39\x3D\x3F\x41-\x5A\x5E-\x7E]+/; |
| 33 |
|
|
$REG{dot_atom} = qr/$REG{atext}(?:$REG{FWS}\x2E$REG{FWS}$REG{atext})*/; |
| 34 |
|
|
$REG{dot_word} = qr/(?:$REG{atext}|$REG{quoted_string})(?:$REG{FWS}\x2E$REG{FWS}(?:$REG{atext}|$REG{quoted_string}))*/; |
| 35 |
|
|
$REG{phrase} = qr/(?:$REG{atext}|$REG{quoted_string})(?:$REG{atext}|$REG{quoted_string}|\.|$REG{FWS})*/; |
| 36 |
|
|
$REG{M_quoted_string} = qr/\x22((?:\x5C[\x00-\xFF]|[\x00-\x0C\x0E-\x21\x23-\x5B\x5D-\xFF])*)\x22/; |
| 37 |
|
|
$REG{NON_atom} = qr/[^\x09\x21\x23-\x27\x2A\x2B\x2D\x2F\x30-\x39\x3D\x3F\x41-\x5A\x5E-\x7E\x2E]/; |
| 38 |
wakaba |
1.4 |
$REG{NON_atext} = qr/[^\x21\x23-\x27\x2A\x2B\x2D\x2F\x30-\x39\x3D\x3F\x41-\x5A\x5E-\x7E]/; |
| 39 |
|
|
$REG{NON_atext_dot} = qr/[^\x21\x23-\x27\x2A\x2B\x2D-\x39\x3D\x3F\x41-\x5A\x5E-\x7E]/; |
| 40 |
|
|
$REG{NON_atext_dot_wsp} = qr/[^\x09\x20\x21\x23-\x27\x2A\x2B\x2D-\x39\x3D\x3F\x41-\x5A\x5E-\x7E]/; |
| 41 |
|
|
$REG{NON_http_token_wsp} = qr/[^\x09\x20\x21\x23-\x27\x2A\x2B\x2D\x2E\x30-\x39\x41-\x5A\x5E-\x7A\x7C\x7E]/; |
| 42 |
|
|
$REG{NON_component} = qr/[^\x2B\x2D\x2E\x30-\x39\x41-\x5A\x5F\x61-\x7A\x80-\xFF\x2F\x3D\x3F]/; |
| 43 |
|
|
$REG{NON_distribution} = qr/[^\x21\x2B\x2D\x30-\x39\x41-\x5A\x5F\x61-\x7A]/; |
| 44 |
wakaba |
1.3 |
$REG{S_encoded_word} = qr/=\x3F$REG{atext_dot}\x3F=/; |
| 45 |
wakaba |
1.1 |
|
| 46 |
|
|
## Keywords: foo, bar, "and so on" |
| 47 |
|
|
## Newsgroups: local.test,local.foo,local.bar |
| 48 |
|
|
## Accept: text/html; q=1.0, text/plain; q=0.03; *; q=0.01 |
| 49 |
|
|
|
| 50 |
|
|
%OPTION = ( |
| 51 |
|
|
field_name => 'keywords', |
| 52 |
wakaba |
1.3 |
encoding_after_encode => '*default', |
| 53 |
|
|
encoding_before_decode => '*default', |
| 54 |
|
|
hook_encode_string => #sub {shift; (value => shift, @_)}, |
| 55 |
|
|
\&Message::Util::encode_header_string, |
| 56 |
|
|
hook_decode_string => #sub {shift; (value => shift, @_)}, |
| 57 |
|
|
\&Message::Util::decode_header_string, |
| 58 |
wakaba |
1.2 |
is_quoted_string => 1, ## Can itself quoted-string? |
| 59 |
wakaba |
1.4 |
long_count => 10, |
| 60 |
|
|
remove_comment => 1, |
| 61 |
wakaba |
1.1 |
separator => ', ', |
| 62 |
wakaba |
1.4 |
separator_long => ', ', |
| 63 |
wakaba |
1.1 |
max => -1, |
| 64 |
wakaba |
1.3 |
value_type => [':none:'], |
| 65 |
wakaba |
1.4 |
value_unsafe_rule => 'NON_http_token_wsp', |
| 66 |
wakaba |
1.1 |
); |
| 67 |
|
|
|
| 68 |
wakaba |
1.4 |
sub _init_option ($) { |
| 69 |
wakaba |
1.1 |
my $self = shift; |
| 70 |
|
|
my %field_type = qw(accept-charset accept accept-encoding accept |
| 71 |
|
|
accept-language accept |
| 72 |
wakaba |
1.2 |
content-language keywords |
| 73 |
wakaba |
1.1 |
followup-to newsgroups |
| 74 |
wakaba |
1.4 |
list-archive list- list-digest list- list-help list- |
| 75 |
|
|
list-owner list- list-post list- list-subscribe list- |
| 76 |
|
|
list-unsubscribe list- list-url list- uri list- |
| 77 |
wakaba |
1.6 |
posted-to newsgroups |
| 78 |
wakaba |
1.2 |
x-brother x-moe x-daughter x-moe |
| 79 |
|
|
x-respect x-moe x-syster x-moe x-wife x-moe); |
| 80 |
wakaba |
1.6 |
my $field_name = lc $self->{option}->{field_name} |
| 81 |
|
|
|| lc $self->{option}->{-field_name}; ## AD HOC |
| 82 |
wakaba |
1.1 |
$field_name = $field_type{$field_name} || $field_name; |
| 83 |
wakaba |
1.2 |
if ($field_name eq 'newsgroups') { |
| 84 |
|
|
$self->{option}->{separator} = ','; |
| 85 |
wakaba |
1.4 |
$self->{option}->{separator_long} = ', '; |
| 86 |
|
|
$self->{option}->{long_count} = 5; |
| 87 |
|
|
$self->{option}->{value_unsafe_rule} = 'NON_component'; |
| 88 |
wakaba |
1.5 |
$self->{option}->{encoding_after_encode} = 'utf-8'; |
| 89 |
wakaba |
1.4 |
} elsif ($field_name eq 'distribution') { |
| 90 |
|
|
$self->{option}->{separator} = ','; |
| 91 |
|
|
$self->{option}->{separator_long} = ', '; |
| 92 |
|
|
$self->{option}->{long_count} = 15; |
| 93 |
|
|
$self->{option}->{value_unsafe_rule} = 'NON_distribution'; |
| 94 |
wakaba |
1.2 |
} elsif ($field_name eq 'x-moe') { |
| 95 |
wakaba |
1.1 |
$self->{option}->{is_quoted_string} = -1; |
| 96 |
wakaba |
1.4 |
$self->{option}->{value_type} = ['Message::Field::ValueParams', |
| 97 |
|
|
{format => $self->{option}->{format}}]; |
| 98 |
wakaba |
1.1 |
} elsif ($field_name eq 'accept') { |
| 99 |
|
|
$self->{option}->{is_quoted_string} = -1; |
| 100 |
wakaba |
1.4 |
$self->{option}->{value_type} = ['Message::Field::ValueParams', |
| 101 |
|
|
{format => $self->{option}->{format}}]; |
| 102 |
|
|
} elsif ($field_name eq 'list-') { |
| 103 |
|
|
$self->{option}->{is_quoted_string} = -1; |
| 104 |
|
|
$self->{option}->{remove_comment} = -1; |
| 105 |
|
|
$self->{option}->{value_type} = ['Message::Field::URI', |
| 106 |
|
|
{field_name => $self->{option}->{field_name}, |
| 107 |
|
|
format => $self->{option}->{format}}]; |
| 108 |
wakaba |
1.1 |
} elsif ($field_name eq 'encrypted') { |
| 109 |
|
|
$self->{option}->{max} = 2; |
| 110 |
|
|
} |
| 111 |
|
|
$self; |
| 112 |
|
|
} |
| 113 |
|
|
|
| 114 |
|
|
=head2 Message::Field::CSV->new () |
| 115 |
|
|
|
| 116 |
|
|
Returns new CSV field body. |
| 117 |
|
|
|
| 118 |
|
|
=cut |
| 119 |
|
|
|
| 120 |
|
|
sub new ($;%) { |
| 121 |
|
|
my $self = bless {}, shift; |
| 122 |
|
|
my %option = @_; |
| 123 |
|
|
for (%OPTION) {$option{$_} ||= $OPTION{$_}} |
| 124 |
|
|
$self->{option} = \%option; |
| 125 |
wakaba |
1.4 |
$self->_init_option (); |
| 126 |
wakaba |
1.1 |
$self; |
| 127 |
|
|
} |
| 128 |
|
|
|
| 129 |
|
|
=head2 Message::Field::CSV->parse ($unfolded_field_body) |
| 130 |
|
|
|
| 131 |
|
|
Parses C<field-body>. |
| 132 |
|
|
|
| 133 |
|
|
=cut |
| 134 |
|
|
|
| 135 |
|
|
sub parse ($$;%) { |
| 136 |
|
|
my $self = bless {}, shift; |
| 137 |
|
|
my $field_body = shift; |
| 138 |
|
|
my %option = @_; |
| 139 |
|
|
for (%OPTION) {$option{$_} ||= $OPTION{$_}} |
| 140 |
|
|
$self->{option} = \%option; |
| 141 |
wakaba |
1.4 |
$self->_init_option (); |
| 142 |
|
|
$field_body = $self->_delete_comment ($field_body) |
| 143 |
|
|
unless $option{remove_comment}<0; |
| 144 |
wakaba |
1.1 |
@{$self->{value}} = $self->_parse_list ($field_body); |
| 145 |
|
|
$self; |
| 146 |
|
|
} |
| 147 |
|
|
|
| 148 |
|
|
sub _parse_list ($$) { |
| 149 |
|
|
my $self = shift; |
| 150 |
|
|
my $fb = shift; |
| 151 |
|
|
my @ids; |
| 152 |
wakaba |
1.4 |
$fb =~ s{((?:$REG{quoted_string}|$REG{uri_literal}|$REG{domain_literal}|$REG{comment}|[^\x22\x28\x2C\x3C\x5B])+)}{ |
| 153 |
wakaba |
1.1 |
my $s = $1; $s =~ s/^$REG{WSP}+//; $s =~ s/$REG{WSP}+$//; |
| 154 |
|
|
if ($self->{option}->{is_quoted_string}>0) { |
| 155 |
wakaba |
1.3 |
push @ids, $self->_value ($self->_decode_quoted_string ($s)); |
| 156 |
wakaba |
1.1 |
} else { |
| 157 |
wakaba |
1.2 |
push @ids, $self->_value ($s); |
| 158 |
wakaba |
1.1 |
} |
| 159 |
|
|
}goex; |
| 160 |
|
|
@ids; |
| 161 |
|
|
} |
| 162 |
|
|
|
| 163 |
|
|
=head2 $self->value () |
| 164 |
|
|
|
| 165 |
wakaba |
1.5 |
Returns array reference to value list. |
| 166 |
wakaba |
1.1 |
|
| 167 |
|
|
=cut |
| 168 |
|
|
|
| 169 |
wakaba |
1.5 |
sub value ($) {shift->{value}} |
| 170 |
wakaba |
1.1 |
|
| 171 |
|
|
=head2 $self->add ($value, [%option]) |
| 172 |
|
|
|
| 173 |
|
|
Adds new value. |
| 174 |
|
|
|
| 175 |
|
|
=cut |
| 176 |
|
|
|
| 177 |
|
|
sub add ($;$%) { |
| 178 |
|
|
my $self = shift; |
| 179 |
|
|
my ($value, %option) = @_; |
| 180 |
wakaba |
1.2 |
push @{$self->{value}}, $self->_value ($value); |
| 181 |
|
|
$value; |
| 182 |
|
|
} |
| 183 |
|
|
|
| 184 |
|
|
## Hook called before returning C<value>. |
| 185 |
|
|
## $self->_param_value ($name, $value); |
| 186 |
|
|
sub _value ($$) { |
| 187 |
|
|
my $self = shift; |
| 188 |
|
|
my $value = shift; |
| 189 |
|
|
my $vtype = $self->{option}->{value_type}->[0]; |
| 190 |
|
|
my %vopt; %vopt = %{$self->{option}->{value_type}->[1]} |
| 191 |
|
|
if ref $self->{option}->{value_type}->[1]; |
| 192 |
|
|
if (ref $value) { |
| 193 |
|
|
return $value; |
| 194 |
|
|
} elsif ($vtype eq ':none:') { |
| 195 |
|
|
return $value; |
| 196 |
|
|
} elsif ($value) { |
| 197 |
|
|
eval "require $vtype"; |
| 198 |
|
|
return $vtype->parse ($value, %vopt); |
| 199 |
|
|
} else { |
| 200 |
|
|
eval "require $vtype"; |
| 201 |
|
|
return $vtype->new (%vopt); |
| 202 |
|
|
} |
| 203 |
wakaba |
1.1 |
} |
| 204 |
|
|
|
| 205 |
|
|
sub stringify ($;%) { |
| 206 |
|
|
my $self = shift; |
| 207 |
|
|
my %option = @_; |
| 208 |
|
|
$option{separator} ||= $self->{option}->{separator}; |
| 209 |
wakaba |
1.4 |
$option{separator_long} ||= $self->{option}->{separator_long}; |
| 210 |
|
|
$option{long_count} ||= $self->{option}->{long_count}; |
| 211 |
wakaba |
1.1 |
$option{max} ||= $self->{option}->{max}; |
| 212 |
|
|
$option{is_quoted_string} ||= $self->{option}->{is_quoted_string}; |
| 213 |
wakaba |
1.4 |
$option{value_unsafe_rule} ||= $self->{option}->{value_unsafe_rule}; |
| 214 |
wakaba |
1.1 |
$self->_delete_empty (); |
| 215 |
|
|
$option{max}--; |
| 216 |
|
|
$option{max} = $#{$self->{value}} if $option{max}<0; |
| 217 |
|
|
$option{max} = $#{$self->{value}} if $#{$self->{value}}<$option{max}; |
| 218 |
wakaba |
1.4 |
$option{separator} = $option{separator_long} |
| 219 |
|
|
if $option{max} >= $option{long_count}; |
| 220 |
wakaba |
1.1 |
join $option{separator}, |
| 221 |
wakaba |
1.3 |
map { |
| 222 |
|
|
if ($option{is_quoted_string}>0) { |
| 223 |
|
|
my %s = &{$self->{option}->{hook_encode_string}} ($self, |
| 224 |
|
|
$_, type => 'phrase'); |
| 225 |
wakaba |
1.4 |
$self->_quote_unsafe_string ($s{value}, |
| 226 |
|
|
unsafe => $option{value_unsafe_rule}); |
| 227 |
wakaba |
1.3 |
} else { |
| 228 |
|
|
$_; |
| 229 |
|
|
} |
| 230 |
|
|
} @{$self->{value}}[0..$option{max}]; |
| 231 |
wakaba |
1.1 |
} |
| 232 |
|
|
|
| 233 |
wakaba |
1.4 |
=head2 $self->option ($option_name, [$option_value]) |
| 234 |
|
|
|
| 235 |
|
|
Set/gets new value of the option. |
| 236 |
|
|
|
| 237 |
|
|
=cut |
| 238 |
|
|
|
| 239 |
|
|
sub option ($$;$) { |
| 240 |
|
|
my $self = shift; |
| 241 |
|
|
my ($name, $value) = @_; |
| 242 |
|
|
if (defined $value) { |
| 243 |
|
|
$self->{option}->{$name} = $value; |
| 244 |
|
|
} |
| 245 |
|
|
$self->{option}->{$name}; |
| 246 |
|
|
} |
| 247 |
|
|
|
| 248 |
wakaba |
1.5 |
sub value_type ($;$%) { |
| 249 |
|
|
my $self = shift; |
| 250 |
|
|
my $new_value_type = shift; |
| 251 |
|
|
if ($new_value_type) { |
| 252 |
|
|
$self->{option}->{value_type}->[0] = $new_value_type; |
| 253 |
|
|
} |
| 254 |
|
|
$self->{option}->{value_type}->[0] || ':none:'; |
| 255 |
|
|
} |
| 256 |
|
|
|
| 257 |
wakaba |
1.1 |
sub _delete_empty ($) { |
| 258 |
|
|
my $self = shift; |
| 259 |
|
|
my @nid; |
| 260 |
|
|
for my $id (@{$self->{value}}) {push @nid, $id if length $id} |
| 261 |
|
|
$self->{value} = \@nid; |
| 262 |
|
|
} |
| 263 |
|
|
|
| 264 |
wakaba |
1.4 |
sub _quote_unsafe_string ($$;%) { |
| 265 |
wakaba |
1.1 |
my $self = shift; |
| 266 |
|
|
my $string = shift; |
| 267 |
wakaba |
1.4 |
my %option = @_; |
| 268 |
|
|
$option{unsafe} ||= 'NON_atext_dot'; |
| 269 |
|
|
if ($string =~ /$REG{$option{unsafe}}/ || $string =~ /$REG{WSP}$REG{WSP}+/) { |
| 270 |
|
|
$string =~ s/([\x22\x5C])([\x21-\x7E])?/"\x5C$1".(defined $2?"\x5C$2":'')/ge; |
| 271 |
wakaba |
1.1 |
$string = '"'.$string.'"'; |
| 272 |
|
|
} |
| 273 |
|
|
$string; |
| 274 |
|
|
} |
| 275 |
|
|
|
| 276 |
|
|
|
| 277 |
|
|
=head2 $self->_unquote_quoted_string ($string) |
| 278 |
|
|
|
| 279 |
|
|
Unquote C<quoted-string>. Get rid of C<DQUOTE>s and |
| 280 |
|
|
C<REVERSED SOLIDUS> included in C<quoted-pair>. |
| 281 |
|
|
This method is intended for internal use. |
| 282 |
|
|
|
| 283 |
|
|
=cut |
| 284 |
|
|
|
| 285 |
|
|
sub _unquote_quoted_string ($$) { |
| 286 |
|
|
my $self = shift; |
| 287 |
|
|
my $quoted_string = shift; |
| 288 |
|
|
$quoted_string =~ s{$REG{M_quoted_string}}{ |
| 289 |
|
|
my $qtext = $1; |
| 290 |
|
|
$qtext =~ s/\x5C([\x00-\xFF])/$1/g; |
| 291 |
|
|
$qtext; |
| 292 |
|
|
}goex; |
| 293 |
|
|
$quoted_string; |
| 294 |
|
|
} |
| 295 |
|
|
|
| 296 |
wakaba |
1.3 |
sub _decode_quoted_string ($$) { |
| 297 |
|
|
my $self = shift; |
| 298 |
|
|
my $quoted_string = shift; |
| 299 |
|
|
$quoted_string =~ s{$REG{M_quoted_string}|([^\x22]+)}{ |
| 300 |
|
|
my ($qtext,$t) = ($1, $2); |
| 301 |
|
|
if ($t) { |
| 302 |
|
|
my %s = &{$self->{option}->{hook_decode_string}} ($self, $t, |
| 303 |
|
|
type => 'phrase'); |
| 304 |
|
|
$s{value}; |
| 305 |
|
|
} else { |
| 306 |
|
|
$qtext =~ s/\x5C([\x00-\xFF])/$1/g; |
| 307 |
|
|
my %s = &{$self->{option}->{hook_decode_string}} ($self, $qtext, |
| 308 |
|
|
type => 'phrase/quoted'); |
| 309 |
|
|
$s{value}; |
| 310 |
|
|
} |
| 311 |
|
|
}goex; |
| 312 |
|
|
$quoted_string; |
| 313 |
|
|
} |
| 314 |
|
|
|
| 315 |
wakaba |
1.1 |
=head2 $self->_delete_comment ($field_body) |
| 316 |
|
|
|
| 317 |
|
|
Remove all C<comment> in given strictured C<field-body>. |
| 318 |
|
|
This method is intended to be used for internal process. |
| 319 |
|
|
|
| 320 |
|
|
=cut |
| 321 |
|
|
|
| 322 |
|
|
sub _delete_comment ($$) { |
| 323 |
|
|
my $self = shift; |
| 324 |
|
|
my $body = shift; |
| 325 |
wakaba |
1.4 |
$body =~ s{($REG{quoted_string}|$REG{uri_literal}|$REG{domain_literal})|$REG{comment}}{ |
| 326 |
wakaba |
1.1 |
my $o = $1; $o? $o : ' '; |
| 327 |
|
|
}gex; |
| 328 |
|
|
$body; |
| 329 |
|
|
} |
| 330 |
|
|
|
| 331 |
|
|
=head1 EXAMPLE |
| 332 |
|
|
|
| 333 |
|
|
|
| 334 |
|
|
=head1 LICENSE |
| 335 |
|
|
|
| 336 |
|
|
Copyright 2002 wakaba E<lt>w@suika.fam.cxE<gt>. |
| 337 |
|
|
|
| 338 |
|
|
This program is free software; you can redistribute it and/or modify |
| 339 |
|
|
it under the terms of the GNU General Public License as published by |
| 340 |
|
|
the Free Software Foundation; either version 2 of the License, or |
| 341 |
|
|
(at your option) any later version. |
| 342 |
|
|
|
| 343 |
|
|
This program is distributed in the hope that it will be useful, |
| 344 |
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 345 |
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 346 |
|
|
GNU General Public License for more details. |
| 347 |
|
|
|
| 348 |
|
|
You should have received a copy of the GNU General Public License |
| 349 |
|
|
along with this program; see the file COPYING. If not, write to |
| 350 |
|
|
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, |
| 351 |
|
|
Boston, MA 02111-1307, USA. |
| 352 |
|
|
|
| 353 |
|
|
=head1 CHANGE |
| 354 |
|
|
|
| 355 |
|
|
See F<ChangeLog>. |
| 356 |
wakaba |
1.6 |
$Date: 2002/04/01 05:32:15 $ |
| 357 |
wakaba |
1.1 |
|
| 358 |
|
|
=cut |
| 359 |
|
|
|
| 360 |
|
|
1; |