this repo has no description
Merge pull request #17 from websages/rss_fix
Fix: rss feed
authored by
Stephen Yeargin
and committed by
GitHub
40d651d5
526890d8
+97
-19
+88
-11
htdocs/lib/tumble.pm
+88
-11
htdocs/lib/tumble.pm
···
6
6
7
7
use DBI;
8
8
use POSIX qw( strftime );
9
+
use Time::Local qw( timelocal timegm );
9
10
use Cwd qw( abs_path getcwd );
10
11
use File::Spec;
11
12
···
34
35
$self->{'arg'}->{'dtype'} ||= 'html';
35
36
36
37
for ( $self->{'arg'}->{'dtype'} ) {
37
-
/rss|xml/ && do { $self->header_props( -type => 'text/xml' ); };
38
+
/rss|xml/ && do { $self->header_props( -type => 'text/xml; charset=UTF-8' ); };
38
39
/html/ && do { $self->header_props( -type => 'text/html; charset=UTF-8' ); };
39
40
}
40
41
···
83
84
$data->{$item}->{'timestamp'} =~
84
85
/(\d{4})-(\d{2})-(\d{2})\s(\d{2}):(\d{2}):(\d{2})/
85
86
) {
87
+
# Parse the timestamp: $1=year, $2=month, $3=day, $4=hour, $5=minute, $6=second
88
+
my ($year, $month, $day, $hour, $minute, $second) = ($1, $2, $3, $4, $5, $6);
89
+
90
+
# Convert to epoch time (month is 0-based in timelocal)
91
+
# Note: timelocal interprets time as local time based on server timezone
92
+
my $epoch = timelocal($second, $minute, $hour, $day, $month - 1, $year - 1900);
93
+
94
+
# Format as RFC 822 date (RSS pubDate format)
95
+
# Get localtime components for the timestamp
96
+
my @lt = localtime($epoch);
97
+
98
+
# Calculate timezone offset by comparing GMT and local time
99
+
# timegm returns GMT epoch for given local time components
100
+
my $gmt_epoch = timegm(@lt);
101
+
my $local_epoch = timelocal(@lt);
102
+
my $tz_offset_seconds = $local_epoch - $gmt_epoch;
103
+
my $tz_offset_minutes = $tz_offset_seconds / 60;
104
+
105
+
# Format timezone offset as +HHMM or -HHMM
106
+
my $tz_sign = $tz_offset_minutes >= 0 ? '+' : '-';
107
+
my $tz_hours = abs(int($tz_offset_minutes / 60));
108
+
my $tz_mins = abs(int($tz_offset_minutes % 60));
109
+
my $tz_offset = sprintf("%s%02d%02d", $tz_sign, $tz_hours, $tz_mins);
110
+
86
111
$data->{$item}->{'timestamp'} =
87
-
POSIX::strftime(
88
-
"%a, %d %b %Y %T -0600", 0, $5, $4, $3, $2 - 1, $1 - 1900
89
-
);
112
+
POSIX::strftime("%a, %d %b %Y %H:%M:%S $tz_offset", @lt);
90
113
91
114
if ( ( !$d ) || ( $3 ne $d ) ) {
92
115
$d = $3;
···
153
176
$data->{$item}->{'url'} .
154
177
qq{" alt="image" />};
155
178
};
179
+
180
+
/quote/ && do {
181
+
# For quote items, build description text
182
+
my $quote_text = $data->{$item}->{'quote'} || '';
183
+
my $author_text = $data->{$item}->{'author'} || '';
184
+
# Build the description text - will be escaped in wrap() function
185
+
$content = '"' . $quote_text . '" --' . $author_text;
186
+
};
156
187
}
157
188
158
-
$c .= $self->wrap(
189
+
# For XML/RSS feeds, wrap HTML content in CDATA sections (for ircLink and image items)
190
+
my $xml_content = $content;
191
+
if ( $self->{'arg'}->{'dtype'} =~ /xml|rss/ && defined $content && $content ne '' && $data->{$item}->{'type'} ne 'quote' ) {
192
+
# Wrap HTML content in CDATA for RSS descriptions (quote already has CDATA)
193
+
$xml_content = '<![CDATA[' . $content . ']]>';
194
+
}
195
+
196
+
my %template_vars = (
159
197
wrapper => 'tumble_item_' . $data->{$item}->{'type'},
160
198
author => $data->{$item}->{'user'},
161
-
baseurl => $CONFIG->{'baseurl'},
162
-
content => $content,
163
-
199
+
baseurl => $CONFIG->{'baseurl'},
164
200
%{$data->{$item}}
165
201
);
202
+
203
+
# Add content or description depending on item type
204
+
if ( $data->{$item}->{'type'} eq 'quote' ) {
205
+
# For quote items, pass description (will be escaped in wrap() function)
206
+
$template_vars{'description'} = $content if defined $content;
207
+
} elsif ( defined $xml_content ) {
208
+
$template_vars{'content'} = $xml_content;
209
+
}
210
+
211
+
$c .= $self->wrap( %template_vars );
166
212
}
167
213
168
-
$c =~ s/\&/\&/g if $c;
214
+
# Escape XML special characters for non-CDATA sections
215
+
if ( $self->{'arg'}->{'dtype'} =~ /xml|rss/ && $c ) {
216
+
# Only escape if not already in CDATA sections
217
+
# Escape & first, then other characters
218
+
$c =~ s/&(?!lt;|gt;|amp;|quot;|apos;|#\d+;|#x[0-9a-f]+;)/&/gi;
219
+
# Don't escape < and > that are already in CDATA sections
220
+
# But we need to escape them outside CDATA and in titles/links
221
+
# Since we're wrapping descriptions in CDATA, we only need to escape in titles and links
222
+
# This is handled by the XML escaping function below for non-CDATA content
223
+
} elsif ( $c ) {
224
+
# For HTML output, escape ampersands
225
+
$c =~ s/\&/\&/g;
226
+
}
169
227
170
228
my ( $nav );
171
229
···
346
404
}
347
405
}
348
406
407
+
# Escape XML special characters for text fields in XML/RSS output
408
+
my $is_xml = $self->{'arg'}->{'dtype'} =~ /xml|rss/;
409
+
349
410
map {
350
-
chomp( $arg->{$_} ) if ref $arg->{$_};
351
-
$template->param( $_ => $arg->{$_} );
411
+
my $value = $arg->{$_};
412
+
chomp( $value ) if ref $value;
413
+
414
+
# For XML output, escape text fields (but not content which should be CDATA)
415
+
if ( $is_xml && !ref $value && $_ ne 'content' && $_ ne 'container' ) {
416
+
# Escape XML special characters for titles, links, descriptions, etc.
417
+
# Must escape & first, then < and >
418
+
$value =~ s/&/&/g;
419
+
$value =~ s/</</g;
420
+
$value =~ s/>/>/g;
421
+
# Escape quotes for attribute safety (though we're using in content, not attributes)
422
+
# Also escape quotes in description text to be safe
423
+
if ( $_ eq 'description' ) {
424
+
$value =~ s/"/"/g;
425
+
}
426
+
}
427
+
428
+
$template->param( $_ => $value );
352
429
} keys %{$arg};
353
430
354
431
return $template->output();
+5
-4
htdocs/thtml/index.txml
+5
-4
htdocs/thtml/index.txml
···
1
-
<?xml version="1.0" encoding="iso-8859-1"?>
2
-
<rss version="2.0">
1
+
<?xml version="1.0" encoding="UTF-8"?>
2
+
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
3
3
<channel>
4
4
<title>tumblefish.</title>
5
5
<link>http://<!-- tmpl_var name="baseurl" --></link>
6
+
<atom:link href="https://<!-- tmpl_var name="baseurl" -->/index.xml" rel="self" type="application/rss+xml" />
6
7
<description>tumblefish.</description>
7
8
<language>en-us</language>
8
9
<generator>lsrfsh v2.00</generator>
9
10
10
-
<managingEditor>scott@loserfish.org</managingEditor>
11
+
<managingEditor>scott@loserfish.org (Scott)</managingEditor>
11
12
<copyright>http://creativecommons.org/licenses/by-sa/1.0</copyright>
12
-
<webMaster>scott@loserfish.org</webMaster>
13
+
<webMaster>scott@loserfish.org (Scott)</webMaster>
13
14
<ttl>15</ttl>
14
15
<!-- tmpl_var name="container" -->
15
16
</channel>
+1
-1
htdocs/thtml/tumble_item_image.txml
+1
-1
htdocs/thtml/tumble_item_image.txml
···
1
1
<item>
2
2
<title><!-- tmpl_var name="title" --></title>
3
3
<link><!-- tmpl_var name="url" --></link>
4
-
<guid>tumble-<!-- tmpl_var name="ircLinkID" --></guid>
4
+
<guid isPermaLink="false">tumble-image-<!-- tmpl_var name="imageID" --></guid>
5
5
<description><!-- tmpl_var name="content" --></description>
6
6
<pubDate><!-- tmpl_var name="timestamp" --></pubDate>
7
7
</item>
+1
-1
htdocs/thtml/tumble_item_ircLink.txml
+1
-1
htdocs/thtml/tumble_item_ircLink.txml
···
1
1
<item>
2
2
<title><!-- tmpl_var name="title" --></title>
3
3
<link>http://<!-- tmpl_var name="baseurl" -->/irclink/?<!-- tmpl_var name="ircLinkID" --></link>
4
-
<guid>tumble-<!-- tmpl_var name="ircLinkID" --></guid>
4
+
<guid isPermaLink="false">tumble-<!-- tmpl_var name="ircLinkID" --></guid>
5
5
<description><!-- tmpl_var name="content" --></description>
6
6
<pubDate><!-- tmpl_var name="timestamp" --></pubDate>
7
7
</item>
+2
-2
htdocs/thtml/tumble_item_quote.txml
+2
-2
htdocs/thtml/tumble_item_quote.txml
···
1
1
<item>
2
2
<title><!-- tmpl_var name="quote" --></title>
3
3
<link>http://<!-- tmpl_var name="baseurl" -->/</link>
4
-
<guid>tumble-<!-- tmpl_var name="ircLinkID" --></guid>
5
-
<description>"<!-- tmpl_var name="quote" -->" --<!-- tmpl_var name="author" --></description>
4
+
<guid isPermaLink="false">tumble-quote-<!-- tmpl_var name="quoteID" --></guid>
5
+
<description><!-- tmpl_var name="description" --></description>
6
6
<pubDate><!-- tmpl_var name="timestamp" --></pubDate>
7
7
</item>