2 # Blog aggregation plugin.
3 package IkiWiki::Plugin::aggregate;
12 use open qw{:utf8 :std};
18 hook(type => "getopt", id => "aggregate", call => \&getopt);
19 hook(type => "checkconfig", id => "aggregate", call => \&checkconfig);
20 hook(type => "filter", id => "aggregate", call => \&filter);
21 hook(type => "preprocess", id => "aggregate", call => \&preprocess);
22 hook(type => "delete", id => "aggregate", call => \&delete);
23 hook(type => "savestate", id => "aggregate", call => \&savestate);
27 eval q{use Getopt::Long};
29 Getopt::Long::Configure('pass_through');
30 GetOptions("aggregate" => \$config{aggregate});
33 sub checkconfig () { #{{{
34 if ($config{aggregate} && ! ($config{post_commit} &&
35 IkiWiki::commit_hook_enabled())) {
36 # don't wait for the lock
37 IkiWiki::lockwiki(0) || exit 1;
45 IkiWiki::unlockwiki();
51 my $page=$params{page};
53 loadstate(); # if not already loaded
54 # Mark all feeds originating on this page as removable;
55 # preprocess will unmark those that still exist.
58 return $params{content};
61 sub preprocess (@) { #{{{
64 foreach my $required (qw{name url}) {
65 if (! exists $params{$required}) {
66 return "[[aggregate ".sprintf(gettext("missing %s parameter"), $required)."]]";
71 my $name=$params{name};
72 if (exists $feeds{$name}) {
79 $feed->{sourcepage}=$params{page};
80 $feed->{url}=$params{url};
81 my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".IkiWiki::titlepage($params{name});
83 ($dir)=$dir=~/$config{wiki_file_regexp}/;
85 $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
86 $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
87 $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
88 $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
89 delete $feed->{remove};
90 delete $feed->{expired};
91 $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
92 $feed->{numposts}=0 unless defined $feed->{numposts};
93 $feed->{newposts}=0 unless defined $feed->{newposts};
94 $feed->{message}=gettext("new feed") unless defined $feed->{message};
95 $feed->{error}=0 unless defined $feed->{error};
101 push @{$feed->{tags}}, $value;
105 return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
106 ($feed->{error} ? "<em>" : "").$feed->{message}.
107 ($feed->{error} ? "</em>" : "").
108 " (".$feed->{numposts}." ".gettext("posts").
109 ($feed->{newposts} ? "; ".$feed->{newposts}.
110 " ".gettext("new") : "").
114 sub delete (@) { #{{{
117 # Remove feed data for removed pages.
118 foreach my $file (@files) {
119 my $page=pagename($file);
125 sub loadstate () { #{{{
126 return if $state_loaded;
127 if (-e "$config{wikistatedir}/aggregate") {
128 open (IN, "$config{wikistatedir}/aggregate" ||
129 die "$config{wikistatedir}/aggregate: $!");
131 $_=IkiWiki::possibly_foolish_untaint($_);
134 foreach my $i (split(/ /, $_)) {
135 my ($field, $val)=split(/=/, $i, 2);
136 if ($field eq "name" || $field eq "feed" ||
137 $field eq "guid" || $field eq "message") {
138 $data->{$field}=decode_entities($val, " \t\n");
140 elsif ($field eq "tag") {
141 push @{$data->{tags}}, $val;
144 $data->{$field}=$val;
148 if (exists $data->{name}) {
149 $feeds{$data->{name}}=$data;
151 elsif (exists $data->{guid}) {
152 $guids{$data->{guid}}=$data;
162 sub savestate () { #{{{
163 eval q{use HTML::Entities};
165 my $newfile="$config{wikistatedir}/aggregate.new";
166 # TODO: This cleanup function could use improvement. Any newly
167 # aggregated files are left behind unrecorded, and should be deleted.
168 my $cleanup = sub { unlink($newfile) };
169 open (OUT, ">$newfile") || error("open $newfile: $!", $cleanup);
170 foreach my $data (values %feeds, values %guids) {
171 if ($data->{remove}) {
173 foreach my $guid (values %guids) {
174 if ($guid->{feed} eq $data->{name}) {
180 unlink pagefile($data->{page});
184 elsif ($data->{expired} && exists $data->{page}) {
185 unlink pagefile($data->{page});
186 delete $data->{page};
191 foreach my $field (keys %$data) {
192 if ($field eq "name" || $field eq "feed" ||
193 $field eq "guid" || $field eq "message") {
194 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
196 elsif ($field eq "tags") {
197 push @line, "tag=$_" foreach @{$data->{tags}};
200 push @line, "$field=".$data->{$field};
203 print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
205 close OUT || error("save $newfile: $!", $cleanup);
206 rename($newfile, "$config{wikistatedir}/aggregate") ||
207 error("rename $newfile: $!", $cleanup);
211 foreach my $feed (values %feeds) {
212 next unless $feed->{expireage} || $feed->{expirecount};
214 foreach my $item (sort { $IkiWiki::pagectime{$b->{page}} <=> $IkiWiki::pagectime{$a->{page}} }
215 grep { exists $_->{page} && $_->{feed} eq $feed->{name} && $IkiWiki::pagectime{$_->{page}} }
217 if ($feed->{expireage}) {
218 my $days_old = (time - $IkiWiki::pagectime{$item->{page}}) / 60 / 60 / 24;
219 if ($days_old > $feed->{expireage}) {
220 debug(sprintf(gettext("expiring %s (%s days old)"),
221 $item->{page}, $days_old));
225 elsif ($feed->{expirecount} &&
226 $count >= $feed->{expirecount}) {
227 debug(sprintf(gettext("expiring %s"), $item->{page}));
237 sub aggregate () { #{{{
238 eval q{use XML::Feed};
240 eval q{use URI::Fetch};
242 eval q{use HTML::Entities};
245 foreach my $feed (values %feeds) {
246 next unless $config{rebuild} ||
247 time - $feed->{lastupdate} >= $feed->{updateinterval};
248 $feed->{lastupdate}=time;
250 $feed->{message}=sprintf(gettext("processed ok at %s"),
251 displaytime($feed->{lastupdate}));
253 $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
255 debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
257 if (! length $feed->{feedurl}) {
258 my @urls=XML::Feed->find_feeds($feed->{url});
260 $feed->{message}=sprintf(gettext("could not find feed at %s"), $feed->{url});
262 debug($feed->{message});
265 $feed->{feedurl}=pop @urls;
267 my $res=URI::Fetch->fetch($feed->{feedurl});
269 $feed->{message}=URI::Fetch->errstr;
271 debug($feed->{message});
274 if ($res->status == URI::Fetch::URI_GONE()) {
275 $feed->{message}=gettext("feed not found");
277 debug($feed->{message});
280 my $content=$res->content;
281 my $f=eval{XML::Feed->parse(\$content)};
283 # One common cause of XML::Feed crashing is a feed
284 # that contains invalid UTF-8 sequences. Convert
285 # feed to ascii to try to work around.
286 $feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)"));
287 $content=Encode::decode_utf8($content);
288 $f=eval{XML::Feed->parse(\$content)};
291 $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
293 debug($feed->{message});
297 $feed->{message}=XML::Feed->errstr;
299 debug($feed->{message});
303 foreach my $entry ($f->entries) {
306 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
307 link => $entry->link,
308 content => defined $entry->content->body ? $entry->content->body : "",
309 guid => defined $entry->id ? $entry->id : time."_".$feed->name,
310 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
316 sub add_page (@) { #{{{
319 my $feed=$params{feed};
322 if (exists $guids{$params{guid}}) {
323 # updating an existing post
324 $guid=$guids{$params{guid}};
325 return if $guid->{expired};
329 $guid->{guid}=$params{guid};
330 $guids{$params{guid}}=$guid;
331 $mtime=$params{ctime};
335 # assign it an unused page
336 my $page=IkiWiki::titlepage($params{title});
337 # escape slashes and periods in title so it doesn't specify
338 # directory name or trigger ".." disallowing code.
339 $page=~s!([/.])!"__".ord($1)."__"!eg;
340 $page=$feed->{dir}."/".$page;
341 ($page)=$page=~/$config{wiki_file_regexp}/;
342 if (! defined $page || ! length $page) {
343 $page=$feed->{dir}."/item";
346 while (exists $IkiWiki::pagecase{lc $page.$c} ||
347 -e pagefile($page.$c)) {
351 # Make sure that the file name isn't too long.
352 # NB: This doesn't check for path length limits.
353 my $max=POSIX::pathconf($config{srcdir}, &POSIX::_PC_NAME_MAX);
354 if (defined $max && length(htmlfn($page)) >= $max) {
356 $page=$feed->{dir}."/item";
357 while (exists $IkiWiki::pagecase{lc $page.$c} ||
358 -e pagefile($page.$c)) {
364 debug(sprintf(gettext("creating new page %s"), $page));
366 $guid->{feed}=$feed->{name};
368 # To write or not to write? Need to avoid writing unchanged pages
369 # to avoid unneccessary rebuilding. The mtime from rss cannot be
370 # trusted; let's use a digest.
371 eval q{use Digest::MD5 'md5_hex'};
374 my $digest=md5_hex(Encode::encode_utf8($params{content}));
375 return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
376 $guid->{md5}=$digest;
379 my $template=template("aggregatepost.tmpl", blind_cache => 1);
380 $template->param(title => $params{title})
381 if defined $params{title} && length($params{title});
382 $template->param(content => htmlescape(htmlabs($params{content}, $feed->{feedurl})));
383 $template->param(name => $feed->{name});
384 $template->param(url => $feed->{url});
385 $template->param(permalink => urlabs($params{link}, $feed->{feedurl}))
386 if defined $params{link};
387 if (ref $feed->{tags}) {
388 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
390 writefile(htmlfn($guid->{page}), $config{srcdir},
393 # Set the mtime, this lets the build process get the right creation
394 # time on record for the new page.
395 utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
398 sub htmlescape ($) { #{{{
399 # escape accidental wikilinks and preprocessor stuff
401 $html=~s/(?<!\\)\[\[/\\\[\[/g;
405 sub urlabs ($$) { #{{{
409 URI->new_abs($url, $urlbase)->as_string;
412 sub htmlabs ($$) { #{{{
413 # Convert links in html from relative to absolute.
414 # Note that this is a heuristic, which is not specified by the rss
415 # spec and may not be right for all feeds. Also, see Debian
421 my $p = HTML::Parser->new(api_version => 3);
422 $p->handler(default => sub { $ret.=join("", @_) }, "text");
423 $p->handler(start => sub {
424 my ($tagname, $pos, $text) = @_;
425 if (ref $HTML::Tagset::linkElements{$tagname}) {
427 # use attribute sets from right to left
428 # to avoid invalidating the offsets
429 # when replacing the values
430 my($k_offset, $k_len, $v_offset, $v_len) =
432 my $attrname = lc(substr($text, $k_offset, $k_len));
433 next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
434 next unless $v_offset; # 0 v_offset means no value
435 my $v = substr($text, $v_offset, $v_len);
436 $v =~ s/^([\'\"])(.*)\1$/$2/;
437 my $new_v=urlabs($v, $urlbase);
438 $new_v =~ s/\"/"/g; # since we quote with ""
439 substr($text, $v_offset, $v_len) = qq("$new_v");
443 }, "tagname, tokenpos, text");
450 sub remove_feeds () { #{{{
454 foreach my $id (keys %feeds) {
455 if ($feeds{$id}->{sourcepage} eq $page) {
456 $feeds{$id}->{remove}=1;
462 sub pagefile ($) { #{{{
465 return "$config{srcdir}/".htmlfn($page);
468 sub htmlfn ($) { #{{{
469 return shift().".html";