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;
46 IkiWiki::unlockwiki();
52 my $page=$params{page};
54 loadstate(); # if not already loaded
55 # Mark all feeds originating on this page as removable;
56 # preprocess will unmark those that still exist.
59 return $params{content};
62 sub preprocess (@) { #{{{
65 foreach my $required (qw{name url}) {
66 if (! exists $params{$required}) {
67 return "[[aggregate ".sprintf(gettext("missing %s parameter"), $required)."]]";
72 my $name=$params{name};
73 if (exists $feeds{$name}) {
80 $feed->{sourcepage}=$params{page};
81 $feed->{url}=$params{url};
82 my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".IkiWiki::titlepage($params{name});
84 ($dir)=$dir=~/$config{wiki_file_regexp}/;
86 $feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
87 $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60;
88 $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0;
89 $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0;
90 delete $feed->{remove};
91 delete $feed->{expired};
92 $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
93 $feed->{numposts}=0 unless defined $feed->{numposts};
94 $feed->{newposts}=0 unless defined $feed->{newposts};
95 $feed->{message}=gettext("new feed") unless defined $feed->{message};
96 $feed->{error}=0 unless defined $feed->{error};
102 push @{$feed->{tags}}, $value;
106 return "<a href=\"".$feed->{url}."\">".$feed->{name}."</a>: ".
107 ($feed->{error} ? "<em>" : "").$feed->{message}.
108 ($feed->{error} ? "</em>" : "").
109 " (".$feed->{numposts}." ".gettext("posts").
110 ($feed->{newposts} ? "; ".$feed->{newposts}.
111 " ".gettext("new") : "").
115 sub delete (@) { #{{{
118 # Remove feed data for removed pages.
119 foreach my $file (@files) {
120 my $page=pagename($file);
126 sub loadstate () { #{{{
127 return if $state_loaded;
128 if (-e "$config{wikistatedir}/aggregate") {
129 open (IN, "$config{wikistatedir}/aggregate" ||
130 die "$config{wikistatedir}/aggregate: $!");
132 $_=IkiWiki::possibly_foolish_untaint($_);
135 foreach my $i (split(/ /, $_)) {
136 my ($field, $val)=split(/=/, $i, 2);
137 if ($field eq "name" || $field eq "feed" ||
138 $field eq "guid" || $field eq "message") {
139 $data->{$field}=decode_entities($val, " \t\n");
141 elsif ($field eq "tag") {
142 push @{$data->{tags}}, $val;
145 $data->{$field}=$val;
149 if (exists $data->{name}) {
150 $feeds{$data->{name}}=$data;
152 elsif (exists $data->{guid}) {
153 $guids{$data->{guid}}=$data;
163 sub savestate () { #{{{
164 eval q{use HTML::Entities};
166 my $newfile="$config{wikistatedir}/aggregate.new";
167 # TODO: This cleanup function could use improvement. Any newly
168 # aggregated files are left behind unrecorded, and should be deleted.
169 my $cleanup = sub { unlink($newfile) };
170 open (OUT, ">$newfile") || error("open $newfile: $!", $cleanup);
171 foreach my $data (values %feeds, values %guids) {
172 if ($data->{remove}) {
174 foreach my $guid (values %guids) {
175 if ($guid->{feed} eq $data->{name}) {
181 unlink pagefile($data->{page});
185 elsif ($data->{expired} && exists $data->{page}) {
186 unlink pagefile($data->{page});
187 delete $data->{page};
192 foreach my $field (keys %$data) {
193 if ($field eq "name" || $field eq "feed" ||
194 $field eq "guid" || $field eq "message") {
195 push @line, "$field=".encode_entities($data->{$field}, " \t\n");
197 elsif ($field eq "tags") {
198 push @line, "tag=$_" foreach @{$data->{tags}};
201 push @line, "$field=".$data->{$field};
204 print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
206 close OUT || error("save $newfile: $!", $cleanup);
207 rename($newfile, "$config{wikistatedir}/aggregate") ||
208 error("rename $newfile: $!", $cleanup);
211 sub clearstate () { #{{{
218 foreach my $feed (values %feeds) {
219 next unless $feed->{expireage} || $feed->{expirecount};
221 foreach my $item (sort { $IkiWiki::pagectime{$b->{page}} <=> $IkiWiki::pagectime{$a->{page}} }
222 grep { exists $_->{page} && $_->{feed} eq $feed->{name} && $IkiWiki::pagectime{$_->{page}} }
224 if ($feed->{expireage}) {
225 my $days_old = (time - $IkiWiki::pagectime{$item->{page}}) / 60 / 60 / 24;
226 if ($days_old > $feed->{expireage}) {
227 debug(sprintf(gettext("expiring %s (%s days old)"),
228 $item->{page}, $days_old));
232 elsif ($feed->{expirecount} &&
233 $count >= $feed->{expirecount}) {
234 debug(sprintf(gettext("expiring %s"), $item->{page}));
244 sub aggregate () { #{{{
245 eval q{use XML::Feed};
247 eval q{use URI::Fetch};
249 eval q{use HTML::Entities};
252 foreach my $feed (values %feeds) {
253 next unless $config{rebuild} ||
254 time - $feed->{lastupdate} >= $feed->{updateinterval};
255 $feed->{lastupdate}=time;
257 $feed->{message}=sprintf(gettext("processed ok at %s"),
258 displaytime($feed->{lastupdate}));
260 $IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
262 debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
264 if (! length $feed->{feedurl}) {
265 my @urls=XML::Feed->find_feeds($feed->{url});
267 $feed->{message}=sprintf(gettext("could not find feed at %s"), $feed->{url});
269 debug($feed->{message});
272 $feed->{feedurl}=pop @urls;
274 my $res=URI::Fetch->fetch($feed->{feedurl});
276 $feed->{message}=URI::Fetch->errstr;
278 debug($feed->{message});
281 if ($res->status == URI::Fetch::URI_GONE()) {
282 $feed->{message}=gettext("feed not found");
284 debug($feed->{message});
287 my $content=$res->content;
288 my $f=eval{XML::Feed->parse(\$content)};
290 # One common cause of XML::Feed crashing is a feed
291 # that contains invalid UTF-8 sequences. Convert
292 # feed to ascii to try to work around.
293 $feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)"));
294 $content=Encode::decode_utf8($content);
295 $f=eval{XML::Feed->parse(\$content)};
298 $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
300 debug($feed->{message});
304 $feed->{message}=XML::Feed->errstr;
306 debug($feed->{message});
310 foreach my $entry ($f->entries) {
313 title => defined $entry->title ? decode_entities($entry->title) : "untitled",
314 link => $entry->link,
315 content => defined $entry->content->body ? $entry->content->body : "",
316 guid => defined $entry->id ? $entry->id : time."_".$feed->name,
317 ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
323 sub add_page (@) { #{{{
326 my $feed=$params{feed};
329 if (exists $guids{$params{guid}}) {
330 # updating an existing post
331 $guid=$guids{$params{guid}};
332 return if $guid->{expired};
336 $guid->{guid}=$params{guid};
337 $guids{$params{guid}}=$guid;
338 $mtime=$params{ctime};
342 # assign it an unused page
343 my $page=IkiWiki::titlepage($params{title});
344 # escape slashes and periods in title so it doesn't specify
345 # directory name or trigger ".." disallowing code.
346 $page=~s!([/.])!"__".ord($1)."__"!eg;
347 $page=$feed->{dir}."/".$page;
348 ($page)=$page=~/$config{wiki_file_regexp}/;
349 if (! defined $page || ! length $page) {
350 $page=$feed->{dir}."/item";
353 while (exists $IkiWiki::pagecase{lc $page.$c} ||
354 -e pagefile($page.$c)) {
358 # Make sure that the file name isn't too long.
359 # NB: This doesn't check for path length limits.
360 my $max=POSIX::pathconf($config{srcdir}, &POSIX::_PC_NAME_MAX);
361 if (defined $max && length(htmlfn($page)) >= $max) {
363 $page=$feed->{dir}."/item";
364 while (exists $IkiWiki::pagecase{lc $page.$c} ||
365 -e pagefile($page.$c)) {
371 debug(sprintf(gettext("creating new page %s"), $page));
373 $guid->{feed}=$feed->{name};
375 # To write or not to write? Need to avoid writing unchanged pages
376 # to avoid unneccessary rebuilding. The mtime from rss cannot be
377 # trusted; let's use a digest.
378 eval q{use Digest::MD5 'md5_hex'};
381 my $digest=md5_hex(Encode::encode_utf8($params{content}));
382 return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
383 $guid->{md5}=$digest;
386 my $template=template("aggregatepost.tmpl", blind_cache => 1);
387 $template->param(title => $params{title})
388 if defined $params{title} && length($params{title});
389 $template->param(content => htmlescape(htmlabs($params{content}, $feed->{feedurl})));
390 $template->param(name => $feed->{name});
391 $template->param(url => $feed->{url});
392 $template->param(permalink => urlabs($params{link}, $feed->{feedurl}))
393 if defined $params{link};
394 if (ref $feed->{tags}) {
395 $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
397 writefile(htmlfn($guid->{page}), $config{srcdir},
400 # Set the mtime, this lets the build process get the right creation
401 # time on record for the new page.
402 utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
405 sub htmlescape ($) { #{{{
406 # escape accidental wikilinks and preprocessor stuff
408 $html=~s/(?<!\\)\[\[/\\\[\[/g;
412 sub urlabs ($$) { #{{{
416 URI->new_abs($url, $urlbase)->as_string;
419 sub htmlabs ($$) { #{{{
420 # Convert links in html from relative to absolute.
421 # Note that this is a heuristic, which is not specified by the rss
422 # spec and may not be right for all feeds. Also, see Debian
428 my $p = HTML::Parser->new(api_version => 3);
429 $p->handler(default => sub { $ret.=join("", @_) }, "text");
430 $p->handler(start => sub {
431 my ($tagname, $pos, $text) = @_;
432 if (ref $HTML::Tagset::linkElements{$tagname}) {
434 # use attribute sets from right to left
435 # to avoid invalidating the offsets
436 # when replacing the values
437 my($k_offset, $k_len, $v_offset, $v_len) =
439 my $attrname = lc(substr($text, $k_offset, $k_len));
440 next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
441 next unless $v_offset; # 0 v_offset means no value
442 my $v = substr($text, $v_offset, $v_len);
443 $v =~ s/^([\'\"])(.*)\1$/$2/;
444 my $new_v=urlabs($v, $urlbase);
445 $new_v =~ s/\"/"/g; # since we quote with ""
446 substr($text, $v_offset, $v_len) = qq("$new_v");
450 }, "tagname, tokenpos, text");
457 sub remove_feeds () { #{{{
461 foreach my $id (keys %feeds) {
462 if ($feeds{$id}->{sourcepage} eq $page) {
463 $feeds{$id}->{remove}=1;
469 sub pagefile ($) { #{{{
472 return "$config{srcdir}/".htmlfn($page);
475 sub htmlfn ($) { #{{{
476 return shift().".html";