X-Git-Url: https://sipb.mit.edu/gitweb.cgi/ikiwiki.git/blobdiff_plain/71d2d930272bcdba0b5d2671506e2386ea416632..f181962dfc24c67f7db840abd3b6a3ca06e39b1b:/IkiWiki/Plugin/aggregate.pm diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm index be2c96369..8044f1ed4 100644 --- a/IkiWiki/Plugin/aggregate.pm +++ b/IkiWiki/Plugin/aggregate.pm @@ -1,13 +1,13 @@ #!/usr/bin/perl -# Blog aggregation plugin. +# Feed aggregation plugin. package IkiWiki::Plugin::aggregate; use warnings; use strict; -use IkiWiki; -use HTML::Entities; +use IkiWiki 2.00; use HTML::Parser; use HTML::Tagset; +use HTML::Entities; use URI; use open qw{:utf8 :std}; @@ -16,42 +16,196 @@ my %guids; sub import { #{{{ hook(type => "getopt", id => "aggregate", call => \&getopt); + hook(type => "getsetup", id => "aggregate", call => \&getsetup); hook(type => "checkconfig", id => "aggregate", call => \&checkconfig); - hook(type => "filter", id => "aggregate", call => \&filter); + hook(type => "needsbuild", id => "aggregate", call => \&needsbuild); hook(type => "preprocess", id => "aggregate", call => \&preprocess); hook(type => "delete", id => "aggregate", call => \&delete); hook(type => "savestate", id => "aggregate", call => \&savestate); + hook(type => "htmlize", id => "_aggregated", call => \&htmlize); + if (exists $config{aggregate_webtrigger} && $config{aggregate_webtrigger}) { + hook(type => "cgi", id => "aggregate", call => \&cgi); + } } # }}} sub getopt () { #{{{ eval q{use Getopt::Long}; error($@) if $@; Getopt::Long::Configure('pass_through'); - GetOptions("aggregate" => \$config{aggregate}); + GetOptions( + "aggregate" => \$config{aggregate}, + "aggregateinternal!" => \$config{aggregateinternal}, + ); +} #}}} + +sub getsetup () { #{{{ + return + plugin => { + safe => 1, + rebuild => undef, + }, + aggregateinternal => { + type => "boolean", + example => 0, + description => "enable aggregation to internal pages?", + safe => 0, # enabling needs manual transition + rebuild => 0, + }, + aggregate_webtrigger => { + type => "boolean", + example => 0, + description => "allow aggregation to be triggered via the web?", + safe => 1, + rebuild => 0, + }, } #}}} sub checkconfig () { #{{{ - my $nolock=($config{post_commit} && ! IkiWiki::commit_hook_enabled()); - IkiWiki::lockwiki() unless $nolock; + if ($config{aggregate} && ! ($config{post_commit} && + IkiWiki::commit_hook_enabled())) { + launchaggregation(); + } +} #}}} + +sub cgi ($) { #{{{ + my $cgi=shift; + + if (defined $cgi->param('do') && + $cgi->param("do") eq "aggregate_webtrigger") { + $|=1; + print "Content-Type: text/plain\n\n"; + $config{cgi}=0; + $config{verbose}=1; + $config{syslog}=0; + print gettext("Aggregation triggered via web.")."\n\n"; + if (launchaggregation()) { + IkiWiki::lockwiki(); + IkiWiki::loadindex(); + require IkiWiki::Render; + IkiWiki::refresh(); + IkiWiki::saveindex(); + } + else { + print gettext("Nothing to do right now, all feeds are up-to-date!")."\n"; + } + exit 0; + } +} #}}} + +sub launchaggregation () { #{{{ + # See if any feeds need aggregation. loadstate(); - if ($config{aggregate} && ! $nolock) { + my @feeds=needsaggregate(); + return unless @feeds; + if (! lockaggregate()) { + debug("an aggregation process is already running"); + return; + } + # force a later rebuild of source pages + $IkiWiki::forcerebuild{$_->{sourcepage}}=1 + foreach @feeds; + + # Fork a child process to handle the aggregation. + # The parent process will then handle building the + # result. This avoids messy code to clear state + # accumulated while aggregating. + defined(my $pid = fork) or error("Can't fork: $!"); + if (! $pid) { IkiWiki::loadindex(); - aggregate(); + # Aggregation happens without the main wiki lock + # being held. This allows editing pages etc while + # aggregation is running. + aggregate(@feeds); + + IkiWiki::lockwiki; + # Merge changes, since aggregation state may have + # changed on disk while the aggregation was happening. + mergestate(); expire(); savestate(); + IkiWiki::unlockwiki; + exit 0; + } + waitpid($pid,0); + if ($?) { + error "aggregation failed with code $?"; } - IkiWiki::unlockwiki() unless $nolock; + + clearstate(); + unlockaggregate(); + + return 1; } #}}} -sub filter (@) { #{{{ +# Pages with extension _aggregated have plain html markup, pass through. +sub htmlize (@) { #{{{ my %params=@_; - my $page=$params{page}; + return $params{content}; +} #}}} - # Mark all feeds originating on this page as removable; - # preprocess will unmark those that still exist. - remove_feeds($page); +# Used by ikiwiki-transition aggregateinternal. +sub migrate_to_internal { #{{{ + if (! lockaggregate()) { + error("an aggregation process is currently running"); + } - return $params{content}; + IkiWiki::lockwiki(); + loadstate(); + $config{verbose}=1; + + foreach my $data (values %guids) { + next unless $data->{page}; + next if $data->{expired}; + + $config{aggregateinternal} = 0; + my $oldname = pagefile($data->{page}); + my $oldoutput = $config{destdir}."/".IkiWiki::htmlpage($data->{page}); + + $config{aggregateinternal} = 1; + my $newname = pagefile($data->{page}); + + debug "moving $oldname -> $newname"; + if (-e $newname) { + if (-e $oldname) { + error("$newname already exists"); + } + else { + debug("already renamed to $newname?"); + } + } + elsif (-e $oldname) { + rename($oldname, $newname) || error("$!"); + } + else { + debug("$oldname not found"); + } + if (-e $oldoutput) { + require IkiWiki::Render; + debug("removing output file $oldoutput"); + IkiWiki::prune($oldoutput); + } + } + + savestate(); + IkiWiki::unlockwiki; + + unlockaggregate(); +} #}}} + +sub needsbuild (@) { #{{{ + my $needsbuild=shift; + + loadstate(); + + foreach my $feed (values %feeds) { + if (exists $pagesources{$feed->{sourcepage}} && + grep { $_ eq $pagesources{$feed->{sourcepage}} } @$needsbuild) { + # Mark all feeds originating on this page as + # not yet seen; preprocess will unmark those that + # still exist. + markunseen($feed->{sourcepage}); + } + } } # }}} sub preprocess (@) { #{{{ @@ -59,7 +213,7 @@ sub preprocess (@) { #{{{ foreach my $required (qw{name url}) { if (! exists $params{$required}) { - return "[[aggregate ".sprintf(gettext("missing %s parameter"), $required)."]]"; + error sprintf(gettext("missing %s parameter"), $required) } } @@ -82,8 +236,14 @@ sub preprocess (@) { #{{{ $feed->{updateinterval}=defined $params{updateinterval} ? $params{updateinterval} * 60 : 15 * 60; $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0; $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0; - delete $feed->{remove}; - delete $feed->{expired}; + if (exists $params{template}) { + $params{template}=~s/[^-_a-zA-Z0-9]+//g; + } + else { + $params{template} = "aggregatepost" + } + $feed->{template}=$params{template} . ".tmpl"; + delete $feed->{unseen}; $feed->{lastupdate}=0 unless defined $feed->{lastupdate}; $feed->{numposts}=0 unless defined $feed->{numposts}; $feed->{newposts}=0 unless defined $feed->{newposts}; @@ -113,14 +273,28 @@ sub delete (@) { #{{{ # Remove feed data for removed pages. foreach my $file (@files) { my $page=pagename($file); - remove_feeds($page); + markunseen($page); } } #}}} +sub markunseen ($) { #{{{ + my $page=shift; + + foreach my $id (keys %feeds) { + if ($feeds{$id}->{sourcepage} eq $page) { + $feeds{$id}->{unseen}=1; + } + } +} #}}} + +my $state_loaded=0; + sub loadstate () { #{{{ + return if $state_loaded; + $state_loaded=1; if (-e "$config{wikistatedir}/aggregate") { - open (IN, "$config{wikistatedir}/aggregate" || - die "$config{wikistatedir}/aggregate: $!"); + open(IN, "$config{wikistatedir}/aggregate") || + die "$config{wikistatedir}/aggregate: $!"; while () { $_=IkiWiki::possibly_foolish_untaint($_); chomp; @@ -152,33 +326,12 @@ sub loadstate () { #{{{ } #}}} sub savestate () { #{{{ - eval q{use HTML::Entities}; - error($@) if $@; + return unless $state_loaded; + garbage_collect(); my $newfile="$config{wikistatedir}/aggregate.new"; - # TODO: This cleanup function could use improvement. Any newly - # aggregated files are left behind unrecorded, and should be deleted. my $cleanup = sub { unlink($newfile) }; open (OUT, ">$newfile") || error("open $newfile: $!", $cleanup); foreach my $data (values %feeds, values %guids) { - if ($data->{remove}) { - if ($data->{name}) { - foreach my $guid (values %guids) { - if ($guid->{feed} eq $data->{name}) { - $guid->{remove}=1; - } - } - } - else { - unlink pagefile($data->{page}); - } - next; - } - elsif ($data->{expired} && exists $data->{page}) { - unlink pagefile($data->{page}); - delete $data->{page}; - delete $data->{md5}; - } - my @line; foreach my $field (keys %$data) { if ($field eq "name" || $field eq "feed" || @@ -199,18 +352,82 @@ sub savestate () { #{{{ error("rename $newfile: $!", $cleanup); } #}}} +sub garbage_collect () { #{{{ + foreach my $name (keys %feeds) { + # remove any feeds that were not seen while building the pages + # that used to contain them + if ($feeds{$name}->{unseen}) { + delete $feeds{$name}; + } + } + + foreach my $guid (values %guids) { + # any guid whose feed is gone should be removed + if (! exists $feeds{$guid->{feed}}) { + unlink pagefile($guid->{page}) + if exists $guid->{page}; + delete $guids{$guid->{guid}}; + } + # handle expired guids + elsif ($guid->{expired} && exists $guid->{page}) { + unlink pagefile($guid->{page}); + delete $guid->{page}; + delete $guid->{md5}; + } + } +} #}}} + +sub mergestate () { #{{{ + # Load the current state in from disk, and merge into it + # values from the state in memory that might have changed + # during aggregation. + my %myfeeds=%feeds; + my %myguids=%guids; + clearstate(); + loadstate(); + + # All that can change in feed state during aggregation is a few + # fields. + foreach my $name (keys %myfeeds) { + if (exists $feeds{$name}) { + foreach my $field (qw{message lastupdate numposts + newposts error}) { + $feeds{$name}->{$field}=$myfeeds{$name}->{$field}; + } + } + } + + # New guids can be created during aggregation. + # It's also possible that guids were removed from the on-disk state + # while the aggregation was in process. That would only happen if + # their feed was also removed, so any removed guids added back here + # will be garbage collected later. + foreach my $guid (keys %myguids) { + if (! exists $guids{$guid}) { + $guids{$guid}=$myguids{$guid}; + } + } +} #}}} + +sub clearstate () { #{{{ + %feeds=(); + %guids=(); + $state_loaded=0; +} #}}} + sub expire () { #{{{ foreach my $feed (values %feeds) { next unless $feed->{expireage} || $feed->{expirecount}; my $count=0; - foreach my $item (sort { $IkiWiki::pagectime{$b->{page}} <=> $IkiWiki::pagectime{$a->{page}} } - grep { exists $_->{page} && $_->{feed} eq $feed->{name} && $IkiWiki::pagectime{$_->{page}} } + my %seen; + foreach my $item (sort { ($IkiWiki::pagectime{$b->{page}} || 0) <=> ($IkiWiki::pagectime{$a->{page}} || 0) } + grep { exists $_->{page} && $_->{feed} eq $feed->{name} } values %guids) { if ($feed->{expireage}) { - my $days_old = (time - $IkiWiki::pagectime{$item->{page}}) / 60 / 60 / 24; + my $days_old = (time - ($IkiWiki::pagectime{$item->{page}} || 0)) / 60 / 60 / 24; if ($days_old > $feed->{expireage}) { debug(sprintf(gettext("expiring %s (%s days old)"), - $item->{page}, $days_old)); + $item->{page}, int($days_old))); $item->{expired}=1; } } @@ -220,29 +437,32 @@ sub expire () { #{{{ $item->{expired}=1; } else { - $count++; + if (! $seen{$item->{page}}) { + $seen{$item->{page}}=1; + $count++; + } } } } } #}}} -sub aggregate () { #{{{ +sub needsaggregate () { #{{{ + return values %feeds if $config{rebuild}; + return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds; +} #}}} + +sub aggregate (@) { #{{{ eval q{use XML::Feed}; error($@) if $@; eval q{use URI::Fetch}; error($@) if $@; - eval q{use HTML::Entities}; - error($@) if $@; - foreach my $feed (values %feeds) { - next unless $config{rebuild} || - time - $feed->{lastupdate} >= $feed->{updateinterval}; + foreach my $feed (@_) { $feed->{lastupdate}=time; $feed->{newposts}=0; $feed->{message}=sprintf(gettext("processed ok at %s"), displaytime($feed->{lastupdate})); $feed->{error}=0; - $IkiWiki::forcerebuild{$feed->{sourcepage}}=1; debug(sprintf(gettext("checking feed %s ..."), $feed->{name})); @@ -275,8 +495,15 @@ sub aggregate () { #{{{ # One common cause of XML::Feed crashing is a feed # that contains invalid UTF-8 sequences. Convert # feed to ascii to try to work around. - $feed->{message}=sprintf(gettext("invalid UTF-8 stripped from feed")); - $content=Encode::decode_utf8($content); + $feed->{message}.=" ".sprintf(gettext("(invalid UTF-8 stripped from feed)")); + $content=Encode::decode_utf8($content, 0); + $f=eval{XML::Feed->parse(\$content)}; + } + if ($@) { + # Another possibility is badly escaped entities. + $feed->{message}.=" ".sprintf(gettext("(feed entities escaped)")); + $content=~s/\&(?!amp)(\w+);/&$1;/g; + $content=Encode::decode_utf8($content, 0); $f=eval{XML::Feed->parse(\$content)}; } if ($@) { @@ -293,12 +520,19 @@ sub aggregate () { #{{{ } foreach my $entry ($f->entries) { + my $content=$content=$entry->content->body; + # atom feeds may have no content, only a summary + if (! defined $content && ref $entry->summary) { + $content=$entry->summary->body; + } + add_page( feed => $feed, + copyright => $f->copyright, title => defined $entry->title ? decode_entities($entry->title) : "untitled", link => $entry->link, - content => defined $entry->content->body ? $entry->content->body : "", - guid => defined $entry->id ? $entry->id : time."_".$feed->name, + content => defined $content ? $content : "", + guid => defined $entry->id ? $entry->id : time."_".$feed->{name}, ctime => $entry->issued ? ($entry->issued->epoch || time) : time, ); } @@ -368,12 +602,14 @@ sub add_page (@) { #{{{ $guid->{md5}=$digest; # Create the page. - my $template=template("aggregatepost.tmpl", blind_cache => 1); + my $template=template($feed->{template}, blind_cache => 1); $template->param(title => $params{title}) if defined $params{title} && length($params{title}); $template->param(content => htmlescape(htmlabs($params{content}, $feed->{feedurl}))); $template->param(name => $feed->{name}); $template->param(url => $feed->{url}); + $template->param(copyright => $params{copyright}) + if defined $params{copyright} && length $params{copyright}; $template->param(permalink => urlabs($params{link}, $feed->{feedurl})) if defined $params{link}; if (ref $feed->{tags}) { @@ -382,9 +618,13 @@ sub add_page (@) { #{{{ writefile(htmlfn($guid->{page}), $config{srcdir}, $template->output); - # Set the mtime, this lets the build process get the right creation - # time on record for the new page. - utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime; + if (defined $mtime && $mtime <= time) { + # Set the mtime, this lets the build process get the right + # creation time on record for the new page. + utime $mtime, $mtime, pagefile($guid->{page}); + # Store it in pagectime for expiry code to use also. + $IkiWiki::pagectime{$guid->{page}}=$mtime; + } } #}}} sub htmlescape ($) { #{{{ @@ -439,18 +679,6 @@ sub htmlabs ($$) { #{{{ return $ret; } #}}} -sub remove_feeds () { #{{{ - my $page=shift; - - my %removed; - foreach my $id (keys %feeds) { - if ($feeds{$id}->{sourcepage} eq $page) { - $feeds{$id}->{remove}=1; - $removed{$id}=1; - } - } -} #}}} - sub pagefile ($) { #{{{ my $page=shift; @@ -458,7 +686,29 @@ sub pagefile ($) { #{{{ } #}}} sub htmlfn ($) { #{{{ - return shift().".html"; + return shift().".".($config{aggregateinternal} ? "_aggregated" : $config{htmlext}); +} #}}} + +my $aggregatelock; + +sub lockaggregate () { #{{{ + # Take an exclusive lock to prevent multiple concurrent aggregators. + # Returns true if the lock was aquired. + if (! -d $config{wikistatedir}) { + mkdir($config{wikistatedir}); + } + open($aggregatelock, '>', "$config{wikistatedir}/aggregatelock") || + error ("cannot open to $config{wikistatedir}/aggregatelock: $!"); + if (! flock($aggregatelock, 2 | 4)) { # LOCK_EX | LOCK_NB + close($aggregatelock) || error("failed closing aggregatelock: $!"); + return 0; + } + return 1; +} #}}} + +sub unlockaggregate () { #{{{ + return close($aggregatelock) if $aggregatelock; + return; } #}}} 1