X-Git-Url: http://git.vanrenterghem.biz/git.ikiwiki.info.git/blobdiff_plain/a80da69cc358b69278005f65dd86aa5a9fbb64cf..8d4342183b1c3a96797def6fff96feebacb90db6:/IkiWiki/Plugin/aggregate.pm diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm index c18784e8b..be7da3a71 100644 --- a/IkiWiki/Plugin/aggregate.pm +++ b/IkiWiki/Plugin/aggregate.pm @@ -4,20 +4,20 @@ package IkiWiki::Plugin::aggregate; use warnings; use strict; -use IkiWiki 2.00; +use IkiWiki 3.00; use HTML::Parser; use HTML::Tagset; use HTML::Entities; -use URI; use open qw{:utf8 :std}; my %feeds; my %guids; -sub import { #{{{ +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 => "checkconfig", id => "aggregate", call => \&checkconfig, + last => 1); hook(type => "needsbuild", id => "aggregate", call => \&needsbuild); hook(type => "preprocess", id => "aggregate", call => \&preprocess); hook(type => "delete", id => "aggregate", call => \&delete); @@ -26,9 +26,9 @@ sub import { #{{{ if (exists $config{aggregate_webtrigger} && $config{aggregate_webtrigger}) { hook(type => "cgi", id => "aggregate", call => \&cgi); } -} # }}} +} -sub getopt () { #{{{ +sub getopt () { eval q{use Getopt::Long}; error($@) if $@; Getopt::Long::Configure('pass_through'); @@ -36,9 +36,9 @@ sub getopt () { #{{{ "aggregate" => \$config{aggregate}, "aggregateinternal!" => \$config{aggregateinternal}, ); -} #}}} +} -sub getsetup () { #{{{ +sub getsetup () { return plugin => { safe => 1, @@ -46,7 +46,7 @@ sub getsetup () { #{{{ }, aggregateinternal => { type => "boolean", - example => 0, + example => 1, description => "enable aggregation to internal pages?", safe => 0, # enabling needs manual transition rebuild => 0, @@ -58,16 +58,22 @@ sub getsetup () { #{{{ safe => 1, rebuild => 0, }, -} #}}} +} + +sub checkconfig () { + if (! defined $config{aggregateinternal}) { + $config{aggregateinternal}=1; + } -sub checkconfig () { #{{{ + # This is done here rather than in a refresh hook because it + # needs to run before the wiki is locked. if ($config{aggregate} && ! ($config{post_commit} && IkiWiki::commit_hook_enabled())) { launchaggregation(); } -} #}}} +} -sub cgi ($) { #{{{ +sub cgi ($) { my $cgi=shift; if (defined $cgi->param('do') && @@ -90,16 +96,15 @@ sub cgi ($) { #{{{ } exit 0; } -} #}}} +} -sub launchaggregation () { #{{{ +sub launchaggregation () { # See if any feeds need aggregation. loadstate(); my @feeds=needsaggregate(); return unless @feeds; if (! lockaggregate()) { - debug("an aggregation process is already running"); - return; + error("an aggregation process is already running"); } # force a later rebuild of source pages $IkiWiki::forcerebuild{$_->{sourcepage}}=1 @@ -135,16 +140,16 @@ sub launchaggregation () { #{{{ unlockaggregate(); return 1; -} #}}} +} # Pages with extension _aggregated have plain html markup, pass through. -sub htmlize (@) { #{{{ +sub htmlize (@) { my %params=@_; return $params{content}; -} #}}} +} # Used by ikiwiki-transition aggregateinternal. -sub migrate_to_internal { #{{{ +sub migrate_to_internal { if (! lockaggregate()) { error("an aggregation process is currently running"); } @@ -159,10 +164,14 @@ sub migrate_to_internal { #{{{ $config{aggregateinternal} = 0; my $oldname = "$config{srcdir}/".htmlfn($data->{page}); + if (! -e $oldname) { + $oldname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page}); + } + my $oldoutput = $config{destdir}."/".IkiWiki::htmlpage($data->{page}); $config{aggregateinternal} = 1; - my $newname = "$config{srcdir}/".htmlfn($data->{page}); + my $newname = $IkiWiki::Plugin::transient::transientdir."/".htmlfn($data->{page}); debug "moving $oldname -> $newname"; if (-e $newname) { @@ -182,7 +191,7 @@ sub migrate_to_internal { #{{{ if (-e $oldoutput) { require IkiWiki::Render; debug("removing output file $oldoutput"); - IkiWiki::prune($oldoutput); + IkiWiki::prune($oldoutput, $config{destdir}); } } @@ -190,9 +199,9 @@ sub migrate_to_internal { #{{{ IkiWiki::unlockwiki; unlockaggregate(); -} #}}} +} -sub needsbuild (@) { #{{{ +sub needsbuild (@) { my $needsbuild=shift; loadstate(); @@ -206,9 +215,11 @@ sub needsbuild (@) { #{{{ markunseen($feed->{sourcepage}); } } -} # }}} -sub preprocess (@) { #{{{ + return $needsbuild; +} + +sub preprocess (@) { my %params=@_; foreach my $required (qw{name url}) { @@ -245,6 +256,7 @@ sub preprocess (@) { #{{{ $feed->{template}=$params{template} . ".tmpl"; delete $feed->{unseen}; $feed->{lastupdate}=0 unless defined $feed->{lastupdate}; + $feed->{lasttry}=$feed->{lastupdate} unless defined $feed->{lasttry}; $feed->{numposts}=0 unless defined $feed->{numposts}; $feed->{newposts}=0 unless defined $feed->{newposts}; $feed->{message}=gettext("new feed") unless defined $feed->{message}; @@ -265,9 +277,9 @@ sub preprocess (@) { #{{{ ($feed->{newposts} ? "; ".$feed->{newposts}. " ".gettext("new") : ""). ")"; -} # }}} +} -sub delete (@) { #{{{ +sub delete (@) { my @files=@_; # Remove feed data for removed pages. @@ -275,9 +287,9 @@ sub delete (@) { #{{{ my $page=pagename($file); markunseen($page); } -} #}}} +} -sub markunseen ($) { #{{{ +sub markunseen ($) { my $page=shift; foreach my $id (keys %feeds) { @@ -285,15 +297,15 @@ sub markunseen ($) { #{{{ $feeds{$id}->{unseen}=1; } } -} #}}} +} my $state_loaded=0; -sub loadstate () { #{{{ +sub loadstate () { return if $state_loaded; $state_loaded=1; if (-e "$config{wikistatedir}/aggregate") { - open(IN, "$config{wikistatedir}/aggregate") || + open(IN, "<", "$config{wikistatedir}/aggregate") || die "$config{wikistatedir}/aggregate: $!"; while () { $_=IkiWiki::possibly_foolish_untaint($_); @@ -323,14 +335,14 @@ sub loadstate () { #{{{ close IN; } -} #}}} +} -sub savestate () { #{{{ +sub savestate () { return unless $state_loaded; garbage_collect(); my $newfile="$config{wikistatedir}/aggregate.new"; my $cleanup = sub { unlink($newfile) }; - open (OUT, ">$newfile") || error("open $newfile: $!", $cleanup); + open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup); foreach my $data (values %feeds, values %guids) { my @line; foreach my $field (keys %$data) { @@ -342,7 +354,8 @@ sub savestate () { #{{{ push @line, "tag=$_" foreach @{$data->{tags}}; } else { - push @line, "$field=".$data->{$field}; + push @line, "$field=".$data->{$field} + if defined $data->{$field}; } } print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup); @@ -350,9 +363,23 @@ sub savestate () { #{{{ close OUT || error("save $newfile: $!", $cleanup); rename($newfile, "$config{wikistatedir}/aggregate") || error("rename $newfile: $!", $cleanup); -} #}}} -sub garbage_collect () { #{{{ + my $timestamp=undef; + foreach my $feed (keys %feeds) { + my $t=$feeds{$feed}->{lastupdate}+$feeds{$feed}->{updateinterval}; + if (! defined $timestamp || $timestamp > $t) { + $timestamp=$t; + } + } + $newfile=~s/\.new$/time/; + open (OUT, ">", $newfile) || error("open $newfile: $!", $cleanup); + if (defined $timestamp) { + print OUT $timestamp."\n"; + } + close OUT || error("save $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 @@ -364,20 +391,23 @@ sub garbage_collect () { #{{{ foreach my $guid (values %guids) { # any guid whose feed is gone should be removed if (! exists $feeds{$guid->{feed}}) { - unlink "$config{srcdir}/".htmlfn($guid->{page}) - if exists $guid->{page}; + if (exists $guid->{page}) { + unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page}) + || unlink "$config{srcdir}/".htmlfn($guid->{page}); + } delete $guids{$guid->{guid}}; } # handle expired guids elsif ($guid->{expired} && exists $guid->{page}) { unlink "$config{srcdir}/".htmlfn($guid->{page}); + unlink $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page}); delete $guid->{page}; delete $guid->{md5}; } } -} #}}} +} -sub mergestate () { #{{{ +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. @@ -390,14 +420,15 @@ sub mergestate () { #{{{ # fields. foreach my $name (keys %myfeeds) { if (exists $feeds{$name}) { - foreach my $field (qw{message lastupdate numposts - newposts error}) { + foreach my $field (qw{message lastupdate lasttry + numposts newposts error}) { $feeds{$name}->{$field}=$myfeeds{$name}->{$field}; } } } # New guids can be created during aggregation. + # Guids have a few fields that may be updated 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 @@ -406,16 +437,21 @@ sub mergestate () { #{{{ if (! exists $guids{$guid}) { $guids{$guid}=$myguids{$guid}; } + else { + foreach my $field (qw{md5}) { + $guids{$guid}->{$field}=$myguids{$guid}->{$field}; + } + } } -} #}}} +} -sub clearstate () { #{{{ +sub clearstate () { %feeds=(); %guids=(); $state_loaded=0; -} #}}} +} -sub expire () { #{{{ +sub expire () { foreach my $feed (values %feeds) { next unless $feed->{expireage} || $feed->{expirecount}; my $count=0; @@ -444,24 +480,25 @@ sub expire () { #{{{ } } } -} #}}} +} -sub needsaggregate () { #{{{ +sub needsaggregate () { return values %feeds if $config{rebuild}; return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds; -} #}}} +} -sub aggregate (@) { #{{{ +sub aggregate (@) { + eval q{use Net::INET6Glue::INET_is_INET6}; # may not be available eval q{use XML::Feed}; error($@) if $@; eval q{use URI::Fetch}; error($@) if $@; foreach my $feed (@_) { - $feed->{lastupdate}=time; + $feed->{lasttry}=time; $feed->{newposts}=0; - $feed->{message}=sprintf(gettext("processed ok at %s"), - displaytime($feed->{lastupdate})); + $feed->{message}=sprintf(gettext("last checked %s"), + displaytime($feed->{lasttry})); $feed->{error}=0; debug(sprintf(gettext("checking feed %s ..."), $feed->{name})); @@ -476,13 +513,18 @@ sub aggregate (@) { #{{{ } $feed->{feedurl}=pop @urls; } - my $res=URI::Fetch->fetch($feed->{feedurl}); + my $ua=useragent(); + my $res=URI::Fetch->fetch($feed->{feedurl}, UserAgent=>$ua); if (! $res) { $feed->{message}=URI::Fetch->errstr; $feed->{error}=1; debug($feed->{message}); next; } + + # lastupdate is only set if we were able to contact the server + $feed->{lastupdate}=$feed->{lasttry}; + if ($res->status == URI::Fetch::URI_GONE()) { $feed->{message}=gettext("feed not found"); $feed->{error}=1; @@ -496,18 +538,24 @@ sub aggregate (@) { #{{{ # 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, 0); - $f=eval{XML::Feed->parse(\$content)}; + $f=eval { + $content=Encode::decode_utf8($content, 0); + 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)}; + $f=eval { + $content=Encode::decode_utf8($content, 0); + XML::Feed->parse(\$content) + }; } if ($@) { - $feed->{message}=gettext("feed crashed XML::Feed!")." ($@)"; + # gettext can clobber $@ + my $error = $@; + $feed->{message}=gettext("feed crashed XML::Feed!")." ($error)"; $feed->{error}=1; debug($feed->{message}); next; @@ -520,26 +568,35 @@ sub aggregate (@) { #{{{ } foreach my $entry ($f->entries) { - my $content=$content=$entry->content->body; + # XML::Feed doesn't work around XML::Atom's bizarre + # API, so we will. Real unicode strings? Yes please. + # See [[bugs/Aggregated_Atom_feeds_are_double-encoded]] + no warnings 'once'; + local $XML::Atom::ForceUnicode = 1; + use warnings; + + my $c=$entry->content; # atom feeds may have no content, only a summary - if (! defined $content && ref $entry->summary) { - $content=$entry->summary->body; + if (! defined $c && ref $entry->summary) { + $c=$entry->summary; } add_page( feed => $feed, copyright => $f->copyright, title => defined $entry->title ? decode_entities($entry->title) : "untitled", + author => defined $entry->author ? decode_entities($entry->author) : "", link => $entry->link, - content => defined $content ? $content : "", + content => (defined $c && defined $c->body) ? $c->body : "", guid => defined $entry->id ? $entry->id : time."_".$feed->{name}, ctime => $entry->issued ? ($entry->issued->epoch || time) : time, + base => (defined $c && $c->can("base")) ? $c->base : undef, ); } } -} #}}} +} -sub add_page (@) { #{{{ +sub add_page (@) { my %params=@_; my $feed=$params{feed}; @@ -549,6 +606,7 @@ sub add_page (@) { #{{{ # updating an existing post $guid=$guids{$params{guid}}; return if $guid->{expired}; + write_page($feed, $guid, $mtime, \%params); } else { # new post @@ -570,25 +628,39 @@ sub add_page (@) { #{{{ } my $c=""; while (exists $IkiWiki::pagecase{lc $page.$c} || + -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) || -e "$config{srcdir}/".htmlfn($page.$c)) { $c++ } + $page=$page.$c; - # Make sure that the file name isn't too long. - # NB: This doesn't check for path length limits. - my $max=POSIX::pathconf($config{srcdir}, &POSIX::_PC_NAME_MAX); - if (defined $max && length(htmlfn($page)) >= $max) { + $guid->{page}=$page; + eval { write_page($feed, $guid, $mtime, \%params) }; + if ($@) { + # assume failure was due to a too long filename $c=""; $page=$feed->{dir}."/item"; while (exists $IkiWiki::pagecase{lc $page.$c} || - -e "$config{srcdir}/".htmlfn($page.$c)) { + -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) || + -e "$config{srcdir}/".htmlfn($page.$c)) { $c++ } + $page=$page.$c; + + $guid->{page}=$page; + write_page($feed, $guid, $mtime, \%params); } - $guid->{page}=$page; debug(sprintf(gettext("creating new page %s"), $page)); } +} + +sub write_page ($$$$$) { + my $feed=shift; + my $guid=shift; + my $mtime=shift; + my %params=%{shift()}; + $guid->{feed}=$feed->{name}; # To write or not to write? Need to avoid writing unchanged pages @@ -602,50 +674,57 @@ sub add_page (@) { #{{{ $guid->{md5}=$digest; # Create the page. - my $template=template($feed->{template}, blind_cache => 1); + my $template; + eval { + $template=template($feed->{template}, blind_cache => 1); + }; + if ($@) { + # gettext can clobber $@ + my $error = $@; + print STDERR gettext("failed to process template:")." $error"; + return; + } $template->param(title => $params{title}) if defined $params{title} && length($params{title}); - $template->param(content => htmlescape(htmlabs($params{content}, $feed->{feedurl}))); + $template->param(author => $params{author}) + if defined $params{author} && length($params{author} + && $params{author} ne $feed->{name}); + $template->param(content => wikiescape(htmlabs($params{content}, + defined $params{base} ? $params{base} : $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})) + $template->param(permalink => IkiWiki::urlabs($params{link}, $feed->{feedurl})) if defined $params{link}; if (ref $feed->{tags}) { $template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]); } - writefile(htmlfn($guid->{page}), $config{srcdir}, - $template->output); + writefile(htmlfn($guid->{page}), + $IkiWiki::Plugin::transient::transientdir, $template->output); 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, $mtime, "$config{srcdir}/".htmlfn($guid->{page}); + utime $mtime, $mtime, + $IkiWiki::Plugin::transient::transientdir."/".htmlfn($guid->{page}); # Store it in pagectime for expiry code to use also. - $IkiWiki::pagectime{$guid->{page}}=$mtime; + $IkiWiki::pagectime{$guid->{page}}=$mtime + unless exists $IkiWiki::pagectime{$guid->{page}}; } else { # Dummy value for expiry code. - $IkiWiki::pagectime{$guid->{page}}=time; + $IkiWiki::pagectime{$guid->{page}}=time + unless exists $IkiWiki::pagectime{$guid->{page}}; } -} #}}} +} -sub htmlescape ($) { #{{{ +sub wikiescape ($) { # escape accidental wikilinks and preprocessor stuff - my $html=shift; - $html=~s/(?new_abs($url, $urlbase)->as_string; -} #}}} + return encode_entities(shift, '\[\]'); +} -sub htmlabs ($$) { #{{{ +sub htmlabs ($$) { # Convert links in html from relative to absolute. # Note that this is a heuristic, which is not specified by the rss # spec and may not be right for all feeds. Also, see Debian @@ -670,7 +749,7 @@ sub htmlabs ($$) { #{{{ next unless $v_offset; # 0 v_offset means no value my $v = substr($text, $v_offset, $v_len); $v =~ s/^([\'\"])(.*)\1$/$2/; - my $new_v=urlabs($v, $urlbase); + my $new_v=IkiWiki::urlabs($v, $urlbase); $new_v =~ s/\"/"/g; # since we quote with "" substr($text, $v_offset, $v_len) = qq("$new_v"); } @@ -681,15 +760,15 @@ sub htmlabs ($$) { #{{{ $p->eof; return $ret; -} #}}} +} -sub htmlfn ($) { #{{{ +sub htmlfn ($) { return shift().".".($config{aggregateinternal} ? "_aggregated" : $config{htmlext}); -} #}}} +} my $aggregatelock; -sub lockaggregate () { #{{{ +sub lockaggregate () { # Take an exclusive lock to prevent multiple concurrent aggregators. # Returns true if the lock was aquired. if (! -d $config{wikistatedir}) { @@ -702,11 +781,11 @@ sub lockaggregate () { #{{{ return 0; } return 1; -} #}}} +} -sub unlockaggregate () { #{{{ +sub unlockaggregate () { return close($aggregatelock) if $aggregatelock; return; -} #}}} +} 1