X-Git-Url: http://git.vanrenterghem.biz/git.ikiwiki.info.git/blobdiff_plain/dae0f48e91304afcb6ebe0936360e51b22a56548..e3f7675264914032a21fca8a50170d6c0dd375f9:/IkiWiki/Plugin/aggregate.pm diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm index ac55dcb1d..2a3f36fce 100644 --- a/IkiWiki/Plugin/aggregate.pm +++ b/IkiWiki/Plugin/aggregate.pm @@ -9,6 +9,7 @@ use HTML::Entities; use HTML::Parser; use HTML::Tagset; use URI; +use open qw{:utf8 :std}; my %feeds; my %guids; @@ -24,6 +25,7 @@ sub import { #{{{ sub getopt () { #{{{ eval q{use Getopt::Long}; + error($@) if $@; Getopt::Long::Configure('pass_through'); GetOptions("aggregate" => \$config{aggregate}); } #}}} @@ -34,6 +36,7 @@ sub checkconfig () { #{{{ if ($config{aggregate}) { IkiWiki::loadindex(); aggregate(); + expire(); savestate(); } IkiWiki::unlockwiki(); @@ -79,10 +82,12 @@ sub preprocess (@) { #{{{ $feed->{expireage}=defined $params{expireage} ? $params{expireage} : 0; $feed->{expirecount}=defined $params{expirecount} ? $params{expirecount} : 0; delete $feed->{remove}; + delete $feed->{expired}; $feed->{lastupdate}=0 unless defined $feed->{lastupdate}; $feed->{numposts}=0 unless defined $feed->{numposts}; $feed->{newposts}=0 unless defined $feed->{newposts}; $feed->{message}="new feed" unless defined $feed->{message}; + $feed->{error}=0 unless defined $feed->{error}; $feed->{tags}=[]; while (@_) { my $key=shift; @@ -93,7 +98,9 @@ sub preprocess (@) { #{{{ } return "{url}."\">".$feed->{name}.": ". - "".$feed->{message}." (".$feed->{numposts}." posts". + ($feed->{error} ? "" : "").$feed->{message}. + ($feed->{error} ? "" : ""). + " (".$feed->{numposts}." posts". ($feed->{newposts} ? "; ".$feed->{newposts}." new" : ""). ")"; } # }}} @@ -144,7 +151,7 @@ sub loadstate () { #{{{ sub savestate () { #{{{ eval q{use HTML::Entities}; - die $@ if $@; + error($@) if $@; open (OUT, ">$config{wikistatedir}/aggregate" || die "$config{wikistatedir}/aggregate: $!"); foreach my $data (values %feeds, values %guids) { @@ -161,6 +168,11 @@ sub savestate () { #{{{ } next; } + elsif ($data->{expired} && exists $data->{page}) { + unlink pagefile($data->{page}); + delete $data->{page}; + delete $data->{md5}; + } my @line; foreach my $field (keys %$data) { @@ -180,11 +192,37 @@ sub savestate () { #{{{ close OUT; } #}}} +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}} } + values %guids) { + if ($feed->{expireage}) { + my $days_old = (time - $IkiWiki::pagectime{$item->{page}}) / 60 / 60 / 24; + if ($days_old > $feed->{expireage}) { + debug("expiring ".$item->{page}." ($days_old days old)"); + $item->{expired}=1; + } + } + elsif ($feed->{expirecount} && + $count >= $feed->{expirecount}) { + debug("expiring ".$item->{page}); + $item->{expired}=1; + } + else { + $count++; + } + } + } +} #}}} + sub aggregate () { #{{{ eval q{use XML::Feed}; - die $@ if $@; + error($@) if $@; eval q{use HTML::Entities}; - die $@ if $@; + error($@) if $@; foreach my $feed (values %feeds) { next unless $config{rebuild} || @@ -199,6 +237,7 @@ sub aggregate () { #{{{ my @urls=XML::Feed->find_feeds($feed->{url}); if (! @urls) { $feed->{message}="could not find feed at ".$feed->{feedurl}; + $feed->{error}=1; debug($feed->{message}); next; } @@ -207,11 +246,13 @@ sub aggregate () { #{{{ my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))}; if ($@) { $feed->{message}="feed crashed XML::Feed! $@"; + $feed->{error}=1; debug($feed->{message}); next; } if (! $f) { $feed->{message}=XML::Feed->errstr; + $feed->{error}=1; debug($feed->{message}); next; } @@ -229,9 +270,8 @@ sub aggregate () { #{{{ $feed->{message}="processed ok at ". displaytime($feed->{lastupdate}); + $feed->{error}=0; } - - # TODO: expiry } #}}} sub add_page (@) { #{{{ @@ -243,6 +283,7 @@ sub add_page (@) { #{{{ if (exists $guids{$params{guid}}) { # updating an existing post $guid=$guids{$params{guid}}; + return if $guid->{expired}; } else { # new post @@ -276,6 +317,7 @@ sub add_page (@) { #{{{ # to avoid unneccessary rebuilding. The mtime from rss cannot be # trusted; let's use a digest. eval q{use Digest::MD5 'md5_hex'}; + error($@) if $@; require Encode; my $digest=md5_hex(Encode::encode_utf8($params{content})); return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};