X-Git-Url: https://sipb.mit.edu/gitweb.cgi/ikiwiki.git/blobdiff_plain/dcaa4b013eddf9b0654e0b1e70f7c3cce8a9cabf..e43cd269d2b492da6fc4bfdc3d6930e88f1dfa0c:/IkiWiki/Plugin/aggregate.pm
diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm
index cb04308e6..2a3f36fce 100644
--- a/IkiWiki/Plugin/aggregate.pm
+++ b/IkiWiki/Plugin/aggregate.pm
@@ -5,38 +5,41 @@ package IkiWiki::Plugin::aggregate;
use warnings;
use strict;
use IkiWiki;
+use HTML::Entities;
+use HTML::Parser;
+use HTML::Tagset;
+use URI;
+use open qw{:utf8 :std};
my %feeds;
my %guids;
sub import { #{{{
- IkiWiki::hook(type => "getopt", id => "aggregate",
- call => \&getopt);
- IkiWiki::hook(type => "checkconfig", id => "aggregate",
- call => \&checkconfig);
- IkiWiki::hook(type => "filter", id => "aggregate",
- call => \&filter);
- IkiWiki::hook(type => "preprocess", id => "aggregate",
- call => \&preprocess);
- IkiWiki::hook(type => "delete", id => "aggregate",
- call => \&delete);
- IkiWiki::hook(type => "savestate", id => "aggregate",
- call => \&savestate);
+ hook(type => "getopt", id => "aggregate", call => \&getopt);
+ hook(type => "checkconfig", id => "aggregate", call => \&checkconfig);
+ hook(type => "filter", id => "aggregate", call => \&filter);
+ hook(type => "preprocess", id => "aggregate", call => \&preprocess);
+ hook(type => "delete", id => "aggregate", call => \&delete);
+ hook(type => "savestate", id => "aggregate", call => \&savestate);
} # }}}
sub getopt () { #{{{
eval q{use Getopt::Long};
+ error($@) if $@;
Getopt::Long::Configure('pass_through');
- GetOptions("aggregate" => \$IkiWiki::config{aggregate});
+ GetOptions("aggregate" => \$config{aggregate});
} #}}}
sub checkconfig () { #{{{
+ IkiWiki::lockwiki();
loadstate();
- if ($IkiWiki::config{aggregate}) {
+ if ($config{aggregate}) {
IkiWiki::loadindex();
aggregate();
+ expire();
savestate();
}
+ IkiWiki::unlockwiki();
} #}}}
sub filter (@) { #{{{
@@ -70,19 +73,22 @@ sub preprocess (@) { #{{{
$feed->{name}=$name;
$feed->{sourcepage}=$params{page};
$feed->{url}=$params{url};
- my $dir=exists $params{dir} ? $params{dir} : IkiWiki::titlepage($params{name});
+ my $dir=exists $params{dir} ? $params{dir} : $params{page}."/".IkiWiki::titlepage($params{name});
$dir=~s/^\/+//;
- ($dir)=$dir=~/$IkiWiki::config{wiki_file_regexp}/;
+ ($dir)=$dir=~/$config{wiki_file_regexp}/;
$feed->{dir}=$dir;
$feed->{feedurl}=defined $params{feedurl} ? $params{feedurl} : "";
$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};
$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;
my $value=shift;
@@ -92,8 +98,11 @@ sub preprocess (@) { #{{{
}
return "{url}."\">".$feed->{name}.": ".
- "".$feed->{message}." (".$feed->{numposts}.
- " stored posts; ".$feed->{newposts}." new)
";
+ ($feed->{error} ? "" : "").$feed->{message}.
+ ($feed->{error} ? "" : "").
+ " (".$feed->{numposts}." posts".
+ ($feed->{newposts} ? "; ".$feed->{newposts}." new" : "").
+ ")";
} # }}}
sub delete (@) { #{{{
@@ -101,23 +110,24 @@ sub delete (@) { #{{{
# Remove feed data for removed pages.
foreach my $file (@files) {
- my $page=IkiWiki::pagename($file);
+ my $page=pagename($file);
remove_feeds($page);
}
} #}}}
sub loadstate () { #{{{
- if (-e "$IkiWiki::config{wikistatedir}/aggregate") {
- open (IN, "$IkiWiki::config{wikistatedir}/aggregate" ||
- die "$IkiWiki::config{wikistatedir}/aggregate: $!");
+ if (-e "$config{wikistatedir}/aggregate") {
+ open (IN, "$config{wikistatedir}/aggregate" ||
+ die "$config{wikistatedir}/aggregate: $!");
while () {
$_=IkiWiki::possibly_foolish_untaint($_);
chomp;
my $data={};
foreach my $i (split(/ /, $_)) {
my ($field, $val)=split(/=/, $i, 2);
- if ($field eq "name" || $field eq "message") {
- $data->{$field}=IkiWiki::pagetitle($val);
+ if ($field eq "name" || $field eq "feed" ||
+ $field eq "guid" || $field eq "message") {
+ $data->{$field}=decode_entities($val, " \t\n");
}
elsif ($field eq "tag") {
push @{$data->{tags}}, $val;
@@ -140,8 +150,10 @@ sub loadstate () { #{{{
} #}}}
sub savestate () { #{{{
- open (OUT, ">$IkiWiki::config{wikistatedir}/aggregate" ||
- die "$IkiWiki::config{wikistatedir}/aggregate: $!");
+ eval q{use HTML::Entities};
+ error($@) if $@;
+ open (OUT, ">$config{wikistatedir}/aggregate" ||
+ die "$config{wikistatedir}/aggregate: $!");
foreach my $data (values %feeds, values %guids) {
if ($data->{remove}) {
if ($data->{name}) {
@@ -156,11 +168,17 @@ sub savestate () { #{{{
}
next;
}
+ elsif ($data->{expired} && exists $data->{page}) {
+ unlink pagefile($data->{page});
+ delete $data->{page};
+ delete $data->{md5};
+ }
my @line;
foreach my $field (keys %$data) {
- if ($field eq "name" || $field eq "message") {
- push @line, "$field=".IkiWiki::titlepage($data->{$field});
+ if ($field eq "name" || $field eq "feed" ||
+ $field eq "guid" || $field eq "message") {
+ push @line, "$field=".encode_entities($data->{$field}, " \t\n");
}
elsif ($field eq "tags") {
push @line, "tag=$_" foreach @{$data->{tags}};
@@ -174,25 +192,53 @@ 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 time - $feed->{lastupdate} >= $feed->{updateinterval};
+ next unless $config{rebuild} ||
+ time - $feed->{lastupdate} >= $feed->{updateinterval};
$feed->{lastupdate}=time;
$feed->{newposts}=0;
$IkiWiki::forcerebuild{$feed->{sourcepage}}=1;
- IkiWiki::debug("checking feed ".$feed->{name}." ...");
+ debug("checking feed ".$feed->{name}." ...");
if (! length $feed->{feedurl}) {
my @urls=XML::Feed->find_feeds($feed->{url});
if (! @urls) {
$feed->{message}="could not find feed at ".$feed->{feedurl};
- IkiWiki::debug($feed->{message});
+ $feed->{error}=1;
+ debug($feed->{message});
next;
}
$feed->{feedurl}=pop @urls;
@@ -200,12 +246,14 @@ sub aggregate () { #{{{
my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))};
if ($@) {
$feed->{message}="feed crashed XML::Feed! $@";
- IkiWiki::debug($feed->{message});
+ $feed->{error}=1;
+ debug($feed->{message});
next;
}
if (! $f) {
$feed->{message}=XML::Feed->errstr;
- IkiWiki::debug($feed->{message});
+ $feed->{error}=1;
+ debug($feed->{message});
next;
}
@@ -220,10 +268,10 @@ sub aggregate () { #{{{
);
}
- $feed->{message}="processed ok";
+ $feed->{message}="processed ok at ".
+ displaytime($feed->{lastupdate});
+ $feed->{error}=0;
}
-
- # TODO: expiry
} #}}}
sub add_page (@) { #{{{
@@ -235,6 +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
@@ -245,19 +294,22 @@ sub add_page (@) { #{{{
$feed->{newposts}++;
# assign it an unused page
- my $page=$feed->{dir}."/".IkiWiki::titlepage($params{title});
- ($page)=$page=~/$IkiWiki::config{wiki_file_regexp}/;
+ my $page=IkiWiki::titlepage($params{title});
+ # escape slashes and periods in title so it doesn't specify
+ # directory name or trigger ".." disallowing code.
+ $page=~s!([/.])!"__".ord($1)."__"!eg;
+ $page=$feed->{dir}."/".$page;
+ ($page)=$page=~/$config{wiki_file_regexp}/;
if (! defined $page || ! length $page) {
$page=$feed->{dir}."/item";
}
- $page=~s/\.\.//g; # avoid ".." directory tricks
my $c="";
- while (exists $IkiWiki::pagesources{$page.$c} ||
+ while (exists $IkiWiki::pagecase{lc $page.$c} ||
-e pagefile($page.$c)) {
$c++
}
$guid->{page}=$page;
- IkiWiki::debug("creating new page $page");
+ debug("creating new page $page");
}
$guid->{feed}=$feed->{name};
@@ -265,24 +317,25 @@ 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;
+ return unless ! exists $guid->{md5} || $guid->{md5} ne $digest || $config{rebuild};
$guid->{md5}=$digest;
# Create the page.
- my $template=IkiWiki::template("aggregatepost.tmpl", blind_cache => 1);
- my $content=$params{content};
- $params{content}=~s/(?param(content => $params{content});
- $template->param(url => $feed->{url});
+ my $template=template("aggregatepost.tmpl", 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(link => $params{link}) if defined $params{link};
+ $template->param(url => $feed->{url});
+ $template->param(permalink => urlabs($params{link}, $feed->{feedurl}))
+ if defined $params{link};
if (ref $feed->{tags}) {
$template->param(tags => [map { tag => $_ }, @{$feed->{tags}}]);
}
- IkiWiki::writefile($guid->{page}.".html", $IkiWiki::config{srcdir},
+ writefile(htmlpage($guid->{page}), $config{srcdir},
$template->output);
# Set the mtime, this lets the build process get the right creation
@@ -290,6 +343,58 @@ sub add_page (@) { #{{{
utime $mtime, $mtime, pagefile($guid->{page}) if defined $mtime;
} #}}}
+sub htmlescape ($) { #{{{
+ # escape accidental wikilinks and preprocessor stuff
+ my $html=shift;
+ $html=~s/(?new_abs($url, $urlbase)->as_string;
+} #}}}
+
+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
+ # bug #381359.
+ my $html=shift;
+ my $urlbase=shift;
+
+ my $ret="";
+ my $p = HTML::Parser->new(api_version => 3);
+ $p->handler(default => sub { $ret.=join("", @_) }, "text");
+ $p->handler(start => sub {
+ my ($tagname, $pos, $text) = @_;
+ if (ref $HTML::Tagset::linkElements{$tagname}) {
+ while (4 <= @$pos) {
+ # use attribute sets from right to left
+ # to avoid invalidating the offsets
+ # when replacing the values
+ my($k_offset, $k_len, $v_offset, $v_len) =
+ splice(@$pos, -4);
+ my $attrname = lc(substr($text, $k_offset, $k_len));
+ next unless grep { $_ eq $attrname } @{$HTML::Tagset::linkElements{$tagname}};
+ 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);
+ $new_v =~ s/\"/"/g; # since we quote with ""
+ substr($text, $v_offset, $v_len) = qq("$new_v");
+ }
+ }
+ $ret.=$text;
+ }, "tagname, tokenpos, text");
+ $p->parse($html);
+ $p->eof;
+
+ return $ret;
+} #}}}
+
sub remove_feeds () { #{{{
my $page=shift;
@@ -305,7 +410,7 @@ sub remove_feeds () { #{{{
sub pagefile ($) { #{{{
my $page=shift;
- return "$IkiWiki::config{srcdir}/$page.html";
+ return "$config{srcdir}/".htmlpage($page);
} #}}}
1