use HTML::Parser;
use HTML::Tagset;
use HTML::Entities;
-use URI;
use open qw{:utf8 :std};
my %feeds;
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);
safe => 1,
rebuild => 0,
},
+ cookiejar => {
+ type => "string",
+ example => { file => "$ENV{HOME}/.ikiwiki/cookies" },
+ safe => 0, # hooks into perl module internals
+ description => "cookie control",
+ },
}
sub checkconfig () {
if (! defined $config{aggregateinternal}) {
$config{aggregateinternal}=1;
}
+ if (! defined $config{cookiejar}) {
+ $config{cookiejar}={ file => "$ENV{HOME}/.ikiwiki/cookies" };
+ }
+ # 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();
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
$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) {
if (-e $oldoutput) {
require IkiWiki::Render;
debug("removing output file $oldoutput");
- IkiWiki::prune($oldoutput);
+ IkiWiki::prune($oldoutput, $config{destdir});
}
}
markunseen($feed->{sourcepage});
}
}
+
+ return $needsbuild;
}
sub preprocess (@) {
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 (<IN>) {
$_=IkiWiki::possibly_foolish_untaint($_);
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) {
close OUT || error("save $newfile: $!", $cleanup);
rename($newfile, "$config{wikistatedir}/aggregate") ||
error("rename $newfile: $!", $cleanup);
+
+ 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 $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};
}
}
# 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
if (! exists $guids{$guid}) {
$guids{$guid}=$myguids{$guid};
}
+ else {
+ foreach my $field (qw{md5}) {
+ $guids{$guid}->{$field}=$myguids{$guid}->{$field};
+ }
+ }
}
}
}
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};
}
$feed->{feedurl}=pop @urls;
}
- my $res=URI::Fetch->fetch($feed->{feedurl});
+ my $res=URI::Fetch->fetch($feed->{feedurl},
+ UserAgent => LWP::UserAgent->new(
+ cookie_jar => $config{cookiejar},
+ ),
+ );
if (! $res) {
$feed->{message}=URI::Fetch->errstr;
$feed->{error}=1;
}
foreach my $entry ($f->entries) {
+ # 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]]
+ local $XML::Atom::ForceUnicode = 1;
+
my $c=$entry->content;
# atom feeds may have no content, only a summary
if (! defined $c && ref $entry->summary) {
# updating an existing post
$guid=$guids{$params{guid}};
return if $guid->{expired};
+ write_page($feed, $guid, $mtime, \%params);
}
else {
# new post
}
my $c="";
while (exists $IkiWiki::pagecase{lc $page.$c} ||
+ -e $IkiWiki::Plugin::transient::transientdir."/".htmlfn($page.$c) ||
-e "$config{srcdir}/".htmlfn($page.$c)) {
$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
+ # (or o
$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++
}
+
+ $guid->{page}=$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
$guid->{md5}=$digest;
# Create the page.
- my $template=template($feed->{template}, blind_cache => 1);
+ my $template;
+ eval {
+ $template=template($feed->{template}, blind_cache => 1);
+ };
+ if ($@) {
+ print STDERR gettext("failed to process template:")." $@";
+ return;
+ }
$template->param(title => $params{title})
if defined $params{title} && length($params{title});
$template->param(content => wikiescape(htmlabs($params{content},
$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}};
}
}
return encode_entities(shift, '\[\]');
}
-sub urlabs ($$) {
- my $url=shift;
- my $urlbase=shift;
-
- URI->new_abs($url, $urlbase)->as_string;
-}
-
sub htmlabs ($$) {
# Convert links in html from relative to absolute.
# Note that this is a heuristic, which is not specified by the rss
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");
}