* Fix aggregator to not warn when a feed contains no body content at all.

* Work around bug #420636 by, if XML::Parser crashes, running the feed
  content though Encode::decode_utf8.
master
joey 2007-04-23 18:36:44 +00:00
parent 6c6413c9f5
commit 71d2d93027
4 changed files with 53 additions and 10 deletions

View File

@ -229,6 +229,8 @@ sub expire () { #{{{
sub aggregate () { #{{{
eval q{use XML::Feed};
error($@) if $@;
eval q{use URI::Fetch};
error($@) if $@;
eval q{use HTML::Entities};
error($@) if $@;
@ -237,6 +239,9 @@ sub aggregate () { #{{{
time - $feed->{lastupdate} >= $feed->{updateinterval};
$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}));
@ -251,7 +256,29 @@ sub aggregate () { #{{{
}
$feed->{feedurl}=pop @urls;
}
my $f=eval{XML::Feed->parse(URI->new($feed->{feedurl}))};
my $res=URI::Fetch->fetch($feed->{feedurl});
if (! $res) {
$feed->{message}=URI::Fetch->errstr;
$feed->{error}=1;
debug($feed->{message});
next;
}
if ($res->status == URI::Fetch::URI_GONE()) {
$feed->{message}=gettext("feed not found");
$feed->{error}=1;
debug($feed->{message});
next;
}
my $content=$res->content;
my $f=eval{XML::Feed->parse(\$content)};
if ($@) {
# 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);
$f=eval{XML::Feed->parse(\$content)};
}
if ($@) {
$feed->{message}=gettext("feed crashed XML::Feed!")." ($@)";
$feed->{error}=1;
@ -270,15 +297,11 @@ sub aggregate () { #{{{
feed => $feed,
title => defined $entry->title ? decode_entities($entry->title) : "untitled",
link => $entry->link,
content => $entry->content->body,
content => defined $entry->content->body ? $entry->content->body : "",
guid => defined $entry->id ? $entry->id : time."_".$feed->name,
ctime => $entry->issued ? ($entry->issued->epoch || time) : time,
);
}
$feed->{message}=sprintf(gettext("processed ok at %s"),
displaytime($feed->{lastupdate}));
$feed->{error}=0;
}
} #}}}

5
debian/changelog vendored
View File

@ -5,8 +5,11 @@ ikiwiki (1.51) UNRELEASED; urgency=low
* Minor template improvements by Alessandro.
* In mercurial backend, untaint ipaddr when using it as the user for the
commit. Thanks, Alexander Wirt. Closes: #420428
* Fix aggregator to not warn when a feed contains no body content at all.
* Work around bug #420636 by, if XML::Parser crashes, running the feed
content though Encode::decode_utf8.
-- Joey Hess <joeyh@debian.org> Sun, 22 Apr 2007 13:43:49 -0400
-- Joey Hess <joeyh@debian.org> Mon, 23 Apr 2007 14:31:57 -0400
ikiwiki (1.50) unstable; urgency=low

View File

@ -13,7 +13,21 @@ When I run ikiwiki with "--aggregate" I get this error:
Use of uninitialized value in subroutine entry at /usr/share/perl5/IkiWiki/Plugin/aggregate.pm line 414.
Use of uninitialized value in subroutine entry at /usr/share/perl5/IkiWiki/Plugin/aggregate.pm line 414.
Also, feeds from DokuWiki seem to crash the aggregate plugin completely, it's not a completely valid feed but presumably crashing is still bad. The feed I'm seeing this with is http://www.wirelesscommons.org/feed.php
> Fixed, this occurred when a feed did not include any body content tag.
> --[[Joey]]
Also, feeds from DokuWiki seem to crash the aggregate plugin completely,
it's not a completely valid feed but presumably crashing is still bad. The
feed I'm seeing this with is http://www.wirelesscommons.org/feed.php
> This is a bug in XML::Parser. Unfortunately, perl does not have a feed
> parser that handles invalid feeds, and in particular, XML::Parser has
> issues with feeds that claim to be encoded in utf-8 and contain invalid
> utf sequences, as well as other encoding issues. See also [[debbug 380426]].
> Note though that this invalid feed does not really crash the aggregate plugin,
> it just notes that XML::Parser crashed on it and continues. This is the
> best I can do in ikiwiki. I have filed a bug on XML::Parser about this,
> it's [[debbug 420636]]. I've also put in a workaround, so [[done]].
-- System Information:
Debian Release: 3.1

View File

@ -1,3 +1,6 @@
I have the same problem here when I use a feed from googles shared feed. http://www.google.com/reader/public/atom/user/04715560304044435944/state/com.google/broadcast
john
> I cannot reproduce any problem with this feed. Can you provide details?
> --[[Joey]]