www: workaround for malformed NNTP links

Some linkifiers to create invalid HTTP links when it sees a
link intended for NNTP services.  This means we may see links
to news.public-inbox.org/inbox.comp.mail.public-inbox.meta
point to "http://" on port 80 instead of 119.  Try to
redirect users to http://public-inbox.org/meta/ in this case.

2 files changed, 121 insertions, 26 deletions

diff --git a/lib/PublicInbox/NewsWWW.pm b/lib/PublicInbox/NewsWWW.pm
new file mode 100644
index 00000000..e19765c9
--- /dev/null
+++ b/lib/PublicInbox/NewsWWW.pm
@@ -0,0 +1,80 @@
+# Copyright (C) 2016 all contributors <meta@public-inbox.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
+#
+# Plack app redirector for mapping /$NEWSGROUP requests to
+# the appropriate /$LISTNAME in PublicInbox::WWW because some
+# auto-linkifiers cannot handle nntp:// redirects properly.
+# This is also used directly by PublicInbox::WWW
+package PublicInbox::NewsWWW;
+use strict;
+use warnings;
+use PublicInbox::Config;
+use URI::Escape qw(uri_escape_utf8);
+
+sub new {
+        my ($class, $pi_config) = @_;
+        $pi_config ||= PublicInbox::Config->new;
+        bless { pi_config => $pi_config }, $class;
+}
+
+sub call {
+        my ($self, $env) = @_;
+        my $ng_map = $self->newsgroup_map;
+        my $path = $env->{PATH_INFO};
+        $path =~ s!\A/+!!;
+        $path =~ s!/+\z!!;
+
+        # some links may have the article number in them:
+        # /inbox.foo.bar/123456
+        my ($ng, $article) = split(m!/+!, $path, 2);
+        if (my $info = $ng_map->{$ng}) {
+                my $url = PublicInbox::Hval::prurl($env, $info->{url});
+                my $code = 301;
+                my $h = [ Location => $url, 'Content-Type' => 'text/plain' ];
+                if (defined $article && $article =~ /\A\d+\z/) {
+                        my $mid = eval { ng_mid_for($ng, $info, $article) };
+                        if (defined $mid) {
+                                # article IDs are not stable across clones,
+                                # do not encourage caching/bookmarking them
+                                $code = 302;
+                                $url .= uri_escape_utf8($mid) . '/';
+                        }
+                }
+
+                return [ $code, $h, [ "Redirecting to $url\n" ] ]
+        }
+        [ 404, [ 'Content-Length' => 'text/plain' ], [] ];
+}
+
+sub ng_mid_for {
+        my ($ng, $info, $article) = @_;
+        # may fail due to lack of Danga::Socket
+        # for defer_weaken:
+        require PublicInbox::NewsGroup;
+        $ng = $info->{ng} ||=
+                PublicInbox::NewsGroup->new($ng, $info->{git_dir}, '');
+        $ng->mm->mid_for($article);
+}
+
+sub newsgroup_map {
+        my ($self) = @_;
+        my $rv;
+        $rv = $self->{ng_map} and return $rv;
+        my $pi_config = $self->{pi_config};
+        my %ng_map;
+        foreach my $k (keys %$pi_config) {
+                $k =~ /\Apublicinbox\.([^\.]+)\.mainrepo\z/ or next;
+                my $listname = $1;
+                my $git_dir = $pi_config->{"publicinbox.$listname.mainrepo"};
+                my $url = $pi_config->{"publicinbox.$listname.url"};
+                defined $url or next;
+                my $ng = $pi_config->{"publicinbox.$listname.newsgroup"};
+                next if (!defined $ng) || ($ng eq ''); # disabled
+
+                $url =~ m!/\z! or $url .= '/';
+                $ng_map{$ng} = { url => $url, git_dir => $git_dir };
+        }
+        $self->{ng_map} = \%ng_map;
+}
+
+1;
diff --git a/lib/PublicInbox/WWW.pm b/lib/PublicInbox/WWW.pm
index 83729273..fd092803 100644
--- a/lib/PublicInbox/WWW.pm
+++ b/lib/PublicInbox/WWW.pm
@@ -45,7 +45,7 @@ sub call {
         if ($method eq 'POST' &&
                  $path_info =~ m!$LISTNAME_RE/(git-upload-pack)\z!) {
                 my $path = $2;
-                return (invalid_list($ctx, $1) ||
+                return (invalid_list($self, $ctx, $1) ||
                         serve_git($cgi, $ctx->{git}, $path));
         }
         elsif ($method !~ /\AGET|HEAD\z/) {
@@ -56,18 +56,19 @@ sub call {
         if ($path_info eq '/') {
                 r404();
         } elsif ($path_info =~ m!$LISTNAME_RE\z!o) {
-                invalid_list($ctx, $1) || r301($ctx, $1);
+                invalid_list($self, $ctx, $1) || r301($ctx, $1);
         } elsif ($path_info =~ m!$LISTNAME_RE(?:/|/index\.html)?\z!o) {
-                invalid_list($ctx, $1) || get_index($ctx);
+                invalid_list($self, $ctx, $1) || get_index($ctx);
         } elsif ($path_info =~ m!$LISTNAME_RE/(?:atom\.xml|new\.atom)\z!o) {
-                invalid_list($ctx, $1) || get_atom($ctx);
+                invalid_list($self, $ctx, $1) || get_atom($ctx);
 
         } elsif ($path_info =~ m!$LISTNAME_RE/
                                 ($PublicInbox::GitHTTPBackend::ANY)\z!ox) {
                 my $path = $2;
-                invalid_list($ctx, $1) || serve_git($cgi, $ctx->{git}, $path);
+                invalid_list($self, $ctx, $1) ||
+                        serve_git($cgi, $ctx->{git}, $path);
         } elsif ($path_info =~ m!$LISTNAME_RE/$MID_RE/$END_RE\z!o) {
-                msg_page($ctx, $1, $2, $3);
+                msg_page($self, $ctx, $1, $2, $3);
 
         # in case people leave off the trailing slash:
         } elsif ($path_info =~ m!$LISTNAME_RE/$MID_RE/(f|T|t|R)\z!o) {
@@ -80,7 +81,7 @@ sub call {
                 r301($ctx, $1, $2);
 
         } else {
-                legacy_redirects($ctx, $path_info);
+                legacy_redirects($self, $ctx, $path_info);
         }
 }
 
@@ -118,7 +119,7 @@ sub r { [ $_[0], ['Content-Type' => 'text/plain'], [ join(' ', @_, "\n") ] ] }
 
 # returns undef if valid, array ref response if invalid
 sub invalid_list {
-        my ($ctx, $listname) = @_;
+        my ($self, $ctx, $listname, $mid) = @_;
         my $git_dir = $ctx->{pi_config}->get($listname, "mainrepo");
         if (defined $git_dir) {
                 $ctx->{git_dir} = $git_dir;
@@ -126,17 +127,23 @@ sub invalid_list {
                 $ctx->{listname} = $listname;
                 return;
         }
-        r404();
+
+        # sometimes linkifiers (not ours!) screw up automatic link
+        # generation and link things intended for nntp:// to https?://,
+        # so try to infer links and redirect them to the appropriate
+        # list URL.
+        $self->news_www->call($ctx->{cgi}->{env});
 }
 
 # returns undef if valid, array ref response if invalid
 sub invalid_list_mid {
-        my ($ctx, $listname, $mid) = @_;
-        my $ret = invalid_list($ctx, $listname, $mid);
+        my ($self, $ctx, $listname, $mid) = @_;
+        my $ret = invalid_list($self, $ctx, $listname, $mid);
         return $ret if $ret;
 
         $ctx->{mid} = $mid = uri_unescape($mid);
         if ($mid =~ /\A[a-f0-9]{40}\z/) {
+                # this is horiffically wasteful for legacy URLs:
                 if ($mid = mid2blob($ctx)) {
                         require Email::Simple;
                         use PublicInbox::MID qw/mid_clean/;
@@ -339,7 +346,7 @@ sub get_thread_atom {
 }
 
 sub legacy_redirects {
-        my ($ctx, $path_info) = @_;
+        my ($self, $ctx, $path_info) = @_;
 
         # single-message pages
         if ($path_info =~ m!$LISTNAME_RE/m/(\S+)/\z!o) {
@@ -384,13 +391,13 @@ sub legacy_redirects {
         # some Message-IDs have slashes in them and the HTTP server
         # may try to be clever and unescape them :<
         } elsif ($path_info =~ m!$LISTNAME_RE/(\S+/\S+)/$END_RE\z!o) {
-                msg_page($ctx, $1, $2, $3);
+                msg_page($self, $ctx, $1, $2, $3);
 
         # in case people leave off the trailing slash:
         } elsif ($path_info =~ m!$LISTNAME_RE/(\S+/\S+)/(f|T|t)\z!o) {
                 r301($ctx, $1, $2, $3 eq 't' ? 't/#u' : $3);
         } else {
-                r404();
+                $self->news_www->call($ctx->{cgi}->{env});
         }
 }
 
@@ -410,18 +417,18 @@ sub r301 {
 }
 
 sub msg_page {
-        my ($ctx, $list, $mid, $e) = @_;
-        unless (invalid_list_mid($ctx, $list, $mid)) {
-                '' eq $e and return get_mid_html($ctx);
-                't/' eq $e and return get_thread($ctx);
-                't.atom' eq $e and return get_thread_atom($ctx);
-                't.mbox' eq $e and return get_thread_mbox($ctx);
-                't.mbox.gz' eq $e and return get_thread_mbox($ctx, '.gz');
-                'T/' eq $e and return get_thread($ctx, 1);
-                'raw' eq $e and return get_mid_txt($ctx);
-                'f/' eq $e and return get_full_html($ctx);
-                'R/' eq $e and return get_reply_html($ctx);
-        }
+        my ($self, $ctx, $list, $mid, $e) = @_;
+        my $ret;
+        $ret = invalid_list_mid($self, $ctx, $list, $mid) and return $ret;
+        '' eq $e and return get_mid_html($ctx);
+        't/' eq $e and return get_thread($ctx);
+        't.atom' eq $e and return get_thread_atom($ctx);
+        't.mbox' eq $e and return get_thread_mbox($ctx);
+        't.mbox.gz' eq $e and return get_thread_mbox($ctx, '.gz');
+        'T/' eq $e and return get_thread($ctx, 1);
+        'raw' eq $e and return get_mid_txt($ctx);
+        'f/' eq $e and return get_full_html($ctx);
+        'R/' eq $e and return get_reply_html($ctx);
         r404($ctx);
 }
 
@@ -430,4 +437,12 @@ sub serve_git {
         PublicInbox::GitHTTPBackend::serve($cgi, $git, $path);
 }
 
+sub news_www {
+        my ($self) = @_;
+        my $nw = $self->{news_www};
+        return $nw if $nw;
+        require PublicInbox::NewsWWW;
+        $self->{news_www} = PublicInbox::NewsWWW->new($self->{pi_config});
+}
+
 1;