1 # Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Used by the web interface to link to messages outside of the our
5 # public-inboxes. Mail threads may cross projects/threads; so
6 # we should ensure users can find more easily find them on other
8 package PublicInbox::ExtMsg;
11 use PublicInbox::Hval qw(ascii_html prurl mid_href);
12 use PublicInbox::WwwStream qw(html_oneshot);
13 use PublicInbox::Smsg;
14 our $MIN_PARTIAL_LEN = 16;
16 # TODO: user-configurable
17 our @EXT_URL = map { ascii_html($_) } (
18 # leading "//" denotes protocol-relative (http:// or https://)
20 '//www.mail-archive.com/search?l=mid&q=%s',
21 'nntp://news.gmane.io/%s',
22 'https://lists.debian.org/msgid-search/%s',
23 '//docs.FreeBSD.org/cgi/mid.cgi?db=mid&id=%s',
24 'https://www.w3.org/mid/%s',
25 'http://www.postgresql.org/message-id/%s',
26 'https://lists.debconf.org/cgi-lurker/keyword.cgi?'.
27 'doc-url=/lurker&format=en.html&query=id:%s'
30 sub PARTIAL_MAX () { 100 }
32 sub search_partial ($$) {
34 return if length($mid) < $MIN_PARTIAL_LEN;
35 my $srch = $ibx->search or return;
36 my $opt = { limit => PARTIAL_MAX, mset => 2 };
37 my @try = ("m:$mid*");
39 if ($chop =~ s/(\W+)(\w*)\z//) {
40 my ($delim, $word) = ($1, $2);
42 push @try, "m:$chop$delim";
43 push @try, "m:$chop$delim*";
46 push @try, "m:$chop*";
49 # break out long words individually to search for, because
50 # too many messages begin with "Pine.LNX." (or "alpine" or "nycvar")
51 if ($mid =~ /\w{9,}/) {
52 my @long = ($mid =~ m!(\w{3,})!g);
53 push(@try, join(' ', map { "m:$_" } @long));
55 # is the last element long enough to not trigger excessive
57 if (length($long[-1]) > 8) {
59 push(@try, join(' ', map { "m:$_" } @long));
63 foreach my $m (@try) {
64 # If Xapian can't handle the wildcard since it
65 # has too many results. $@ can be
66 # Search::Xapian::QueryParserError or even:
67 # "something terrible happened at ../Search/Xapian/Enquire.pm"
68 my $mset = eval { $srch->query($m, $opt) } or next;
71 } @{$ibx->over->get_all(@{$srch->mset_to_artnums($mset)})};
72 return \@mids if scalar(@mids);
77 my ($other, $arg) = @_;
78 my ($cur, $mid, $ibxs, $found) = @$arg;
80 return if $other->{name} eq $cur->{name} || !$other->base_url;
82 my $mm = $other->mm or return;
84 # try to find the URL with Msgmap to avoid forking
85 my $num = $mm->num_for($mid);
89 # no point in trying the fork fallback if we
90 # know Xapian is up-to-date but missing the
91 # message in the current repo
98 my $cur = $ctx->{-inbox};
99 my $mid = $ctx->{mid};
101 eval { require PublicInbox::Msgmap };
104 my $arg = [ $cur, $mid, $ibxs, $found ];
106 $ctx->{www}->{pi_config}->each_inbox(\&ext_msg_i, $arg);
108 return exact($ctx, $found, $mid) if @$found;
110 # fall back to partial MID matching
113 my $mids = search_partial($cur, $mid);
115 $n_partial = scalar(@$mids);
116 push @partial, [ $cur, $mids ];
119 # can't find a partial match in current inbox, try the others:
120 if (!$n_partial && length($mid) >= $MIN_PARTIAL_LEN) {
121 foreach my $ibx (@$ibxs) {
122 $mids = search_partial($ibx, $mid) or next;
123 $n_partial += scalar(@$mids);
124 push @partial, [ $ibx, $mids];
125 last if $n_partial >= PARTIAL_MAX;
130 my $href = mid_href($mid);
131 my $html = ascii_html($mid);
132 my $title = "<$html> not found";
133 my $s = "<pre>Message-ID <$html>\nnot found\n";
136 my $es = $n_partial == 1 ? '' : 'es';
137 $n_partial .= '+' if ($n_partial == PARTIAL_MAX);
138 $s .= "\n$n_partial partial match$es found:\n\n";
139 my $cur_name = $cur->{name};
140 foreach my $pair (@partial) {
141 my ($ibx, $res) = @$pair;
142 my $env = $ctx->{env} if $ibx->{name} eq $cur_name;
143 my $u = $ibx->base_url($env) or next;
144 foreach my $m (@$res) {
145 my $href = mid_href($m);
146 my $html = ascii_html($m);
147 $s .= qq{<a\nhref="$u$href/">$u$html/</a>\n};
151 my $ext = ext_urls($ctx, $mid, $href, $html);
156 $ctx->{-html_tip} = $s .= '</pre>';
157 $ctx->{-title_html} = $title;
158 $ctx->{-upfx} = '../';
159 html_oneshot($ctx, $code);
163 my ($ctx, $mid, $href, $html) = @_;
165 # Fall back to external repos if configured
166 if (@EXT_URL && index($mid, '@') >= 0) {
167 my $env = $ctx->{env};
168 my $e = "\nPerhaps try an external site:\n\n";
169 foreach my $url (@EXT_URL) {
170 my $u = prurl($env, $url);
171 my $r = sprintf($u, $href);
172 my $t = sprintf($u, $html);
173 $e .= qq{<a\nhref="$r">$t</a>\n};
181 my ($ctx, $found, $mid) = @_;
182 my $href = mid_href($mid);
183 my $html = ascii_html($mid);
184 my $title = "<$html> found in ";
185 my $end = @$found == 1 ? 'another inbox' : 'other inboxes';
186 $ctx->{-title_html} = $title . $end;
187 $ctx->{-upfx} = '../';
188 my $ext_urls = ext_urls($ctx, $mid, $href, $html);
189 my $code = (@$found == 1 && $ext_urls eq '') ? 200 : 300;
190 $ctx->{-html_tip} = join('',
191 "<pre>Message-ID: <$html>\nfound in $end:\n\n",
193 my $u = $_->base_url;
194 qq(<a\nhref="$u$href/">$u$html/</a>\n)
196 $ext_urls, '</pre>');
197 html_oneshot($ctx, $code);