1 # Copyright (C) 2015-2021 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->isrch or return;
36 my $opt = { limit => PARTIAL_MAX, relevance => -1 };
37 my @try = ("m:$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 { $srch->mset($m, $opt) } or next;
71 } @{$srch->mset_to_smsg($ibx, $mset)};
72 return \@mids if scalar(@mids);
77 my ($other, $ctx) = @_;
79 return if $other->{name} eq $ctx->{ibx}->{name} || !$other->base_url;
81 my $mm = $other->mm or return;
83 # try to find the URL with Msgmap to avoid forking
84 my $num = $mm->num_for($ctx->{mid});
86 push @{$ctx->{found}}, $other;
88 # no point in trying the fork fallback if we
89 # know Xapian is up-to-date but missing the
90 # message in the current repo
91 push @{$ctx->{again}}, $other;
96 my ($pi_cfg, $section, $ctx) = @_;
97 if (defined($section)) {
98 return if $section !~ m!\Apublicinbox\.([^/]+)\z!;
99 my $ibx = $pi_cfg->lookup_name($1) or return;
100 ext_msg_i($ibx, $ctx);
101 } else { # undef == "EOF"
102 finalize_exact($ctx);
106 sub ext_msg_ALL ($) {
108 my $ALL = $ctx->{www}->{pi_cfg}->ALL or return;
109 my $by_eidx_key = $ctx->{www}->{pi_cfg}->{-by_eidx_key};
110 my $cur_key = eval { $ctx->{ibx}->eidx_key } //
111 return partial_response($ctx); # $cur->{ibx} == $ALL
112 my %seen = ($cur_key => 1);
114 while (my $x = $ALL->over->next_by_mid($ctx->{mid}, \$id, \$prev)) {
115 my $xr3 = $ALL->over->get_xref3($x->{num});
117 $k =~ s/:[0-9]+:$x->{blob}\z// or next;
118 next if $k eq $cur_key;
119 my $ibx = $by_eidx_key->{$k} // next;
120 $ibx->base_url or next;
121 push(@{$ctx->{found}}, $ibx) unless $seen{$k}++;
124 return exact($ctx) if $ctx->{found};
126 # fall back to partial MID matching
127 for my $ibxish ($ctx->{ibx}, $ALL) {
128 my $mids = search_partial($ibxish, $ctx->{mid}) or next;
129 push @{$ctx->{partial}}, [ $ibxish, $mids ];
130 last if ($ctx->{n_partial} += scalar(@$mids)) >= PARTIAL_MAX;
132 partial_response($ctx);
137 ext_msg_ALL($ctx) // sub {
138 $ctx->{-wcb} = $_[0]; # HTTP server write callback
140 if ($ctx->{env}->{'pi-httpd.async'}) {
141 require PublicInbox::ConfigIter;
142 my $iter = PublicInbox::ConfigIter->new(
143 $ctx->{www}->{pi_cfg},
144 \&ext_msg_step, $ctx);
147 $ctx->{www}->{pi_cfg}->each_inbox(\&ext_msg_i, $ctx);
148 finalize_exact($ctx);
153 # called via PublicInbox::DS::event_loop
155 my ($ctx, $sync) = @_;
156 # can't find a partial match in current inbox, try the others:
157 my $ibx = shift @{$ctx->{again}} or return finalize_partial($ctx);
158 my $mids = search_partial($ibx, $ctx->{mid}) or
159 return ($sync ? undef : PublicInbox::DS::requeue($ctx));
160 $ctx->{n_partial} += scalar(@$mids);
161 push @{$ctx->{partial}}, [ $ibx, $mids ];
162 $ctx->{n_partial} >= PARTIAL_MAX ? finalize_partial($ctx)
163 : ($sync ? undef : PublicInbox::DS::requeue($ctx));
169 return $ctx->{-wcb}->(exact($ctx)) if $ctx->{found};
171 # fall back to partial MID matching
172 my $mid = $ctx->{mid};
173 my $cur = $ctx->{ibx};
174 my $mids = search_partial($cur, $mid);
176 $ctx->{n_partial} = scalar(@$mids);
177 push @{$ctx->{partial}}, [ $cur, $mids ];
178 } elsif ($ctx->{again} && length($mid) >= $MIN_PARTIAL_LEN) {
179 bless $ctx, __PACKAGE__;
180 if ($ctx->{env}->{'pi-httpd.async'}) {
185 # synchronous fall-through
186 $ctx->event_step while @{$ctx->{again}};
188 finalize_partial($ctx);
193 (index($u, '://') < 0 && index($u, '/') != 0) ?
194 "$ctx->{-upfx}../$u" : $u;
197 sub partial_response ($) {
199 my $mid = $ctx->{mid};
201 my $href = mid_href($mid);
202 my $html = ascii_html($mid);
203 my $title = "<$html> not found";
204 my $s = "<pre>Message-ID <$html>\nnot found\n";
205 $ctx->{-upfx} //= '../';
206 if (my $n_partial = $ctx->{n_partial}) {
208 my $es = $n_partial == 1 ? '' : 'es';
209 $n_partial .= '+' if ($n_partial == PARTIAL_MAX);
210 $s .= "\n$n_partial partial match$es found:\n\n";
211 my $cur_name = $ctx->{ibx}->{name};
212 foreach my $pair (@{$ctx->{partial}}) {
213 my ($ibx, $res) = @$pair;
214 my $e = $ibx->{name} eq $cur_name ? $ctx->{env} : undef;
215 my $u = _url_pfx($ctx, $ibx->base_url($e) // next);
216 foreach my $m (@$res) {
217 my $href = mid_href($m);
218 my $html = ascii_html($m);
219 $s .= qq{<a\nhref="$u$href/">$u$html/</a>\n};
223 my $ext = ext_urls($ctx, $mid, $href, $html);
228 $ctx->{-html_tip} = $s .= '</pre>';
229 $ctx->{-title_html} = $title;
230 html_oneshot($ctx, $code);
233 sub finalize_partial ($) { $_[0]->{-wcb}->(partial_response($_[0])) }
236 my ($ctx, $mid, $href, $html) = @_;
238 # Fall back to external repos if configured
239 if (@EXT_URL && index($mid, '@') >= 0) {
240 my $env = $ctx->{env};
241 my $e = "\nPerhaps try an external site:\n\n";
242 foreach my $url (@EXT_URL) {
243 my $u = prurl($env, $url);
244 my $r = sprintf($u, $href);
245 my $t = sprintf($u, $html);
246 $e .= qq{<a\nhref="$r">$t</a>\n};
255 my $mid = $ctx->{mid};
256 my $found = $ctx->{found};
257 my $href = mid_href($mid);
258 my $html = ascii_html($mid);
259 my $title = "<$html> found in ";
260 my $end = @$found == 1 ? 'another inbox' : 'other inboxes';
261 $ctx->{-title_html} = $title . $end;
262 $ctx->{-upfx} //= '../';
263 my $ext_urls = ext_urls($ctx, $mid, $href, $html);
264 my $code = (@$found == 1 && $ext_urls eq '') ? 200 : 300;
265 $ctx->{-html_tip} = join('',
266 "<pre>Message-ID: <$html>\nfound in $end:\n\n",
268 my $u = _url_pfx($ctx, $_->base_url);
269 qq(<a\nhref="$u$href/">$u$html/</a>\n)
271 $ext_urls, '</pre>');
272 html_oneshot($ctx, $code);