1 # Copyright (C) all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Streaming interface for mboxrd HTTP responses
5 # See PublicInbox::GzipFilter for details.
6 package PublicInbox::Mbox;
8 use parent 'PublicInbox::GzipFilter';
9 use PublicInbox::MID qw/mid_escape/;
10 use PublicInbox::Hval qw/to_filename/;
11 use PublicInbox::Smsg;
14 # called by PSGI server as body response
15 # this gets called twice for every message, once to return the header,
16 # once to retrieve the body
19 my $smsg = $ctx->{smsg} or return;
20 my $ibx = $ctx->{ibx};
21 my $eml = delete($ctx->{eml}) // $ibx->smsg_eml($smsg) // return;
22 if (($ctx->{smsg} = $ibx->over->next_by_mid(@{$ctx->{next_arg}}))) {
23 $ctx->translate(msg_hdr($ctx, $eml), msg_body($eml));
24 } else { # last message
25 $ctx->zflush(msg_hdr($ctx, $eml), msg_body($eml));
29 # called by PublicInbox::DS::write after http->next_step
31 my ($http) = @_; # PublicInbox::HTTP
32 my $ctx = $http->{forward} or return; # client aborted
34 my $smsg = $ctx->{smsg} or return $ctx->close;
35 $ctx->smsg_blob($smsg);
40 sub async_eml { # for async_blob_cb
42 my $smsg = delete $ctx->{smsg};
44 $ctx->{smsg} = $ctx->{ibx}->over->next_by_mid(@{$ctx->{next_arg}});
45 local $ctx->{eml} = $eml; # for mbox_hdr
46 $ctx->write(msg_hdr($ctx, $eml), msg_body($eml));
51 my $eml = $ctx->{eml} //= $ctx->{ibx}->smsg_eml($ctx->{smsg});
52 my $fn = $eml->header_str('Subject') // '';
54 $fn = to_filename($fn) // 'no-subject';
55 my @hdr = ('Content-Type');
56 if ($ctx->{ibx}->{obfuscate}) {
57 # obfuscation is stupid, but maybe scrapers are, too...
58 push @hdr, 'application/mbox';
61 push @hdr, 'text/plain';
64 my $cs = $ctx->{eml}->ct->{attributes}->{charset} // 'UTF-8';
65 $cs = 'UTF-8' if $cs =~ /[^a-zA-Z0-9\-\_]/; # avoid header injection
66 $hdr[-1] .= "; charset=$cs";
67 push @hdr, 'Content-Disposition', "inline; filename=$fn";
71 # for rare cases where v1 inboxes aren't indexed w/ ->over at all
74 my $mref = $ctx->{ibx}->msg_by_mid($ctx->{mid}) or return;
75 my $eml = $ctx->{eml} = PublicInbox::Eml->new($mref);
76 [ @{mbox_hdr($ctx)}, [ msg_hdr($ctx, $eml) . msg_body($eml) ] ]
79 # /$INBOX/$MESSAGE_ID/raw
82 my $over = $ctx->{ibx}->over or return no_over_raw($ctx);
84 my $mip = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
85 my $smsg = $ctx->{smsg} = $over->next_by_mid(@$mip) or return;
86 bless $ctx, __PACKAGE__;
87 $ctx->psgi_response(\&mbox_hdr);
92 my $header_obj = $eml->header_obj;
94 # drop potentially confusing headers, ssoma already should've dropped
95 # Lines and Content-Length
96 foreach my $d (qw(Lines Bytes Content-Length Status)) {
97 $header_obj->header_set($d);
99 my $crlf = $header_obj->crlf;
100 my $buf = $header_obj->as_string;
101 # fixup old bug from import (pre-a0c07cba0e5d8b6a)
102 $buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
103 "From mboxrd\@z Thu Jan 1 00:00:00 1970" . $crlf . $buf . $crlf;
107 my $bdy = $_[0]->{bdy} // return "\n";
108 # mboxrd quoting style
109 # https://en.wikipedia.org/wiki/Mbox#Modified_mbox
110 # https://www.loc.gov/preservation/digital/formats/fdd/fdd000385.shtml
111 # https://web.archive.org/http://www.qmail.org/man/man5/mbox.html
112 $$bdy =~ s/^(>*From )/>$1/gm;
118 my $msgs = $ctx->{msgs};
120 if (my $smsg = shift @$msgs) {
124 my $over = $ctx->{ibx}->over or return $ctx->gone('over');
125 $ctx->{msgs} = $msgs = $over->get_thread($ctx->{mid},
127 return unless @$msgs;
128 $ctx->{prev} = $msgs->[-1];
133 my ($ctx, $over, $sfx) = @_;
134 my $msgs = $ctx->{msgs} = $over->get_thread($ctx->{mid}, {});
135 return [404, [qw(Content-Type text/plain)], []] if !@$msgs;
136 $ctx->{prev} = $msgs->[-1];
137 require PublicInbox::MboxGz;
138 PublicInbox::MboxGz::mbox_gz($ctx, \&thread_cb, $msgs->[0]->{subject});
142 my ($ctx, $range) = @_;
145 if ($range eq 'all') { # TODO: YYYY[-MM]
148 return [404, [qw(Content-Type text/plain)], []];
150 mbox_all($ctx, { q => $q });
155 my $over = $ctx->{ibx}->over or return $ctx->gone('over');
156 my $ids = $ctx->{ids};
158 while ((my $num = shift @$ids)) {
159 my $smsg = $over->get_art($num) or next;
162 $ctx->{ids} = $ids = $over->ids_after(\($ctx->{prev}));
169 my $over = $ctx->{ibx}->over or
170 return PublicInbox::WWW::need($ctx, 'Overview');
171 my $ids = $over->ids_after(\$prev) or return
172 [404, [qw(Content-Type text/plain)], ["No results found\n"]];
174 $ctx->{prev} = $prev;
175 $ctx->{-low_prio} = 1;
176 require PublicInbox::MboxGz;
177 PublicInbox::MboxGz::mbox_gz($ctx, \&all_ids_cb, 'all');
182 my $over = $ctx->{ibx}->over or return $ctx->gone('over');
184 while (defined(my $num = shift(@{$ctx->{ids}}))) {
185 my $smsg = $over->get_art($num) or next;
188 # refill result set, deprioritize since there's many results
189 my $srch = $ctx->{ibx}->isrch or return $ctx->gone('search');
190 my $mset = $srch->mset($ctx->{query}, $ctx->{qopts});
191 my $size = $mset->size or return;
192 $ctx->{qopts}->{offset} += $size;
193 $ctx->{ids} = $srch->mset_to_artnums($mset, $ctx->{qopts});
194 $ctx->{-low_prio} = 1;
198 sub results_thread_cb {
201 my $over = $ctx->{ibx}->over or return $ctx->gone('over');
203 while (defined(my $num = shift(@{$ctx->{xids}}))) {
204 my $smsg = $over->get_art($num) or next;
208 # refills ctx->{xids}
209 next if $over->expand_thread($ctx);
211 # refill result set, deprioritize since there's many results
212 my $srch = $ctx->{ibx}->isrch or return $ctx->gone('search');
213 my $mset = $srch->mset($ctx->{query}, $ctx->{qopts});
214 my $size = $mset->size or return;
215 $ctx->{qopts}->{offset} += $size;
216 $ctx->{ids} = $srch->mset_to_artnums($mset, $ctx->{qopts});
217 $ctx->{-low_prio} = 1;
224 my $q_string = $q->{'q'};
225 return mbox_all_ids($ctx) if $q_string !~ /\S/;
226 my $srch = $ctx->{ibx}->isrch or
227 return PublicInbox::WWW::need($ctx, 'Search');
228 my $over = $ctx->{ibx}->over or
229 return PublicInbox::WWW::need($ctx, 'Overview');
231 my $qopts = $ctx->{qopts} = { relevance => -2 }; # ORDER BY docid DESC
232 $qopts->{threads} = 1 if $q->{t};
233 $srch->query_approxidate($ctx->{ibx}->git, $q_string);
234 my $mset = $srch->mset($q_string, $qopts);
235 $qopts->{offset} = $mset->size or
236 return [404, [qw(Content-Type text/plain)],
237 ["No results found\n"]];
238 $ctx->{query} = $q_string;
239 $ctx->{ids} = $srch->mset_to_artnums($mset, $qopts);
240 require PublicInbox::MboxGz;
242 if ($q->{t} && $srch->has_threadid) {
243 $fn = 'results-thread-'.$q_string;
244 PublicInbox::MboxGz::mbox_gz($ctx, \&results_thread_cb, $fn);
246 $fn = 'results-'.$q_string;
247 PublicInbox::MboxGz::mbox_gz($ctx, \&results_cb, $fn);