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 # 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;
13 use PublicInbox::Search qw(mdocid);
15 # called by PSGI server as body response
16 # this gets called twice for every message, once to return the header,
17 # once to retrieve the body
20 my $smsg = $ctx->{smsg} or return;
21 my $ibx = $ctx->{-inbox};
22 my $eml = $ibx->smsg_eml($smsg) or return;
23 my $n = $ctx->{smsg} = $ibx->over->next_by_mid(@{$ctx->{next_arg}});
24 $ctx->zmore(msg_hdr($ctx, $eml, $smsg->{mid}));
26 $ctx->translate(msg_body($eml));
27 } else { # last message
28 $ctx->zmore(msg_body($eml));
33 # called by PublicInbox::DS::write after http->next_step
35 my ($http) = @_; # PublicInbox::HTTP
36 my $ctx = $http->{forward} or return; # client aborted
38 my $smsg = $ctx->{smsg} or return $ctx->close;
39 $ctx->smsg_blob($smsg);
44 sub async_eml { # for async_blob_cb
46 my $smsg = delete $ctx->{smsg};
48 $ctx->{smsg} = $ctx->{-inbox}->over->next_by_mid(@{$ctx->{next_arg}});
50 $ctx->zmore(msg_hdr($ctx, $eml, $smsg->{mid}));
51 $ctx->{http_out}->write($ctx->translate(msg_body($eml)));
55 my ($ctx, $subject) = @_;
56 my $fn = $subject // '';
58 $fn = to_filename($fn) // 'no-subject';
59 my @hdr = ('Content-Type');
60 if ($ctx->{-inbox}->{obfuscate}) {
61 # obfuscation is stupid, but maybe scrapers are, too...
62 push @hdr, 'application/mbox';
65 push @hdr, 'text/plain';
68 push @hdr, 'Content-Disposition', "inline; filename=$fn";
72 # for rare cases where v1 inboxes aren't indexed w/ ->over at all
75 my $mref = $ctx->{-inbox}->msg_by_mid($ctx->{mid}) or return;
76 my $eml = PublicInbox::Eml->new($mref);
77 [ 200, res_hdr($ctx, $eml->header_str('Subject')),
78 [ msg_hdr($ctx, $eml, $ctx->{mid}) . msg_body($eml) ] ]
81 # /$INBOX/$MESSAGE_ID/raw
84 $ctx->{base_url} = $ctx->{-inbox}->base_url($ctx->{env});
85 my $over = $ctx->{-inbox}->over or return no_over_raw($ctx);
87 my $mip = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
88 my $smsg = $ctx->{smsg} = $over->next_by_mid(@$mip) or return;
89 my $res_hdr = res_hdr($ctx, $smsg->{subject});
90 bless $ctx, __PACKAGE__;
91 $ctx->psgi_response(200, $res_hdr);
95 my ($ctx, $eml, $mid) = @_;
96 my $header_obj = $eml->header_obj;
98 # drop potentially confusing headers, ssoma already should've dropped
99 # Lines and Content-Length
100 foreach my $d (qw(Lines Bytes Content-Length Status)) {
101 $header_obj->header_set($d);
103 my $ibx = $ctx->{-inbox};
104 my $base = $ctx->{base_url};
105 $mid = $ctx->{mid} unless defined $mid;
106 $mid = mid_escape($mid);
108 'Archived-At', "<$base$mid/>",
109 'List-Archive', "<$base>",
110 'List-Post', "<mailto:$ibx->{-primary_address}>",
112 my $crlf = $header_obj->crlf;
113 my $buf = $header_obj->as_string;
114 # fixup old bug from import (pre-a0c07cba0e5d8b6a)
115 $buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
116 $buf = "From mboxrd\@z Thu Jan 1 00:00:00 1970" . $crlf . $buf;
118 for (my $i = 0; $i < @append; $i += 2) {
120 my $v = $append[$i + 1];
121 my @v = $header_obj->header_raw($k);
128 $buf .= "$k: $v$crlf" if defined $v;
134 my $bdy = $_[0]->{bdy} // return "\n";
135 # mboxrd quoting style
136 # https://en.wikipedia.org/wiki/Mbox#Modified_mbox
137 # https://www.loc.gov/preservation/digital/formats/fdd/fdd000385.shtml
138 # https://web.archive.org/http://www.qmail.org/man/man5/mbox.html
139 $$bdy =~ s/^(>*From )/>$1/gm;
145 my $msgs = $ctx->{msgs};
147 if (my $smsg = shift @$msgs) {
151 $ctx->{msgs} = $msgs = $ctx->{over}->get_thread($ctx->{mid},
153 return unless @$msgs;
154 $ctx->{prev} = $msgs->[-1];
159 my ($ctx, $over, $sfx) = @_;
160 my $msgs = $ctx->{msgs} = $over->get_thread($ctx->{mid}, {});
161 return [404, [qw(Content-Type text/plain)], []] if !@$msgs;
162 $ctx->{prev} = $msgs->[-1];
163 $ctx->{over} = $over; # bump refcnt
164 require PublicInbox::MboxGz;
165 PublicInbox::MboxGz::mbox_gz($ctx, \&thread_cb, $msgs->[0]->{subject});
169 my ($ctx, $range) = @_;
172 if ($range eq 'all') { # TODO: YYYY[-MM]
175 return [404, [qw(Content-Type text/plain)], []];
177 mbox_all($ctx, $query);
182 my $ids = $ctx->{ids};
184 while ((my $num = shift @$ids)) {
185 my $smsg = $ctx->{over}->get_art($num) or next;
188 $ctx->{ids} = $ids = $ctx->{mm}->ids_after(\($ctx->{prev}));
194 my $ibx = $ctx->{-inbox};
196 my $mm = $ctx->{mm} = $ibx->mm;
197 my $ids = $mm->ids_after(\$prev) or return
198 [404, [qw(Content-Type text/plain)], ["No results found\n"]];
199 $ctx->{over} = $ibx->over or
200 return PublicInbox::WWW::need($ctx, 'Overview');
202 $ctx->{prev} = $prev;
203 require PublicInbox::MboxGz;
204 PublicInbox::MboxGz::mbox_gz($ctx, \&all_ids_cb, 'all');
209 my $over = $ctx->{-inbox}->over or return;
211 while (defined(my $num = shift(@{$ctx->{ids}}))) {
212 my $smsg = $over->get_art($num) or next;
216 my $srch = $ctx->{-inbox}->search(undef, $ctx) or return;
217 my $mset = $srch->query($ctx->{query}, $ctx->{qopts});
218 my $size = $mset->size or return;
219 $ctx->{qopts}->{offset} += $size;
220 my $nshard = $srch->{nshard} // 1;
221 $ctx->{ids} = [ map { mdocid($nshard, $_) } $mset->items ];
226 my ($ctx, $query) = @_;
228 return mbox_all_ids($ctx) if $query eq '';
229 my $qopts = $ctx->{qopts} = { mset => 2 }; # order by docid
230 my $srch = $ctx->{-inbox}->search or
231 return PublicInbox::WWW::need($ctx, 'Search');
232 my $mset = $srch->query($query, $qopts);
233 $qopts->{offset} = $mset->size or
234 return [404, [qw(Content-Type text/plain)],
235 ["No results found\n"]];
236 $ctx->{query} = $query;
237 my $nshard = $srch->{nshard} // 1;
238 $ctx->{ids} = [ map { mdocid($nshard, $_) } $mset->items ];
239 require PublicInbox::MboxGz;
240 PublicInbox::MboxGz::mbox_gz($ctx, \&results_cb, 'results-'.$query);