1 # Copyright (C) 2014-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Used for displaying the HTML web interface.
5 # See Documentation/design_www.txt for this.
6 package PublicInbox::View;
9 use bytes (); # only for bytes::length
10 use PublicInbox::MsgTime qw(msg_datestamp);
11 use PublicInbox::Hval qw(ascii_html obfuscate_addrs prurl mid_href);
12 use PublicInbox::Linkify;
13 use PublicInbox::MID qw(id_compress mids mids_for_index references
15 use PublicInbox::MsgIter;
16 use PublicInbox::Address;
17 use PublicInbox::WwwStream;
18 use PublicInbox::Reply;
19 use PublicInbox::ViewDiff qw(flush_diff);
20 use POSIX qw(strftime);
21 use Time::Local qw(timegm);
22 use PublicInbox::Smsg qw(subject_normalized);
23 use constant COLS => 72;
24 use constant INDENT => ' ';
25 use constant TCHILD => '` ';
26 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
30 if (my $more = delete $ctx->{more}) { # unlikely
31 # fake an EOF if $more retrieval fails;
32 eval { msg_page_more($ctx, $nr, @$more) };
33 } elsif (my $hdr = delete $ctx->{hdr}) {
34 # fake an EOF if generating the footer fails;
35 # we want to at least show the message if something
37 eval { html_footer($ctx, $hdr) };
43 # public functions: (unstable)
47 my $mid = $ctx->{mid};
48 my $ibx = $ctx->{-inbox};
49 my ($smsg, $first, $next);
50 if (my $over = $ibx->over) {
52 $smsg = $over->next_by_mid($mid, \$id, \$prev) or return;
53 $first = $ibx->msg_by_smsg($smsg) or return;
54 $next = $over->next_by_mid($mid, \$id, \$prev);
55 $ctx->{more} = [ $id, $prev, $next ] if $next;
57 $first = $ibx->msg_by_mid($mid) or return;
59 my $mime = PublicInbox::MIME->new($first);
60 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
61 my $hdr = $ctx->{hdr} = $mime->header_obj;
62 $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx, 0);
64 # $next cannot be true w/o $smsg being defined:
65 $ctx->{mhref} = $next ? '../'.mid_href($smsg->{mid}).'/' : '';
66 multipart_text_as_html($mime, $ctx);
67 $ctx->{-html_tip} = (${delete $ctx->{obuf}} .= '</pre><hr>');
68 PublicInbox::WwwStream->response($ctx, 200, \&msg_page_i);
71 sub msg_page_more { # cold
72 my ($ctx, $nr, $id, $prev, $smsg) = @_;
73 my $ibx = $ctx->{-inbox};
74 my $next = $ibx->over->next_by_mid($ctx->{mid}, \$id, \$prev);
75 $ctx->{more} = [ $id, $prev, $next ] if $next;
76 $smsg = $ibx->smsg_mime($smsg) or return '';
77 $ctx->{mhref} = '../' . mid_href($smsg->{mid}) . '/';
78 my $mime = delete $smsg->{mime};
79 $ctx->{obuf} = _msg_page_prepare_obuf($mime->header_obj, $ctx, $nr);
80 multipart_text_as_html($mime, $ctx);
81 ${delete $ctx->{obuf}} .= '</pre><hr>';
84 # /$INBOX/$MESSAGE_ID/#R
88 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
90 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
93 my $ibx = $ctx->{-inbox};
94 if (my $url = $ibx->{infourl}) {
95 $url = prurl($ctx->{env}, $url);
96 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
99 my ($arg, $link, $reply_to_all) =
100 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
101 if (ref($arg) eq 'SCALAR') {
102 return '<pre id=R>'.ascii_html($$arg).'</pre>';
105 # mailto: link only works if address obfuscation is disabled
109 * If your mail client supports setting the <b>In-Reply-To</b> header
110 via mailto: links, try the <a
111 href="$link">mailto: link</a>
115 push @$arg, '/path/to/YOUR_REPLY';
116 $arg = ascii_html(join(" \\\n ", '', @$arg));
119 id=R><b>Reply instructions:</b>
121 You may reply publicly to <a
122 href=#t>this message</a> via plain-text email
123 using any one of the following methods:
125 * Save the following mbox file, import it into your mail client,
126 and $reply_to_all from there: <a
129 Avoid top-posting and favor interleaved quoting:
131 href="$p_url">$p_url</a>
133 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
134 switches of git-send-email(1):
139 href="$se_url">$se_url</a>
146 my $refs = references($hdr);
150 sub fold_addresses ($) {
151 return $_[0] if length($_[0]) <= COLS;
152 # try to fold on commas after non-word chars before $lim chars,
153 # Try to get the "," preceded by ">" or ")", but avoid folding
154 # on the comma where somebody uses "Lastname, Firstname".
155 # We also try to keep the last and penultimate addresses in
156 # the list on the same line if possible, hence the extra \z
157 # Fall back to folding on spaces at $lim + 1 chars
158 my $lim = COLS - 8; # 8 = "\t" display width
159 my $too_long = $lim + 1;
160 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
162 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
165 .{$too_long,}?)(?:\s|\z)/xgo));
168 sub _hdr_names_html ($$) {
169 my ($hdr, $field) = @_;
170 my @vals = $hdr->header($field) or return '';
171 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
175 my ($nr, $singular, $plural) = @_;
176 return "0 $plural" if $nr == 0;
177 $nr == 1 ? "$nr $singular" : "$nr $plural";
180 # human-friendly format
181 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
183 # this is already inside a <pre>
185 my ($smsg, $ctx, $more) = @_;
186 my $subj = $smsg->subject;
187 my $mid_raw = $smsg->mid;
188 my $id = id_compress($mid_raw, 1);
191 my $root_anchor = $ctx->{root_anchor} || '';
193 my $obfs_ibx = $ctx->{-obfs_ibx};
195 $subj = '(no subject)' if $subj eq '';
196 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
197 $subj = '<b>'.ascii_html($subj).'</b>';
198 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
199 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
201 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
203 my $ds = $smsg->ds; # for v1 non-Xapian/SQLite users
204 # deleting {mime} is critical to memory use,
205 # the rest of the fields saves about 400K as we iterate across 1K msgs
206 my ($mime) = delete @$smsg{qw(mime ds ts blob subject)};
208 my $hdr = $mime->header_obj;
209 my $from = _hdr_names_html($hdr, 'From');
210 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
211 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
212 my $upfx = $ctx->{-upfx};
213 my $mhref = $upfx . mid_href($mid_raw) . '/';
214 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
215 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
216 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
217 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
218 my ($tlen, $clen) = (length($to), length($cc));
220 if (($tlen + $clen) > COLS) {
221 $to_cc .= ' To: '.$to."\n" if $tlen;
222 $to_cc .= ' Cc: '.$cc."\n" if $clen;
225 $to_cc .= ' To: '.$to;
226 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
228 $to_cc .= ' Cc: '.$cc if $clen;
232 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
235 my $mapping = $ctx->{mapping};
236 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
237 my $href = $upfx . mid_href($irt) . '/';
238 my $html = ascii_html($irt);
239 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
243 # scan through all parts, looking for displayable text
244 $ctx->{mhref} = $mhref;
246 msg_iter($mime, \&add_text_body, $ctx, 1);
250 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
251 "<a\nhref=\"$mhref\">permalink</a>" .
252 " <a\nhref=\"${mhref}raw\">raw</a>" .
253 " <a\nhref=\"${mhref}#R\">reply</a>";
256 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
257 $rv .= "\t[relevance $pct%]";
260 my $nested = 'nested';
265 $flat = "<b>$flat</b>";
267 $nested = "<b>$nested</b>";
269 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
270 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
271 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
276 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
280 sub pad_link ($$;$) {
281 my ($mid, $level, $s) = @_;
283 my $id = id_compress($mid, 1);
284 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
288 my ($mid_raw, $irt, $id, $ctx) = @_;
290 my $mapping = $ctx->{mapping} or return $rv;
292 my $mid_map = $mapping->{$mid_raw};
294 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
295 my ($attr, $node, $idx, $level) = @$mid_map;
296 my $children = $node->{children};
297 my $nr_c = scalar @$children;
300 # delete saves about 200KB on a 1K message thread
301 if (my $refs = delete $node->{references}) {
302 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
304 my $irt_map = $mapping->{$$irt} if defined $$irt;
305 if (defined $irt_map) {
306 $siblings = $irt_map->[1]->{children};
307 $nr_s = scalar(@$siblings) - 1;
308 $rv .= $pad . $irt_map->[0];
310 my $prev = $siblings->[$idx - 1];
311 my $pmid = $prev->{mid};
313 my $s = ($idx - 1). ' preceding siblings ...';
314 $rv .= pad_link($pmid, $level, $s);
315 } elsif ($idx == 2) {
316 my $ppmid = $siblings->[0]->{mid};
317 $rv .= $pad . $mapping->{$ppmid}->[0];
319 $rv .= $pad . $mapping->{$pmid}->[0];
322 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
323 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
324 $attr =~ s!\n\z!</b>\n!s;
325 $attr =~ s!<a\nhref.*</a> !!s; # no point in duplicating subject
326 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
329 my $cmid = $children->[0]->{mid};
330 $rv .= $pad . $mapping->{$cmid}->[0];
332 my $s = ($nr_c - 1). ' more replies';
333 $rv .= pad_link($cmid, $level + 1, $s);
334 } elsif (my $cn = $children->[1]) {
335 $rv .= $pad . $mapping->{$cn->{mid}}->[0];
339 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
341 my $nmid = $next->{mid};
342 $rv .= $pad . $mapping->{$nmid}->[0];
343 my $nnext = $nr_s - $idx;
345 my $s = ($nnext - 1).' subsequent siblings';
346 $rv .= pad_link($nmid, $level, $s);
347 } elsif (my $nn = $siblings->[$idx + 2]) {
348 $rv .= $pad . $mapping->{$nn->{mid}}->[0];
351 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
354 # non-recursive thread walker
355 sub walk_thread ($$$) {
356 my ($rootset, $ctx, $cb) = @_;
357 my @q = map { (0, $_, -1) } @$rootset;
359 my ($level, $node, $i) = splice(@q, 0, 3);
360 defined $node or next;
361 $cb->($ctx, $level, $node, $i) or return;
364 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
368 sub pre_thread { # walk_thread callback
369 my ($ctx, $level, $node, $idx) = @_;
370 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
371 skel_dump($ctx, $level, $node);
374 sub thread_index_entry {
375 my ($ctx, $level, $smsg) = @_;
376 my ($beg, $end) = thread_adj_level($ctx, $level);
377 $beg . '<pre>' . index_entry($smsg, $ctx, 0) . '</pre>' . $end;
380 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
382 return unless exists($ctx->{skel});
383 my $q = $ctx->{-queue};
385 my $level = shift @$q;
386 my $node = shift @$q or next;
388 unshift @$q, map { ($cl, $_) } @{$node->{children}};
389 if ($ctx->{-inbox}->smsg_mime($node)) {
390 return thread_index_entry($ctx, $level, $node);
392 return ghost_index_entry($ctx, $level, $node);
395 join('', thread_adj_level($ctx, 0)) . ${delete $ctx->{skel}};
398 sub stream_thread ($$) {
399 my ($rootset, $ctx) = @_;
400 my $ibx = $ctx->{-inbox};
401 my @q = map { (0, $_) } @$rootset;
405 my $node = shift @q or next;
407 unshift @q, map { ($cl, $_) } @{$node->{children}};
408 $smsg = $ibx->smsg_mime($node) and last;
410 return missing_thread($ctx) unless $smsg;
412 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
413 $ctx->{-title_html} = ascii_html($smsg->{subject});
414 $ctx->{-html_tip} = thread_index_entry($ctx, $level, $smsg);
415 $ctx->{-queue} = \@q;
416 PublicInbox::WwwStream->response($ctx, 200, \&stream_thread_i);
419 # /$INBOX/$MESSAGE_ID/t/
422 my $mid = $ctx->{mid};
423 my $ibx = $ctx->{-inbox};
424 my ($nr, $msgs) = $ibx->over->get_thread($mid);
425 return missing_thread($ctx) if $nr == 0;
426 my $skel = '<hr><pre>';
427 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
428 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
429 $skel .= "<b\nid=t>Thread overview:</b> ";
430 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
431 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
432 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
433 $skel .= "-- links below jump to the message on this page --\n";
434 $ctx->{-upfx} = '../../';
435 $ctx->{cur_level} = 0;
436 $ctx->{skel} = \$skel;
437 $ctx->{prev_attr} = '';
438 $ctx->{prev_level} = 0;
439 $ctx->{root_anchor} = anchor_for($mid);
440 $ctx->{mapping} = {};
441 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
444 my $rootset = thread_results($ctx, $msgs);
446 # reduce hash lookups in pre_thread->skel_dump
447 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
448 walk_thread($rootset, $ctx, \&pre_thread);
451 return stream_thread($rootset, $ctx) unless $ctx->{flat};
453 # flat display: lazy load the full message from smsg
455 while (my $m = shift @$msgs) {
456 $smsg = $ibx->smsg_mime($m) and last;
458 return missing_thread($ctx) unless $smsg;
459 $ctx->{-title_html} = ascii_html($smsg->{subject});
460 $ctx->{-html_tip} = '<pre>'.index_entry($smsg, $ctx, scalar @$msgs);
461 $ctx->{msgs} = $msgs;
462 PublicInbox::WwwStream->response($ctx, 200, \&thread_html_i);
465 sub thread_html_i { # PublicInbox::WwwStream::getline callback
467 my $msgs = $ctx->{msgs} or return;
468 while (my $smsg = shift @$msgs) {
469 $ctx->{-inbox}->smsg_mime($smsg) or next;
470 return index_entry($smsg, $ctx, scalar @$msgs);
472 my ($skel) = delete @$ctx{qw(skel msgs)};
476 sub multipart_text_as_html {
477 # ($mime, $ctx) = @_; # msg_iter will do "$_[0] = undef"
479 # scan through all parts, looking for displayable text
480 msg_iter($_[0], \&add_text_body, $_[1], 1);
483 sub attach_link ($$$$;$) {
484 my ($ctx, $ct, $p, $fn, $err) = @_;
485 my ($part, $depth, @idx) = @$p;
486 my $nl = $idx[-1] > 1 ? "\n" : '';
487 my $idx = join('.', @idx);
488 my $size = bytes::length($part->body);
490 # hide attributes normally, unless we want to aid users in
491 # spotting MUA problems:
492 $ct =~ s/;.*// unless $err;
493 $ct = ascii_html($ct);
495 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
497 } elsif ($ct eq 'text/plain') {
502 my $rv = $ctx->{obuf};
503 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
506 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
509 $$rv .= "[-- Attachment #$idx: ";
510 my $ts = "Type: $ct, Size: $size bytes";
511 my $desc = $part->header('Content-Description') // $fn // '';
512 $desc = ascii_html($desc);
513 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
518 sub add_text_body { # callback for msg_iter
520 my $upfx = $ctx->{mhref};
521 my $ibx = $ctx->{-inbox};
522 # $p - from msg_iter: [ Email::MIME, depth, @idx ]
523 my ($part, $depth, @idx) = @$p;
524 my $ct = $part->content_type || 'text/plain';
525 my $fn = $part->filename;
526 my ($s, $err) = msg_part_text($part, $ct);
527 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
529 # makes no difference to browsers, and don't screw up filename
530 # link generation in diffs with the extra '%0D'
533 # will be escaped to `•' in HTML
534 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
536 # always support diff-highlighting, but we can't linkify hunk
537 # headers for solver unless some coderepo are configured:
539 if ($s =~ /^(?:diff|---|\+{3}) /ms) {
540 # diffstat anchors do not link across attachments or messages:
541 $idx[0] = $upfx . $idx[0] if $upfx ne '';
542 $ctx->{-apfx} = join('/', @idx);
543 $ctx->{-anchors} = {}; # attr => filename
545 delete $ctx->{-long_path};
547 if ($ibx->{-repo_objs}) {
548 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
550 $spfx =~ s!/([^/]*)/\z!/!;
552 my $n_slash = $upfx =~ tr!/!/!;
555 } elsif ($n_slash == 1) {
557 } else { # nslash == 2
562 $ctx->{-spfx} = $spfx;
565 # some editors don't put trailing newlines at the end:
566 $s .= "\n" unless $s =~ /\n\z/s;
568 # split off quoted and unquoted blocks:
569 my @sections = PublicInbox::MsgIter::split_quotes($s);
570 undef $s; # free memory
571 my $rv = $ctx->{obuf};
572 if (defined($fn) || $depth > 0 || $err) {
573 # badly-encoded message with $err? tell the world about it!
574 attach_link($ctx, $ct, $p, $fn, $err);
577 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
578 foreach my $cur (@sections) {
580 # we use a <span> here to allow users to specify
581 # their own color for quoted text
582 $$rv .= qq(<span\nclass="q">);
583 $$rv .= $l->to_html($cur);
586 flush_diff($ctx, \$cur);
589 $$rv .= $l->to_html($cur);
591 undef $cur; # free memory
595 sub _msg_page_prepare_obuf {
596 my ($hdr, $ctx, $nr) = @_;
597 my $over = $ctx->{-inbox}->over;
598 my $obfs_ibx = $ctx->{-obfs_ibx};
600 my $mids = mids_for_index($hdr);
604 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
606 $rv .= "<pre\nid=b>"; # anchor for body start
611 $ctx->{-upfx} = '../';
613 my @title; # (Subject[0], From[0])
614 for my $v ($hdr->header('From')) {
615 my @n = PublicInbox::Address::names($v);
617 $title[1] //= ascii_html(join(', ', @n));
619 obfuscate_addrs($obfs_ibx, $v);
620 obfuscate_addrs($obfs_ibx, $title[1]);
622 $rv .= "From: $v\n" if $v ne '';
624 foreach my $h (qw(To Cc)) {
625 for my $v ($hdr->header($h)) {
628 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
629 $rv .= "$h: $v\n" if $v ne '';
632 my @subj = $hdr->header('Subject');
634 my $v = ascii_html(shift @subj);
635 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
637 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
639 for $v (@subj) { # multi-Subject message :<
641 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
642 $rv .= "Subject: $v\n";
644 } else { # dummy anchor for thread skeleton at bottom of page
645 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
646 $title[0] = '(no subject)';
648 for my $v ($hdr->header('Date')) {
650 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
653 $ctx->{-title_html} = join(' - ', @title);
654 if (scalar(@$mids) == 1) { # common case
655 my $mhtml = ascii_html($mids->[0]);
656 $rv .= "Message-ID: <$mhtml> ";
657 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
659 # X-Alt-Message-ID can happen if a message is injected from
660 # public-inbox-nntpd because of multiple Message-ID headers.
661 my $lnk = PublicInbox::Linkify->new;
663 for my $h (qw(Message-ID X-Alt-Message-ID)) {
664 $s .= "$h: $_\n" for ($hdr->header_raw($h));
666 $lnk->linkify_mids('..', \$s, 1);
669 $rv .= _parent_headers($hdr, $over);
675 qq(expand[<a\nhref="T/#u">flat</a>) .
676 qq(|<a\nhref="t/#u">nested</a>] ) .
677 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
678 qq(<a\nhref="t.atom">Atom feed</a>);
681 sub thread_skel ($$$) {
682 my ($skel, $ctx, $hdr) = @_;
683 my $mid = mids($hdr)->[0];
684 my $ibx = $ctx->{-inbox};
685 my ($nr, $msgs) = $ibx->over->get_thread($mid);
686 my $parent = in_reply_to($hdr);
687 $$skel .= "\n<b>Thread overview: </b>";
689 if (defined $parent) {
690 $$skel .= SKEL_EXPAND."\n ";
691 $$skel .= ghost_parent('../', $parent) . "\n";
693 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
695 $ctx->{next_msg} = undef;
696 $ctx->{parent_msg} = $parent;
701 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
703 # nb: mutt only shows the first Subject in the index pane
704 # when multiple Subject: headers are present, so we follow suit:
705 my $subj = $hdr->header('Subject') // '';
706 $subj = '(no subject)' if $subj eq '';
707 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
709 $ctx->{prev_attr} = '';
710 $ctx->{prev_level} = 0;
711 $ctx->{skel} = $skel;
713 # reduce hash lookups in skel_dump
714 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
715 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
717 $ctx->{parent_msg} = $parent;
720 sub _parent_headers {
721 my ($hdr, $over) = @_;
723 my @irt = $hdr->header_raw('In-Reply-To');
726 my $lnk = PublicInbox::Linkify->new;
727 $rv .= "In-Reply-To: $_\n" for @irt;
728 $lnk->linkify_mids('..', \$rv);
730 $refs = references($hdr);
731 my $irt = pop @$refs;
733 my $html = ascii_html($irt);
734 my $href = mid_href($irt);
735 $rv .= "In-Reply-To: <";
736 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
740 # do not display References: if search is present,
741 # we show the thread skeleton at the bottom, instead.
744 $refs //= references($hdr);
746 @$refs = map { linkify_ref_no_over($_) } @$refs;
747 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
752 # returns a string buffer via ->getline
754 my ($ctx, $hdr) = @_;
755 my $ibx = $ctx->{-inbox};
757 my $skel = " <a\nhref=\"$upfx\">index</a>";
761 thread_skel(\$skel, $ctx, $hdr);
766 if (my $n = $ctx->{next_msg}) {
768 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
771 my $par = $ctx->{parent_msg};
776 if (my $p = $ctx->{prev_msg}) {
777 $prev = mid_href($p);
778 if ($p && $par && $p eq $par) {
779 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
780 'rel=prev>prev parent</a>';
783 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
785 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
787 } elsif ($u) { # unlikely
788 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
790 $rv .= "$next $prev$parent ";
792 $rv .= qq(<a\nhref="#R">reply</a>);
795 $rv .= msg_reply($ctx, $hdr);
798 sub linkify_ref_no_over {
800 my $href = mid_href($mid);
801 my $html = ascii_html($mid);
802 "<<a\nhref=\"../$href/\">$html</a>>";
807 'm' . id_compress($msgid, 1);
811 my ($upfx, $mid) = @_;
813 my $href = mid_href($mid);
814 my $html = ascii_html($mid);
815 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
820 $level ? INDENT x ($level - 1) : '';
824 my ($ctx, $level, $node, $idx) = @_;
825 ++$ctx->{root_idx} if $level == 0;
826 if ($node->{mid} eq $ctx->{mid}) {
827 $ctx->{found_mid_at} = $ctx->{root_idx};
833 sub strict_loose_note ($) {
836 " -- strict thread matches above, loose matches on Subject: below --\n";
838 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
840 " -- use mbox.gz link to download all $nr messages --\n";
846 my ($ctx, $msgs) = @_;
847 require PublicInbox::SearchThread;
848 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
850 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
851 # and preserved in the future. This bug is hidden by `sid' matches
852 # in get_thread, so we never noticed it until now. And even when
853 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
854 # bump since reindexing is expensive and users may not do it
856 # loose threading could've returned too many results,
857 # put the root the message we care about at the top:
858 my $mid = $ctx->{mid};
859 if (defined($mid) && scalar(@$rootset) > 1) {
860 $ctx->{root_idx} = -1;
861 my $nr = scalar @$msgs;
862 walk_thread($rootset, $ctx, \&find_mid_root);
863 my $idx = $ctx->{found_mid_at};
864 if (defined($idx) && $idx != 0) {
865 my $tip = splice(@$rootset, $idx, 1);
866 @$rootset = reverse @$rootset;
867 unshift @$rootset, $tip;
868 $ctx->{sl_note} = strict_loose_note($nr);
876 require PublicInbox::ExtMsg;
877 PublicInbox::ExtMsg::ext_msg($ctx);
881 my ($prev_subj, $subj, $val) = @_;
883 my $omit; # '"' denotes identical text omitted
884 my (@prev_pop, @curr_pop);
885 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
886 push(@prev_pop, pop(@$prev_subj));
887 push(@curr_pop, pop(@$subj));
890 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
891 if (scalar(@curr_pop) == 1) {
893 push @$prev_subj, @prev_pop;
894 push @$subj, @curr_pop;
899 sub skel_dump { # walk_thread callback
900 my ($ctx, $level, $smsg) = @_;
901 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
903 my $skel = $ctx->{skel};
904 my $cur = $ctx->{cur};
905 my $mid = $smsg->{mid};
907 if ($level == 0 && $ctx->{skel_dump_roots}++) {
908 $$skel .= delete($ctx->{sl_note}) || '';
911 my $f = ascii_html($smsg->{from_name});
912 my $obfs_ibx = $ctx->{-obfs_ibx};
913 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
915 my $d = fmt_ts($smsg->{ds});
916 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
917 if (exists $ctx->{searchview}) {
918 if (defined(my $pct = $smsg->{pct})) {
919 $d .= (sprintf(' % 2u', $pct) . '%');
925 $d .= ' ' . indent_for($level) . th_pfx($level);
927 $ctx->{first_level} ||= $level;
929 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
930 $ctx->{prev_attr} = $attr;
932 $ctx->{prev_level} = $level;
937 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
938 "$attr [this message]</a></b>\n";
941 $ctx->{prev_msg} = $mid;
944 $ctx->{next_msg} ||= $mid;
947 # Subject is never undef, this mail was loaded from
948 # our Xapian which would've resulted in '' if it were
949 # really missing (and Filter rejects empty subjects)
950 my @subj = split(/ /, subject_normalized($smsg->{subject}));
951 # remove common suffixes from the subject if it matches the previous,
952 # so we do not show redundant text at the end.
953 my $prev_subj = $ctx->{prev_subj} || [];
954 $ctx->{prev_subj} = [ @subj ];
955 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
958 my $subj = join(' ', @subj);
959 $subj = ascii_html($subj);
960 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
961 $end = "$subj</a> $omit$f\n"
967 my $mapping = $unmatched ? undef : $ctx->{mapping};
969 my $map = $mapping->{$mid};
970 $id = id_compress($mid, 1);
972 $map->[0] = "$d<a\nhref=\"$m\">$end";
975 $m = $ctx->{-upfx}.mid_href($mid).'/';
977 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
982 my ($ctx, $level, $node) = @_;
984 my $mid = $node->{mid};
985 my $d = ' [not found] ';
986 $d .= ' ' if exists $ctx->{searchview};
987 $d .= indent_for($level) . th_pfx($level);
988 my $upfx = $ctx->{-upfx};
989 my $href = $upfx . mid_href($mid) . '/';
990 my $html = ascii_html($mid);
992 my $mapping = $ctx->{mapping};
993 my $map = $mapping->{$mid} if $mapping;
995 my $id = id_compress($mid, 1);
996 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
997 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
999 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1001 ${$ctx->{skel}} .= $d;
1007 (eval { $a->topmost->{ds} } || 0) <=>
1008 (eval { $b->topmost->{ds} } || 0)
1012 # accumulate recent topics if search is supported
1013 # returns 200 if done, 404 if not
1014 sub acc_topic { # walk_thread callback
1015 my ($ctx, $level, $smsg) = @_;
1016 my $mid = $smsg->{mid};
1017 my $has_blob = $smsg->{blob} // do {
1018 if (my $by_mid = $ctx->{-inbox}->smsg_by_mid($mid)) {
1019 %$smsg = (%$smsg, %$by_mid);
1024 my $subj = subject_normalized($smsg->{subject});
1025 $subj = '(no subject)' if $subj eq '';
1026 my $ds = $smsg->{ds};
1027 if ($level == 0) { # new, top-level topic
1028 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1029 $ctx->{-cur_topic} = $topic;
1030 push @{$ctx->{order}}, $topic;
1034 # continue existing topic
1035 my $topic = $ctx->{-cur_topic}; # should never be undef
1036 $topic->[0] = $ds if $ds > $topic->[0];
1037 $topic->[1]++; # bump N+ message counter
1038 my $seen = $topic->[2];
1039 if (scalar(@$topic) == 3) { # parent was a ghost
1040 push @$topic, $subj;
1041 } elsif (!defined($seen->{$subj})) {
1042 push @$topic, $level, $subj; # @extra messages
1044 $seen->{$subj} = $mid; # latest for subject
1045 } else { # ghost message
1046 return 1 if $level != 0; # ignore child ghosts
1047 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1048 push @{$ctx->{order}}, $topic;
1055 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1057 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1062 my $ibx = $ctx->{-inbox};
1063 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1065 # sort by recency, this allows new posts to "bump" old topics...
1066 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1067 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1069 next unless defined $top_subj; # ghost topic
1070 my $mid = delete $seen->{$top_subj};
1071 my $href = mid_href($mid);
1072 my $prev_subj = [ split(/ /, $top_subj) ];
1073 $top_subj = ascii_html($top_subj);
1076 # $n isn't the total number of posts on the topic,
1077 # just the number of posts in the current results window
1081 $anchor = '#u'; # top of only message
1083 $n = " ($n+ messages)";
1084 $anchor = '#t'; # thread skeleton
1087 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1088 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1089 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1090 " $ds UTC $n - $mbox / $atom\n";
1091 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1092 my $level = $extra[$i];
1093 my $subj = $extra[$i + 1]; # already normalized
1094 $mid = delete $seen->{$subj};
1095 my @subj = split(/ /, $subj);
1096 my @next_prev = @subj; # full copy
1097 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1098 $prev_subj = \@next_prev;
1099 $subj = ascii_html($subj);
1100 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1101 $href = mid_href($mid);
1102 $s .= indent_for($level) . TCHILD;
1103 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1107 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1111 # only for the t= query parameter passed to overview DB
1112 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1115 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1116 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1119 sub pagination_footer ($$) {
1120 my ($ctx, $latest) = @_;
1121 delete $ctx->{qp} or return;
1122 my $next = $ctx->{next_page} || '';
1123 my $prev = $ctx->{prev_page} || '';
1125 $next = $next ? "$next " : ' ';
1126 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1128 "<hr><pre>page: $next$prev</pre>";
1131 sub index_nav { # callback for WwwStream
1132 my (undef, $ctx) = @_;
1133 pagination_footer($ctx, '.')
1136 sub paginate_recent ($$) {
1137 my ($ctx, $lim) = @_;
1138 my $t = $ctx->{qp}->{t} || '';
1139 my $opts = { limit => $lim };
1140 my ($after, $before);
1142 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1143 # if only $after exists "YYYYMMDD.." because "." could be skipped
1144 # if interpreted as an end-of-sentence
1145 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1146 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1148 my $ibx = $ctx->{-inbox};
1149 my $msgs = $ibx->recent($opts, $after, $before);
1150 my $nr = scalar @$msgs;
1151 if ($nr < $lim && defined($after)) {
1152 $after = $before = undef;
1153 $msgs = $ibx->recent($opts);
1154 $nr = scalar @$msgs;
1156 my $more = $nr == $lim;
1157 my ($newest, $oldest);
1159 $newest = $msgs->[0]->{ts};
1160 $oldest = $msgs->[-1]->{ts};
1161 # if we only had $after, our SQL query in ->recent ordered
1162 if ($newest < $oldest) {
1163 ($oldest, $newest) = ($newest, $oldest);
1164 $more = 0 if defined($after) && $after < $oldest;
1167 if (defined($oldest) && $more) {
1168 my $s = ts2str($oldest);
1169 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1171 if (defined($newest) && (defined($before) || defined($after))) {
1172 my $s = ts2str($newest);
1173 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1180 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1182 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1184 PublicInbox::WwwStream->response($ctx, dump_topics($ctx), \&index_nav);
1187 sub thread_adj_level {
1188 my ($ctx, $level) = @_;
1190 my $max = $ctx->{cur_level};
1192 return ('', '') if $max == 0; # flat output
1194 # reset existing lists
1195 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1196 $ctx->{cur_level} = 0;
1198 } elsif ($level == $max) { # continue existing list
1200 } elsif ($level < $max) {
1201 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1202 $ctx->{cur_level} = $level;
1203 ("$beg<li>", '</li>');
1204 } else { # ($level > $max) # start a new level
1205 $ctx->{cur_level} = $level;
1206 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1211 sub ghost_index_entry {
1212 my ($ctx, $level, $node) = @_;
1213 my ($beg, $end) = thread_adj_level($ctx, $level);
1214 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid})