1 # Copyright (C) 2014-2019 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);
12 use PublicInbox::Linkify;
13 use PublicInbox::MID qw/id_compress mid_escape mids mids_for_index references/;
14 use PublicInbox::MsgIter;
15 use PublicInbox::Address;
16 use PublicInbox::WwwStream;
17 use PublicInbox::Reply;
18 use PublicInbox::ViewDiff qw(flush_diff);
19 use POSIX qw(strftime);
20 use Time::Local qw(timegm);
21 use PublicInbox::SearchMsg qw(subject_normalized);
22 use constant COLS => 72;
23 use constant INDENT => ' ';
24 use constant TCHILD => '` ';
25 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
29 my $more = $ctx->{more};
31 # $more cannot be true w/o $smsg being defined:
32 my $upfx = $more ? '../'.mid_escape($ctx->{smsg}->mid).'/' : '';
34 multipart_text_as_html(delete $ctx->{mime}, $upfx,
36 } elsif ($more && @$more) {
38 msg_html_more($ctx, $more, $nr);
39 } elsif ($nr == $ctx->{end_nr}) {
40 # fake an EOF if generating the footer fails;
41 # we want to at least show the message if something
44 my $hdr = delete($ctx->{hdr});
45 '<pre>' . html_footer($hdr, 1, $ctx) .
46 '</pre>' . msg_reply($ctx, $hdr)
53 # public functions: (unstable)
56 my ($ctx, $mime, $more, $smsg) = @_;
57 my $ibx = $ctx->{-inbox};
58 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
59 my $hdr = $ctx->{hdr} = $mime->header_obj;
60 $ctx->{tip} = _msg_html_prepare($hdr, $ctx, $more, 0);
65 PublicInbox::WwwStream->response($ctx, 200, \&msg_html_i);
70 my $mid = $ctx->{mid};
71 my $ibx = $ctx->{-inbox};
74 if (my $over = $ibx->over) {
76 $smsg = $over->next_by_mid($mid, \$id, \$prev);
77 $first = $ibx->msg_by_smsg($smsg) if $smsg;
79 my $next = $over->next_by_mid($mid, \$id, \$prev);
80 $more = [ $id, $prev, $next ] if $next;
84 $first = $ibx->msg_by_mid($mid) or return;
86 msg_html($ctx, PublicInbox::MIME->new($first), $more, $smsg);
90 my ($ctx, $more, $nr) = @_;
92 my ($id, $prev, $smsg) = @$more;
93 my $mid = $ctx->{mid};
94 my $ibx = $ctx->{-inbox};
95 $smsg = $ibx->smsg_mime($smsg);
96 my $next = $ibx->over->next_by_mid($mid, \$id, \$prev);
97 @$more = $next ? ($id, $prev, $next) : ();
99 my $upfx = '../' . mid_escape($smsg->mid) . '/';
100 my $mime = delete $smsg->{mime};
101 _msg_html_prepare($mime->header_obj, $ctx, $more, $nr) .
102 multipart_text_as_html($mime, $upfx, $ctx) .
109 warn "Error lookup up additional messages: $@\n";
110 $str = '<pre>Error looking up additional messages</pre>';
115 # /$INBOX/$MESSAGE_ID/#R
117 my ($ctx, $hdr) = @_;
119 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
121 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
124 my $ibx = $ctx->{-inbox};
125 if (my $url = $ibx->{infourl}) {
126 $url = prurl($ctx->{env}, $url);
127 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
130 my ($arg, $link, $reply_to_all) =
131 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
132 if (ref($arg) eq 'SCALAR') {
133 return '<pre id=R>'.ascii_html($$arg).'</pre>';
136 # mailto: link only works if address obfuscation is disabled
140 * If your mail client supports setting the <b>In-Reply-To</b> header
141 via mailto: links, try the <a
142 href="$link">mailto: link</a>
146 push @$arg, '/path/to/YOUR_REPLY';
147 $arg = ascii_html(join(" \\\n ", '', @$arg));
150 id=R><b>Reply instructions:</b>
152 You may reply publically to <a
153 href=#t>this message</a> via plain-text email
154 using any one of the following methods:
156 * Save the following mbox file, import it into your mail client,
157 and $reply_to_all from there: <a
160 Avoid top-posting and favor interleaved quoting:
162 href="$p_url">$p_url</a>
164 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
165 switches of git-send-email(1):
170 href="$se_url">$se_url</a>
177 my $refs = references($hdr);
181 sub fold_addresses ($) {
182 return $_[0] if length($_[0]) <= COLS;
183 # try to fold on commas after non-word chars before $lim chars,
184 # Try to get the "," preceeded by ">" or ")", but avoid folding
185 # on the comma where somebody uses "Lastname, Firstname".
186 # We also try to keep the last and penultimate addresses in
187 # the list on the same line if possible, hence the extra \z
188 # Fall back to folding on spaces at $lim + 1 chars
189 my $lim = COLS - 8; # 8 = "\t" display width
190 my $too_long = $lim + 1;
191 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
193 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
196 .{$too_long,}?)(?:\s|\z)/xgo));
199 sub _hdr_names_html ($$) {
200 my ($hdr, $field) = @_;
201 my @vals = $hdr->header($field) or return '';
202 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
206 my ($nr, $singular, $plural) = @_;
207 return "0 $plural" if $nr == 0;
208 $nr == 1 ? "$nr $singular" : "$nr $plural";
211 # human-friendly format
212 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
214 # this is already inside a <pre>
216 my ($smsg, $ctx, $more) = @_;
217 my $subj = $smsg->subject;
218 my $mid_raw = $smsg->mid;
219 my $id = id_compress($mid_raw, 1);
222 my $root_anchor = $ctx->{root_anchor} || '';
224 my $obfs_ibx = $ctx->{-obfs_ibx};
226 $subj = '(no subject)' if $subj eq '';
227 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
228 $subj = '<b>'.ascii_html($subj).'</b>';
229 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
230 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
232 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
234 my $ds = $smsg->ds; # for v1 non-Xapian/SQLite users
235 # deleting {mime} is critical to memory use,
236 # the rest of the fields saves about 400K as we iterate across 1K msgs
237 my ($mime) = delete @$smsg{qw(mime ds ts blob subject)};
239 my $hdr = $mime->header_obj;
240 my $from = _hdr_names_html($hdr, 'From');
241 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
242 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
243 my $upfx = $ctx->{-upfx};
244 my $mhref = $upfx . mid_escape($mid_raw) . '/';
245 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
246 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
247 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
248 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
249 my ($tlen, $clen) = (length($to), length($cc));
251 if (($tlen + $clen) > COLS) {
252 $to_cc .= ' To: '.$to."\n" if $tlen;
253 $to_cc .= ' Cc: '.$cc."\n" if $clen;
256 $to_cc .= ' To: '.$to;
257 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
259 $to_cc .= ' Cc: '.$cc if $clen;
263 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
266 my $mapping = $ctx->{mapping};
267 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
268 my $mirt = PublicInbox::Hval->new_msgid($irt);
269 my $href = $upfx . $mirt->{href}. '/';
270 my $html = $mirt->as_html;
271 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
275 # scan through all parts, looking for displayable text
276 $ctx->{mhref} = $mhref;
278 msg_iter($mime, \&add_text_body, $ctx);
282 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
283 "<a\nhref=\"$mhref\">permalink</a>" .
284 " <a\nhref=\"${mhref}raw\">raw</a>" .
285 " <a\nhref=\"${mhref}#R\">reply</a>";
288 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
289 $rv .= "\t[relevance $pct%]";
292 my $nested = 'nested';
297 $flat = "<b>$flat</b>";
299 $nested = "<b>$nested</b>";
301 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
302 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
303 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
308 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
312 sub pad_link ($$;$) {
313 my ($mid, $level, $s) = @_;
315 my $id = id_compress($mid, 1);
316 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
320 my ($mid_raw, $irt, $id, $ctx) = @_;
322 my $mapping = $ctx->{mapping} or return $rv;
324 my $mid_map = $mapping->{$mid_raw};
326 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
327 my ($attr, $node, $idx, $level) = @$mid_map;
328 my $children = $node->{children};
329 my $nr_c = scalar @$children;
332 if (my $smsg = $node->{smsg}) {
333 # delete saves about 200KB on a 1K message thread
334 if (my $refs = delete $smsg->{references}) {
335 ($$irt) = ($refs =~ m/<([^>]+)>\z/);
338 my $irt_map = $mapping->{$$irt} if defined $$irt;
339 if (defined $irt_map) {
340 $siblings = $irt_map->[1]->{children};
341 $nr_s = scalar(@$siblings) - 1;
342 $rv .= $pad . $irt_map->[0];
344 my $prev = $siblings->[$idx - 1];
345 my $pmid = $prev->{id};
347 my $s = ($idx - 1). ' preceding siblings ...';
348 $rv .= pad_link($pmid, $level, $s);
349 } elsif ($idx == 2) {
350 my $ppmid = $siblings->[0]->{id};
351 $rv .= $pad . $mapping->{$ppmid}->[0];
353 $rv .= $pad . $mapping->{$pmid}->[0];
356 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
357 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
358 $attr =~ s!\n\z!</b>\n!s;
359 $attr =~ s!<a\nhref.*</a> !!s; # no point in duplicating subject
360 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
363 my $cmid = $children->[0]->{id};
364 $rv .= $pad . $mapping->{$cmid}->[0];
366 my $s = ($nr_c - 1). ' more replies';
367 $rv .= pad_link($cmid, $level + 1, $s);
368 } elsif (my $cn = $children->[1]) {
369 $rv .= $pad . $mapping->{$cn->{id}}->[0];
373 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
375 my $nmid = $next->{id};
376 $rv .= $pad . $mapping->{$nmid}->[0];
377 my $nnext = $nr_s - $idx;
379 my $s = ($nnext - 1).' subsequent siblings';
380 $rv .= pad_link($nmid, $level, $s);
381 } elsif (my $nn = $siblings->[$idx + 2]) {
382 $rv .= $pad . $mapping->{$nn->{id}}->[0];
385 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
389 my ($rootset, $ctx, $cb) = @_;
390 my @q = map { (0, $_, -1) } @$rootset;
392 my ($level, $node, $i) = splice(@q, 0, 3);
393 defined $node or next;
394 $cb->($ctx, $level, $node, $i) or return;
397 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
402 my ($ctx, $level, $node, $idx) = @_;
403 $ctx->{mapping}->{$node->{id}} = [ '', $node, $idx, $level ];
404 skel_dump($ctx, $level, $node);
407 sub thread_index_entry {
408 my ($ctx, $level, $smsg) = @_;
409 my ($beg, $end) = thread_adj_level($ctx, $level);
410 $beg . '<pre>' . index_entry($smsg, $ctx, 0) . '</pre>' . $end;
413 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
415 return unless exists($ctx->{dst});
416 my $q = $ctx->{-queue};
418 my $level = shift @$q;
419 my $node = shift @$q or next;
421 unshift @$q, map { ($cl, $_) } @{$node->{children}};
422 if (my $smsg = $ctx->{-inbox}->smsg_mime($node->{smsg})) {
423 return thread_index_entry($ctx, $level, $smsg);
425 return ghost_index_entry($ctx, $level, $node);
428 join('', thread_adj_level($ctx, 0)) . ${delete $ctx->{dst}}; # skel
431 sub stream_thread ($$) {
432 my ($rootset, $ctx) = @_;
433 my $ibx = $ctx->{-inbox};
434 my @q = map { (0, $_) } @$rootset;
438 my $node = shift @q or next;
440 unshift @q, map { ($cl, $_) } @{$node->{children}};
441 $smsg = $ibx->smsg_mime($node->{smsg}) and last;
443 return missing_thread($ctx) unless $smsg;
445 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
446 $ctx->{-title_html} = ascii_html($smsg->subject);
447 $ctx->{-html_tip} = thread_index_entry($ctx, $level, $smsg);
448 $ctx->{-queue} = \@q;
449 PublicInbox::WwwStream->response($ctx, 200, \&stream_thread_i);
454 my $mid = $ctx->{mid};
455 my $ibx = $ctx->{-inbox};
456 my ($nr, $msgs) = $ibx->over->get_thread($mid);
457 return missing_thread($ctx) if $nr == 0;
458 my $skel = '<hr><pre>';
459 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
460 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
461 $skel .= "<b\nid=t>Thread overview:</b> ";
462 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
463 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
464 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
465 $skel .= "-- links below jump to the message on this page --\n";
466 $ctx->{-upfx} = '../../';
467 $ctx->{cur_level} = 0;
468 $ctx->{dst} = \$skel;
469 $ctx->{prev_attr} = '';
470 $ctx->{prev_level} = 0;
471 $ctx->{root_anchor} = anchor_for($mid);
472 $ctx->{mapping} = {};
473 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
476 my $rootset = thread_results($ctx, $msgs);
478 # reduce hash lookups in pre_thread->skel_dump
479 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
480 walk_thread($rootset, $ctx, *pre_thread);
483 return stream_thread($rootset, $ctx) unless $ctx->{flat};
485 # flat display: lazy load the full message from smsg
487 while (my $m = shift @$msgs) {
488 $smsg = $ibx->smsg_mime($m) and last;
490 return missing_thread($ctx) unless $smsg;
491 $ctx->{-title_html} = ascii_html($smsg->subject);
492 $ctx->{-html_tip} = '<pre>'.index_entry($smsg, $ctx, scalar @$msgs);
493 $ctx->{msgs} = $msgs;
494 PublicInbox::WwwStream->response($ctx, 200, \&thread_html_i);
497 sub thread_html_i { # PublicInbox::WwwStream::getline callback
499 my $msgs = $ctx->{msgs} or return;
500 while (my $smsg = shift @$msgs) {
501 $ctx->{-inbox}->smsg_mime($smsg) or next;
502 return index_entry($smsg, $ctx, scalar @$msgs);
504 my ($skel) = delete @$ctx{qw(dst msgs)};
508 sub multipart_text_as_html {
509 my ($mime, $mhref, $ctx) = @_;
510 $ctx->{mhref} = $mhref;
511 $ctx->{rv} = \(my $rv = '');
513 # scan through all parts, looking for displayable text
514 msg_iter($mime, \&add_text_body, $ctx);
515 ${delete $ctx->{rv}};
519 my ($s, $l, $quot) = @_;
521 # show everything in the full version with anchor from
522 # short version (see above)
523 my $rv = $l->linkify_1($$quot);
525 # we use a <span> here to allow users to specify their own
526 # color for quoted text
527 $rv = $l->linkify_2(ascii_html($rv));
529 $$s .= qq(<span\nclass="q">) . $rv . '</span>'
532 sub attach_link ($$$$;$) {
533 my ($ctx, $ct, $p, $fn, $err) = @_;
534 my ($part, $depth, @idx) = @$p;
535 my $nl = $idx[-1] > 1 ? "\n" : '';
536 my $idx = join('.', @idx);
537 my $size = bytes::length($part->body);
539 # hide attributes normally, unless we want to aid users in
540 # spotting MUA problems:
541 $ct =~ s/;.*// unless $err;
542 $ct = ascii_html($ct);
543 my $desc = $part->header('Content-Description');
544 $desc = $fn unless defined $desc;
545 $desc = '' unless defined $desc;
547 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
549 } elsif ($ct eq 'text/plain') {
555 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
557 $$rv .= "[-- Warning: decoded text below may be mangled --]\n";
559 $$rv .= "[-- Attachment #$idx: ";
560 my $ts = "Type: $ct, Size: $size bytes";
561 $desc = ascii_html($desc);
562 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
567 sub add_text_body { # callback for msg_iter
569 my $upfx = $ctx->{mhref};
570 my $ibx = $ctx->{-inbox};
571 # $p - from msg_iter: [ Email::MIME, depth, @idx ]
572 my ($part, $depth, @idx) = @$p;
573 my $ct = $part->content_type || 'text/plain';
574 my $fn = $part->filename;
575 my ($s, $err) = msg_part_text($part, $ct);
576 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
578 # makes no difference to browsers, and don't screw up filename
579 # link generation in diffs with the extra '%0D'
582 # always support diff-highlighting, but we can't linkify hunk
583 # headers for solver unless some coderepo are configured:
585 if ($s =~ /^(?:diff|---|\+{3}) /ms) {
586 # diffstat anchors do not link across attachments or messages:
587 $idx[0] = $upfx . $idx[0] if $upfx ne '';
588 $ctx->{-apfx} = join('/', @idx);
589 $ctx->{-anchors} = {}; # attr => filename
590 $ctx->{-diff} = $diff = [];
591 delete $ctx->{-long_path};
593 if ($ibx->{-repo_objs}) {
594 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
596 $spfx =~ s!/([^/]*)/\z!/!;
598 my $n_slash = $upfx =~ tr!/!/!;
601 } elsif ($n_slash == 1) {
603 } else { # nslash == 2
608 $ctx->{-spfx} = $spfx;
611 # some editors don't put trailing newlines at the end:
612 $s .= "\n" unless $s =~ /\n\z/s;
614 # split off quoted and unquoted blocks:
615 my @sections = split(/((?:^>[^\n]*\n)+)/sm, $s);
618 if (defined($fn) || $depth > 0 || $err) {
619 # badly-encoded message with $err? tell the world about it!
620 attach_link($ctx, $ct, $p, $fn, $err);
623 my $l = PublicInbox::Linkify->new;
624 foreach my $cur (@sections) {
626 flush_quote($rv, $l, \$cur);
628 @$diff = split(/^/m, $cur);
630 flush_diff($rv, $ctx, $l);
634 $$rv .= $l->linkify_2(ascii_html($cur));
639 obfuscate_addrs($ibx, $$rv) if $ibx->{obfuscate};
642 sub _msg_html_prepare {
643 my ($hdr, $ctx, $more, $nr) = @_;
645 my $over = $ctx->{-inbox}->over;
646 my $obfs_ibx = $ctx->{-obfs_ibx};
648 my $mids = mids_for_index($hdr);
652 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
654 $rv .= "<pre\nid=b>"; # anchor for body start
659 $ctx->{-upfx} = '../';
661 my @title; # (Subject[0], From[0])
662 for my $v ($hdr->header('From')) {
663 $v = PublicInbox::Hval->new($v);
664 my @n = PublicInbox::Address::names($v->raw);
665 $title[1] //= ascii_html(join(', ', @n));
668 obfuscate_addrs($obfs_ibx, $v);
669 obfuscate_addrs($obfs_ibx, $title[1]);
671 $rv .= "From: $v\n" if $v ne '';
673 foreach my $h (qw(To Cc)) {
674 for my $v ($hdr->header($h)) {
677 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
678 $rv .= "$h: $v\n" if $v ne '';
681 my @subj = $hdr->header('Subject');
685 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
688 $rv .= qq(<a\nhref="#r"\nid=t>$v</a>\n);
694 } else { # dummy anchor for thread skeleton at bottom of page
695 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
696 $title[0] = '(no subject)';
698 for my $v ($hdr->header('Date')) {
700 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
703 $ctx->{-title_html} = join(' - ', @title);
704 if (scalar(@$mids) == 1) { # common case
705 my $mid = PublicInbox::Hval->new_msgid($mids->[0]);
706 my $mhtml = $mid->as_html;
707 $rv .= "Message-ID: <$mhtml> ";
708 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
710 # X-Alt-Message-ID can happen if a message is injected from
711 # public-inbox-nntpd because of multiple Message-ID headers.
712 my $lnk = PublicInbox::Linkify->new;
714 for my $h (qw(Message-ID X-Alt-Message-ID)) {
715 $s .= "$h: $_\n" for ($hdr->header_raw($h));
717 $lnk->linkify_mids('..', \$s, 1);
720 $rv .= _parent_headers($hdr, $over);
725 my ($dst, $ctx, $hdr, $tpfx) = @_;
726 my $mid = mids($hdr)->[0];
727 my $ibx = $ctx->{-inbox};
728 my ($nr, $msgs) = $ibx->over->get_thread($mid);
729 my $expand = qq(expand[<a\nhref="${tpfx}T/#u">flat</a>) .
730 qq(|<a\nhref="${tpfx}t/#u">nested</a>] ) .
731 qq(<a\nhref="${tpfx}t.mbox.gz">mbox.gz</a> ) .
732 qq(<a\nhref="${tpfx}t.atom">Atom feed</a>);
734 my $parent = in_reply_to($hdr);
735 $$dst .= "\n<b>Thread overview: </b>";
737 if (defined $parent) {
738 $$dst .= "$expand\n ";
739 $$dst .= ghost_parent("$tpfx../", $parent) . "\n";
741 $$dst .= "[no followups] $expand\n";
743 $ctx->{next_msg} = undef;
744 $ctx->{parent_msg} = $parent;
748 $$dst .= "$nr+ messages / $expand";
749 $$dst .= qq! <a\nhref="#b">top</a>\n!;
751 # nb: mutt only shows the first Subject in the index pane
752 # when multiple Subject: headers are present, so we follow suit:
753 my $subj = $hdr->header('Subject') // '';
754 $subj = '(no subject)' if $subj eq '';
755 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
757 $ctx->{prev_attr} = '';
758 $ctx->{prev_level} = 0;
761 # reduce hash lookups in skel_dump
762 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
763 walk_thread(thread_results($ctx, $msgs), $ctx, *skel_dump);
765 $ctx->{parent_msg} = $parent;
768 sub _parent_headers {
769 my ($hdr, $over) = @_;
771 my @irt = $hdr->header_raw('In-Reply-To');
774 my $lnk = PublicInbox::Linkify->new;
775 $rv .= "In-Reply-To: $_\n" for @irt;
776 $lnk->linkify_mids('..', \$rv);
778 $refs = references($hdr);
779 my $irt = pop @$refs;
781 my $v = PublicInbox::Hval->new_msgid($irt);
782 my $html = $v->as_html;
783 my $href = $v->{href};
784 $rv .= "In-Reply-To: <";
785 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
789 # do not display References: if search is present,
790 # we show the thread skeleton at the bottom, instead.
793 $refs //= references($hdr);
795 @$refs = map { linkify_ref_no_over($_) } @$refs;
796 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
802 my ($hdr, $standalone, $ctx, $rhref) = @_;
804 my $ibx = $ctx->{-inbox} if $ctx;
807 my $idx = $standalone ? " <a\nhref=\"$upfx\">index</a>" : '';
809 if ($idx && $ibx->over) {
811 thread_skel(\$idx, $ctx, $hdr, $tpfx);
816 if (my $n = $ctx->{next_msg}) {
817 $n = PublicInbox::Hval->new_msgid($n)->{href};
818 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
821 my $par = $ctx->{parent_msg};
823 $u = PublicInbox::Hval->new_msgid($par)->{href};
826 if (my $p = $ctx->{prev_msg}) {
827 $prev = PublicInbox::Hval->new_msgid($p)->{href};
828 if ($p && $par && $p eq $par) {
829 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
830 'rel=prev>prev parent</a>';
833 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
835 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
837 } elsif ($u) { # unlikely
838 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
840 $irt = "$next $prev$parent ";
845 $irt .= qq(<a\nhref="$rhref">reply</a>);
849 sub linkify_ref_no_over {
850 my $v = PublicInbox::Hval->new_msgid($_[0]);
851 my $html = $v->as_html;
852 my $href = $v->{href};
853 "<<a\nhref=\"../$href/\">$html</a>>";
858 'm' . id_compress($msgid, 1);
862 my ($upfx, $mid) = @_;
864 $mid = PublicInbox::Hval->new_msgid($mid);
865 my $href = $mid->{href};
866 my $html = $mid->as_html;
867 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
872 $level ? INDENT x ($level - 1) : '';
876 my ($ctx, $level, $node, $idx) = @_;
877 ++$ctx->{root_idx} if $level == 0;
878 if ($node->{id} eq $ctx->{mid}) {
879 $ctx->{found_mid_at} = $ctx->{root_idx};
885 sub strict_loose_note ($) {
888 " -- strict thread matches above, loose matches on Subject: below --\n";
890 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
892 " -- use mbox.gz link to download all $nr messages --\n";
898 my ($ctx, $msgs) = @_;
899 require PublicInbox::SearchThread;
900 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
902 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
903 # and preserved in the future. This bug is hidden by `sid' matches
904 # in get_thread, so we never noticed it until now. And even when
905 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
906 # bump since reindexing is expensive and users may not do it
908 # loose threading could've returned too many results,
909 # put the root the message we care about at the top:
910 my $mid = $ctx->{mid};
911 if (defined($mid) && scalar(@$rootset) > 1) {
912 $ctx->{root_idx} = -1;
913 my $nr = scalar @$msgs;
914 walk_thread($rootset, $ctx, *find_mid_root);
915 my $idx = $ctx->{found_mid_at};
916 if (defined($idx) && $idx != 0) {
917 my $tip = splice(@$rootset, $idx, 1);
918 @$rootset = reverse @$rootset;
919 unshift @$rootset, $tip;
920 $ctx->{sl_note} = strict_loose_note($nr);
928 require PublicInbox::ExtMsg;
929 PublicInbox::ExtMsg::ext_msg($ctx);
933 my ($prev_subj, $subj, $val) = @_;
935 my $omit = ''; # '"' denotes identical text omitted
936 my (@prev_pop, @curr_pop);
937 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
938 push(@prev_pop, pop(@$prev_subj));
939 push(@curr_pop, pop(@$subj));
942 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
943 if (scalar(@curr_pop) == 1) {
945 push @$prev_subj, @prev_pop;
946 push @$subj, @curr_pop;
952 my ($ctx, $level, $node) = @_;
953 my $smsg = $node->{smsg} or return _skel_ghost($ctx, $level, $node);
955 my $dst = $ctx->{dst};
956 my $cur = $ctx->{cur};
957 my $mid = $smsg->{mid};
959 if ($level == 0 && $ctx->{skel_dump_roots}++) {
960 $$dst .= delete $ctx->{sl_note} || '';
963 my $f = ascii_html($smsg->from_name);
964 my $obfs_ibx = $ctx->{-obfs_ibx};
965 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
967 my $d = fmt_ts($smsg->{ds});
968 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
969 if (exists $ctx->{searchview}) {
970 if (defined(my $pct = $smsg->{pct})) {
971 $d .= (sprintf(' % 2u', $pct) . '%');
977 $d .= ' ' . indent_for($level) . th_pfx($level);
979 $ctx->{first_level} ||= $level;
981 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
982 $ctx->{prev_attr} = $attr;
984 $ctx->{prev_level} = $level;
989 $$dst .= "<b>$d<a\nid=r\nhref=\"#t\">".
990 "$attr [this message]</a></b>\n";
993 $ctx->{prev_msg} = $mid;
996 $ctx->{next_msg} ||= $mid;
999 # Subject is never undef, this mail was loaded from
1000 # our Xapian which would've resulted in '' if it were
1001 # really missing (and Filter rejects empty subjects)
1002 my @subj = split(/ /, subject_normalized($smsg->subject));
1003 # remove common suffixes from the subject if it matches the previous,
1004 # so we do not show redundant text at the end.
1005 my $prev_subj = $ctx->{prev_subj} || [];
1006 $ctx->{prev_subj} = [ @subj ];
1007 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
1010 my $subj = join(' ', @subj);
1011 $subj = ascii_html($subj);
1012 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1013 $end = "$subj</a> $omit$f\n"
1019 my $mapping = $unmatched ? undef : $ctx->{mapping};
1021 my $map = $mapping->{$mid};
1022 $id = id_compress($mid, 1);
1024 $map->[0] = "$d<a\nhref=\"$m\">$end";
1027 $m = $ctx->{-upfx}.mid_escape($mid).'/';
1029 $$dst .= $d . "<a\nhref=\"$m\"$id>" . $end;
1034 my ($ctx, $level, $node) = @_;
1036 my $mid = $node->{id};
1037 my $d = ' [not found] ';
1038 $d .= ' ' if exists $ctx->{searchview};
1039 $d .= indent_for($level) . th_pfx($level);
1040 my $upfx = $ctx->{-upfx};
1041 my $m = PublicInbox::Hval->new_msgid($mid);
1042 my $href = $upfx . $m->{href} . '/';
1043 my $html = $m->as_html;
1045 my $mapping = $ctx->{mapping};
1046 my $map = $mapping->{$mid} if $mapping;
1048 my $id = id_compress($mid, 1);
1049 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1050 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1052 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1054 my $dst = $ctx->{dst};
1061 (eval { $a->topmost->{smsg}->ds } || 0) <=>
1062 (eval { $b->topmost->{smsg}->ds } || 0)
1066 # accumulate recent topics if search is supported
1067 # returns 200 if done, 404 if not
1069 my ($ctx, $level, $node) = @_;
1070 my $mid = $node->{id};
1071 my $x = $node->{smsg} || $ctx->{-inbox}->smsg_by_mid($mid);
1075 $subj = $x->subject;
1076 $subj = subject_normalized($subj);
1077 $subj = '(no subject)' if $subj eq '';
1080 $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1081 $ctx->{-cur_topic} = $topic;
1082 push @{$ctx->{order}}, $topic;
1086 $topic = $ctx->{-cur_topic}; # should never be undef
1087 $topic->[0] = $ds if $ds > $topic->[0];
1089 my $seen = $topic->[2];
1090 if (scalar(@$topic) == 3) { # parent was a ghost
1091 push @$topic, $subj;
1092 } elsif (!$seen->{$subj}) {
1093 push @$topic, $level, $subj;
1095 $seen->{$subj} = $mid; # latest for subject
1096 } else { # ghost message
1097 return 1 if $level != 0; # ignore child ghosts
1098 $topic = [ -666, 0, {} ];
1099 $ctx->{-cur_topic} = $topic;
1100 push @{$ctx->{order}}, $topic;
1107 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1109 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1114 my $ibx = $ctx->{-inbox};
1115 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1117 # sort by recency, this allows new posts to "bump" old topics...
1118 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1119 my ($ds, $n, $seen, $top, @ex) = @$topic;
1121 next unless defined $top; # ghost topic
1122 my $mid = delete $seen->{$top};
1123 my $href = mid_escape($mid);
1124 my $prev_subj = [ split(/ /, $top) ];
1125 $top = PublicInbox::Hval->new($top)->as_html;
1128 # $n isn't the total number of posts on the topic,
1129 # just the number of posts in the current results window
1133 $anchor = '#u'; # top of only message
1135 $n = " ($n+ messages)";
1136 $anchor = '#t'; # thread skeleton
1139 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1140 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1141 my $s = "<a\nhref=\"$href/T/$anchor\">$top</a>\n" .
1142 " $ds UTC $n - $mbox / $atom\n";
1143 for (my $i = 0; $i < scalar(@ex); $i += 2) {
1144 my $level = $ex[$i];
1145 my $subj = $ex[$i + 1];
1146 $mid = delete $seen->{$subj};
1147 my @subj = split(/ /, subject_normalized($subj));
1148 my @next_prev = @subj; # full copy
1149 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1150 $prev_subj = \@next_prev;
1151 $subj = ascii_html(join(' ', @subj));
1152 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1153 $href = mid_escape($mid);
1154 $s .= indent_for($level) . TCHILD;
1155 $s .= "<a\nhref=\"$href/T/#u\">$subj</a>$omit\n";
1159 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1163 # only for the t= query parameter passed to overview DB
1164 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1167 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1168 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1171 sub pagination_footer ($$) {
1172 my ($ctx, $latest) = @_;
1173 delete $ctx->{qp} or return;
1174 my $next = $ctx->{next_page} || '';
1175 my $prev = $ctx->{prev_page} || '';
1177 $next = $next ? "$next " : ' ';
1178 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1180 "<hr><pre>page: $next$prev</pre>";
1183 sub index_nav { # callback for WwwStream
1184 my (undef, $ctx) = @_;
1185 pagination_footer($ctx, '.')
1188 sub paginate_recent ($$) {
1189 my ($ctx, $lim) = @_;
1190 my $t = $ctx->{qp}->{t} || '';
1191 my $opts = { limit => $lim };
1192 my ($after, $before);
1194 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1195 # if only $after exists "YYYYMMDD.." because "." could be skipped
1196 # if interpreted as an end-of-sentence
1197 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1198 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1200 my $ibx = $ctx->{-inbox};
1201 my $msgs = $ibx->recent($opts, $after, $before);
1202 my $nr = scalar @$msgs;
1203 if ($nr < $lim && defined($after)) {
1204 $after = $before = undef;
1205 $msgs = $ibx->recent($opts);
1206 $nr = scalar @$msgs;
1208 my $more = $nr == $lim;
1209 my ($newest, $oldest);
1211 $newest = $msgs->[0]->{ts};
1212 $oldest = $msgs->[-1]->{ts};
1213 # if we only had $after, our SQL query in ->recent ordered
1214 if ($newest < $oldest) {
1215 ($oldest, $newest) = ($newest, $oldest);
1216 $more = 0 if defined($after) && $after < $oldest;
1219 if (defined($oldest) && $more) {
1220 my $s = ts2str($oldest);
1221 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1223 if (defined($newest) && (defined($before) || defined($after))) {
1224 my $s = ts2str($newest);
1225 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1232 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1234 walk_thread(thread_results($ctx, $msgs), $ctx, *acc_topic);
1236 PublicInbox::WwwStream->response($ctx, dump_topics($ctx), *index_nav);
1239 sub thread_adj_level {
1240 my ($ctx, $level) = @_;
1242 my $max = $ctx->{cur_level};
1244 return ('', '') if $max == 0; # flat output
1246 # reset existing lists
1247 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1248 $ctx->{cur_level} = 0;
1250 } elsif ($level == $max) { # continue existing list
1252 } elsif ($level < $max) {
1253 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1254 $ctx->{cur_level} = $level;
1255 ("$beg<li>", '</li>');
1256 } else { # ($level > $max) # start a new level
1257 $ctx->{cur_level} = $level;
1258 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1263 sub ghost_index_entry {
1264 my ($ctx, $level, $node) = @_;
1265 my ($beg, $end) = thread_adj_level($ctx, $level);
1266 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{id})