1 # Copyright (C) 2014-2021 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 List::Util qw(max);
10 use PublicInbox::MsgTime qw(msg_datestamp);
11 use PublicInbox::Hval qw(ascii_html obfuscate_addrs prurl mid_href
13 use PublicInbox::Linkify;
14 use PublicInbox::MID qw(id_compress mids mids_for_index references
16 use PublicInbox::MsgIter;
17 use PublicInbox::Address;
18 use PublicInbox::WwwStream qw(html_oneshot);
19 use PublicInbox::Reply;
20 use PublicInbox::ViewDiff qw(flush_diff);
22 use Time::Local qw(timegm);
23 use PublicInbox::Smsg qw(subject_normalized);
24 use PublicInbox::ContentHash qw(content_hash);
25 use constant COLS => 72;
26 use constant INDENT => ' ';
27 use constant TCHILD => '` ';
28 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
32 if ($eml) { # called by WwwStream::async_eml or getline
33 my $smsg = $ctx->{smsg};
34 my $over = $ctx->{ibx}->over;
35 $ctx->{smsg} = $over ? $over->next_by_mid(@{$ctx->{next_arg}})
37 $ctx->{mhref} = ($ctx->{nr} || $ctx->{smsg}) ?
38 "../${\mid_href($smsg->{mid})}/" : '';
39 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($eml, $ctx);
41 multipart_text_as_html($eml, $ctx);
42 $$obuf .= '</pre><hr>';
45 $$obuf .= html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
47 } else { # called by WwwStream::async_next or getline
48 $ctx->{smsg}; # may be undef
52 # /$INBOX/$MSGID/ for unindexed v1 inboxes
53 sub no_over_html ($) {
55 my $bref = $ctx->{ibx}->msg_by_mid($ctx->{mid}) or return; # 404
56 my $eml = PublicInbox::Eml->new($bref);
58 PublicInbox::WwwStream::init($ctx);
59 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($eml, $ctx);
61 multipart_text_as_html($eml, $ctx);
62 $$obuf .= '</pre><hr>';
65 eval { $$obuf .= html_footer($ctx, $eml) };
66 html_oneshot($ctx, 200, $obuf);
69 # public functions: (unstable)
73 my $ibx = $ctx->{ibx};
74 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
75 my $over = $ibx->over or return no_over_html($ctx);
77 my $next_arg = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
79 my $smsg = $ctx->{smsg} = $over->next_by_mid(@$next_arg) or
80 return; # undef == 404
82 # allow user to easily browse the range around this message if
84 $ctx->{-t_max} = $smsg->{ts};
85 PublicInbox::WwwStream::aresponse($ctx, 200, \&msg_page_i);
88 # /$INBOX/$MESSAGE_ID/#R
92 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
94 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
97 my $ibx = $ctx->{ibx};
98 if (my $url = $ibx->{infourl}) {
99 $url = prurl($ctx->{env}, $url);
100 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
103 my ($arg, $link, $reply_to_all) =
104 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
105 if (ref($arg) eq 'SCALAR') {
106 return '<pre id=R>'.ascii_html($$arg).'</pre>';
109 # mailto: link only works if address obfuscation is disabled
113 * If your mail client supports setting the <b>In-Reply-To</b> header
114 via mailto: links, try the <a
115 href="$link">mailto: link</a>
119 push @$arg, '/path/to/YOUR_REPLY';
120 $arg = ascii_html(join(" \\\n ", '', @$arg));
123 id=R><b>Reply instructions:</b>
125 You may reply publicly to <a
126 href=#t>this message</a> via plain-text email
127 using any one of the following methods:
129 * Save the following mbox file, import it into your mail client,
130 and $reply_to_all from there: <a
133 Avoid top-posting and favor interleaved quoting:
135 href="$p_url">$p_url</a>
137 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
138 switches of git-send-email(1):
143 href="$se_url">$se_url</a>
150 my $refs = references($hdr);
154 sub fold_addresses ($) {
155 return $_[0] if length($_[0]) <= COLS;
156 # try to fold on commas after non-word chars before $lim chars,
157 # Try to get the "," preceded by ">" or ")", but avoid folding
158 # on the comma where somebody uses "Lastname, Firstname".
159 # We also try to keep the last and penultimate addresses in
160 # the list on the same line if possible, hence the extra \z
161 # Fall back to folding on spaces at $lim + 1 chars
162 my $lim = COLS - 8; # 8 = "\t" display width
163 my $too_long = $lim + 1;
164 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
166 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
169 .{$too_long,}?)(?:\s|\z)/xgo));
172 sub _hdr_names_html ($$) {
173 my ($hdr, $field) = @_;
174 my @vals = $hdr->header($field) or return '';
175 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
179 my ($nr, $singular, $plural) = @_;
180 return "0 $plural" if $nr == 0;
181 $nr == 1 ? "$nr $singular" : "$nr $plural";
184 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
185 # this is already inside a <pre>
187 my ($ctx, $eml) = @_;
188 my $smsg = delete $ctx->{smsg};
189 my $subj = delete $smsg->{subject};
190 my $mid_raw = $smsg->{mid};
191 my $id = id_compress($mid_raw, 1);
193 my $root_anchor = $ctx->{root_anchor} || '';
195 my $obfs_ibx = $ctx->{-obfs_ibx};
197 $subj = '(no subject)' if $subj eq '';
198 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
199 $subj = '<b>'.ascii_html($subj).'</b>';
200 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
201 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
203 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
205 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
207 # Deleting these fields saves about 400K as we iterate across 1K msgs
208 delete @$smsg{qw(ts blob)};
210 my $from = _hdr_names_html($eml, 'From');
211 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
212 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
213 my $upfx = $ctx->{-upfx};
214 my $mhref = $upfx . mid_href($mid_raw) . '/';
215 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
216 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
217 my $to = fold_addresses(_hdr_names_html($eml, 'To'));
218 my $cc = fold_addresses(_hdr_names_html($eml, 'Cc'));
219 my ($tlen, $clen) = (length($to), length($cc));
221 if (($tlen + $clen) > COLS) {
222 $to_cc .= ' To: '.$to."\n" if $tlen;
223 $to_cc .= ' Cc: '.$cc."\n" if $clen;
226 $to_cc .= ' To: '.$to;
227 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
229 $to_cc .= ' Cc: '.$cc if $clen;
233 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
236 my $mapping = $ctx->{mapping};
237 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($eml)))) {
238 my $href = $upfx . mid_href($irt) . '/';
239 my $html = ascii_html($irt);
240 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
244 # scan through all parts, looking for displayable text
245 $ctx->{mhref} = $mhref;
247 $eml->each_part(\&add_text_body, $ctx, 1);
251 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
252 "<a\nhref=\"$mhref\">permalink</a>" .
253 " <a\nhref=\"${mhref}raw\">raw</a>" .
254 " <a\nhref=\"${mhref}#R\">reply</a>";
257 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
258 $rv .= "\t[relevance $pct%]";
261 my $nested = 'nested';
266 $flat = "<b>$flat</b>";
268 $nested = "<b>$nested</b>";
270 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
271 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
272 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
277 # do we have more messages? start a new <pre> if so
278 $rv .= scalar(@{$ctx->{msgs}}) ? '</pre><hr><pre>' : '</pre>' if $hr;
282 sub pad_link ($$;$) {
283 my ($mid, $level, $s) = @_;
285 my $href = defined($mid) ?
286 ("<a\nhref=#r".id_compress($mid, 1).">($s)</a>\n") :
288 (' 'x19).indent_for($level).th_pfx($level).$href;
292 # my ($mapping, $mid) = @_;
293 ($_[0]->{$_[1] // \'bogus'} // [ "(?)\n" ])->[0];
297 my ($mid_raw, $irt, $id, $ctx) = @_;
299 my $mapping = $ctx->{mapping} or return $rv;
301 my $mid_map = $mapping->{$mid_raw};
303 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
304 my ($attr, $node, $idx, $level) = @$mid_map;
305 my $children = $node->{children};
306 my $nr_c = scalar @$children;
309 # delete saves about 200KB on a 1K message thread
310 if (my $refs = delete $node->{references}) {
311 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
313 my $irt_map = $mapping->{$$irt} if defined $$irt;
314 if (defined $irt_map) {
315 $siblings = $irt_map->[1]->{children};
316 $nr_s = scalar(@$siblings) - 1;
317 $rv .= $pad . $irt_map->[0];
319 my $prev = $siblings->[$idx - 1];
320 my $pmid = $prev->{mid};
322 my $s = ($idx - 1). ' preceding siblings ...';
323 $rv .= pad_link($pmid, $level, $s);
324 } elsif ($idx == 2) {
325 $rv .= $pad . _skel_hdr($mapping,
327 $siblings->[0]->{mid} : undef);
329 $rv .= $pad . _skel_hdr($mapping, $pmid);
332 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
333 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
334 $attr =~ s!\n\z!</b>\n!s;
335 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
336 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
339 my $cmid = $children->[0] ? $children->[0]->{mid} : undef;
340 $rv .= $pad . _skel_hdr($mapping, $cmid);
342 my $s = ($nr_c - 1). ' more replies';
343 $rv .= pad_link($cmid, $level + 1, $s);
344 } elsif (my $cn = $children->[1]) {
345 $rv .= $pad . _skel_hdr($mapping, $cn->{mid});
349 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
351 my $nmid = $next->{mid};
352 $rv .= $pad . _skel_hdr($mapping, $nmid);
353 my $nnext = $nr_s - $idx;
355 my $s = ($nnext - 1).' subsequent siblings';
356 $rv .= pad_link($nmid, $level, $s);
357 } elsif (my $nn = $siblings->[$idx + 2]) {
358 $rv .= $pad . _skel_hdr($mapping, $nn->{mid});
361 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
364 # non-recursive thread walker
365 sub walk_thread ($$$) {
366 my ($rootset, $ctx, $cb) = @_;
367 my @q = map { (0, $_, -1) } @$rootset;
369 my ($level, $node, $i) = splice(@q, 0, 3);
370 defined $node or next;
371 $cb->($ctx, $level, $node, $i) or return;
374 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
378 sub pre_thread { # walk_thread callback
379 my ($ctx, $level, $node, $idx) = @_;
380 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
381 skel_dump($ctx, $level, $node);
384 sub thread_eml_entry {
385 my ($ctx, $eml) = @_;
386 my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
387 $beg . '<pre>' . eml_entry($ctx, $eml) . '</pre>' . $end;
390 sub next_in_queue ($$) {
391 my ($q, $ghost_ok) = @_;
393 my ($level, $smsg) = splice(@$q, 0, 2);
395 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
396 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
401 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
402 my ($ctx, $eml) = @_;
403 return thread_eml_entry($ctx, $eml) if $eml;
404 return unless exists($ctx->{skel});
405 my $ghost_ok = $ctx->{nr}++;
407 my ($lvl, $smsg) = next_in_queue($ctx->{-queue}, $ghost_ok);
409 if (exists $smsg->{blob}) { # next message for cat-file
410 $ctx->{level} = $lvl;
411 if (!$ghost_ok) { # first non-ghost
412 $ctx->{-title_html} =
413 ascii_html($smsg->{subject});
414 $ctx->zmore($ctx->html_top);
418 # buffer the ghost entry and loop
419 $ctx->zmore(ghost_index_entry($ctx, $lvl, $smsg));
421 $ctx->zmore(join('', thread_adj_level($ctx, 0)));
422 $ctx->zmore(${delete($ctx->{skel})});
428 sub stream_thread ($$) {
429 my ($rootset, $ctx) = @_;
430 $ctx->{-queue} = [ map { (0, $_) } @$rootset ];
431 PublicInbox::WwwStream::aresponse($ctx, 200, \&stream_thread_i);
434 # /$INBOX/$MSGID/t/ and /$INBOX/$MSGID/T/
437 $ctx->{-upfx} = '../../';
438 my $mid = $ctx->{mid};
439 my $ibx = $ctx->{ibx};
440 my ($nr, $msgs) = $ibx->over->get_thread($mid);
441 return missing_thread($ctx) if $nr == 0;
443 # link $INBOX_DIR/description text to "index_topics" view around
444 # the newest message in this thread
445 my $t = ts2str($ctx->{-t_max} = max(map { delete $_->{ts} } @$msgs));
446 my $t_fmt = fmt_ts($ctx->{-t_max});
448 my $skel = '<hr><pre>';
449 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
452 href="../../?t=$t">~$t_fmt UTC</a> | <a
453 href="../../">newest</a>]
456 $skel .= "<b\nid=t>Thread overview:</b> ";
457 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
458 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
459 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
460 $skel .= "-- links below jump to the message on this page --\n";
461 $ctx->{cur_level} = 0;
462 $ctx->{skel} = \$skel;
463 $ctx->{prev_attr} = '';
464 $ctx->{prev_level} = 0;
465 $ctx->{root_anchor} = 'm' . id_compress($mid, 1);
466 $ctx->{mapping} = {}; # mid -> [ header_summary, node, idx, level ]
467 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
470 my $rootset = thread_results($ctx, $msgs);
472 # reduce hash lookups in pre_thread->skel_dump
473 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
474 walk_thread($rootset, $ctx, \&pre_thread);
477 return stream_thread($rootset, $ctx) unless $ctx->{flat};
479 # flat display: lazy load the full message from smsg
480 $ctx->{msgs} = $msgs;
481 $ctx->{-html_tip} = '<pre>';
482 PublicInbox::WwwStream::aresponse($ctx, 200, \&thread_html_i);
485 sub thread_html_i { # PublicInbox::WwwStream::getline callback
486 my ($ctx, $eml) = @_;
488 my $smsg = $ctx->{smsg};
489 if (exists $ctx->{-html_tip}) {
490 $ctx->{-title_html} = ascii_html($smsg->{subject});
491 $ctx->zmore($ctx->html_top);
493 return eml_entry($ctx, $eml);
495 while (my $smsg = shift @{$ctx->{msgs}}) {
496 return $smsg if exists($smsg->{blob});
498 my $skel = delete($ctx->{skel}) or return; # all done
504 sub multipart_text_as_html {
505 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
507 # scan through all parts, looking for displayable text
508 $_[0]->each_part(\&add_text_body, $_[1], 1);
511 sub submsg_hdr ($$) {
512 my ($ctx, $eml) = @_;
513 my $obfs_ibx = $ctx->{-obfs_ibx};
514 my $rv = $ctx->{obuf};
516 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
517 my @v = $eml->header($h);
519 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
526 sub attach_link ($$$$;$) {
527 my ($ctx, $ct, $p, $fn, $err) = @_;
528 my ($part, $depth, $idx) = @$p;
530 # Eml iteration clobbers multipart ->{bdy}, so do not offer
531 # downloads for 0-byte multipart attachments
532 return unless $part->{bdy};
534 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
535 my $size = length($part->body);
536 delete $part->{bdy}; # save memory
538 # hide attributes normally, unless we want to aid users in
539 # spotting MUA problems:
540 $ct =~ s/;.*// unless $err;
541 $ct = ascii_html($ct);
543 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
545 } elsif ($ct eq 'text/plain') {
550 my $rv = $ctx->{obuf};
551 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
554 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
557 $$rv .= "[-- Attachment #$idx: ";
558 my $ts = "Type: $ct, Size: $size bytes";
559 my $desc = $part->header('Content-Description') // $fn // '';
560 $desc = ascii_html($desc);
561 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
564 submsg_hdr($ctx, $part) if $part->{is_submsg};
569 sub add_text_body { # callback for each_part
571 my $upfx = $ctx->{mhref};
572 my $ibx = $ctx->{ibx};
573 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
574 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
575 my ($part, $depth, $idx) = @$p;
576 my $ct = $part->content_type || 'text/plain';
577 my $fn = $part->filename;
578 my ($s, $err) = msg_part_text($part, $ct);
579 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
581 my $rv = $ctx->{obuf};
582 if ($part->{is_submsg}) {
583 submsg_hdr($ctx, $part);
587 # makes no difference to browsers, and don't screw up filename
588 # link generation in diffs with the extra '%0D'
591 # will be escaped to `•' in HTML
592 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
594 # always support diff-highlighting, but we can't linkify hunk
595 # headers for solver unless some coderepo are configured:
597 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
598 # diffstat anchors do not link across attachments or messages,
599 # -apfx is just a stable prefix for making diffstat anchors
600 # linkable to the first diff hunk w/o crossing attachments
601 $idx =~ tr!.!/!; # compatibility with previous versions
602 $ctx->{-apfx} = $upfx . $idx;
604 # do attr => filename mappings for diffstats in git diffs:
605 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
607 delete $ctx->{-long_path};
609 # absolute URL (Atom feeds)
610 if ($ibx->{coderepo}) {
611 if (index($upfx, '//') >= 0) {
613 $spfx =~ s!/([^/]*)/\z!/!;
615 my $n_slash = $upfx =~ tr!/!/!;
618 } elsif ($n_slash == 1) {
620 } else { # nslash == 2
625 $ctx->{-spfx} = $spfx;
628 # split off quoted and unquoted blocks:
629 my @sections = PublicInbox::MsgIter::split_quotes($s);
630 undef $s; # free memory
631 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
632 # badly-encoded message with $err? tell the world about it!
633 attach_link($ctx, $ct, $p, $fn, $err);
636 delete $part->{bdy}; # save memory
637 foreach my $cur (@sections) {
639 # we use a <span> here to allow users to specify
640 # their own color for quoted text
641 $$rv .= qq(<span\nclass="q">);
642 $$rv .= $l->to_html($cur);
645 flush_diff($ctx, \$cur);
648 $$rv .= $l->to_html($cur);
650 undef $cur; # free memory
654 sub _msg_page_prepare_obuf {
655 my ($eml, $ctx) = @_;
656 my $over = $ctx->{ibx}->over;
657 my $obfs_ibx = $ctx->{-obfs_ibx};
659 my $mids = mids_for_index($eml);
660 my $nr = $ctx->{nr}++;
661 if ($nr) { # unlikely
662 if ($ctx->{chash} eq content_hash($eml)) {
663 warn "W: BUG? @$mids not deduplicated properly\n";
667 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
670 $ctx->{first_hdr} = $eml->header_obj;
671 $ctx->{chash} = content_hash($eml) if $ctx->{smsg}; # reused MID
672 $rv .= "<pre\nid=b>"; # anchor for body start
674 $ctx->{-upfx} = '../' if $over;
675 my @title; # (Subject[0], From[0])
676 for my $v ($eml->header('From')) {
677 my @n = PublicInbox::Address::names($v);
679 $title[1] //= ascii_html(join(', ', @n));
681 obfuscate_addrs($obfs_ibx, $v);
682 obfuscate_addrs($obfs_ibx, $title[1]);
684 $rv .= "From: $v\n" if $v ne '';
686 foreach my $h (qw(To Cc)) {
687 for my $v ($eml->header($h)) {
690 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
691 $rv .= "$h: $v\n" if $v ne '';
694 my @subj = $eml->header('Subject');
696 my $v = ascii_html(shift @subj);
697 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
699 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
701 for $v (@subj) { # multi-Subject message :<
703 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
704 $rv .= "Subject: $v\n";
706 } else { # dummy anchor for thread skeleton at bottom of page
707 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
708 $title[0] = '(no subject)';
710 for my $v ($eml->header('Date')) {
712 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
713 $rv .= qq{Date: $v\t<a\nhref="#r">[thread overview]</a>\n};
715 if (!$nr) { # first (and only) message, common case
716 $ctx->{-title_html} = join(' - ', @title);
717 $rv = $ctx->html_top . $rv;
719 if (scalar(@$mids) == 1) { # common case
720 my $mhtml = ascii_html($mids->[0]);
721 $rv .= "Message-ID: <$mhtml> ";
722 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
724 # X-Alt-Message-ID can happen if a message is injected from
725 # public-inbox-nntpd because of multiple Message-ID headers.
726 my $lnk = PublicInbox::Linkify->new;
728 for my $h (qw(Message-ID X-Alt-Message-ID)) {
729 $s .= "$h: $_\n" for ($eml->header_raw($h));
731 $lnk->linkify_mids('..', \$s, 1);
734 $rv .= _parent_headers($eml, $over);
740 qq(expand[<a\nhref="T/#u">flat</a>) .
741 qq(|<a\nhref="t/#u">nested</a>] ) .
742 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
743 qq(<a\nhref="t.atom">Atom feed</a>);
746 sub thread_skel ($$$) {
747 my ($skel, $ctx, $hdr) = @_;
748 my $mid = mids($hdr)->[0];
749 my $ibx = $ctx->{ibx};
750 my ($nr, $msgs) = $ibx->over->get_thread($mid);
751 my $parent = in_reply_to($hdr);
752 $$skel .= "\n<b>Thread overview: </b>";
754 if (defined $parent) {
755 $$skel .= SKEL_EXPAND."\n ";
756 $$skel .= ghost_parent('../', $parent) . "\n";
758 $$skel .= "<a\nid=r>[no followups]</a> ".
761 $ctx->{next_msg} = undef;
762 $ctx->{parent_msg} = $parent;
767 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
769 # nb: mutt only shows the first Subject in the index pane
770 # when multiple Subject: headers are present, so we follow suit:
771 my $subj = $hdr->header('Subject') // '';
772 $subj = '(no subject)' if $subj eq '';
773 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
775 $ctx->{prev_attr} = '';
776 $ctx->{prev_level} = 0;
777 $ctx->{skel} = $skel;
779 # reduce hash lookups in skel_dump
780 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
781 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
783 $ctx->{parent_msg} = $parent;
786 sub _parent_headers {
787 my ($hdr, $over) = @_;
789 my @irt = $hdr->header_raw('In-Reply-To');
792 my $lnk = PublicInbox::Linkify->new;
793 $rv .= "In-Reply-To: $_\n" for @irt;
794 $lnk->linkify_mids('..', \$rv);
796 $refs = references($hdr);
797 my $irt = pop @$refs;
799 my $html = ascii_html($irt);
800 my $href = mid_href($irt);
801 $rv .= "In-Reply-To: <";
802 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
806 # do not display References: if search is present,
807 # we show the thread skeleton at the bottom, instead.
810 $refs //= references($hdr);
812 @$refs = map { linkify_ref_no_over($_) } @$refs;
813 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
818 # returns a string buffer
820 my ($ctx, $hdr) = @_;
821 my $ibx = $ctx->{ibx};
826 my $t = ts2str($ctx->{-t_max});
827 my $t_fmt = fmt_ts($ctx->{-t_max});
830 href="$upfx?t=$t">~$t_fmt UTC</a>|<a
831 href="$upfx">newest</a>]
834 thread_skel(\$skel, $ctx, $hdr);
839 if (my $n = $ctx->{next_msg}) {
841 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
844 my $par = $ctx->{parent_msg};
849 if (my $p = $ctx->{prev_msg}) {
850 $prev = mid_href($p);
851 if ($p && $par && $p eq $par) {
852 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
853 'rel=prev>prev parent</a>';
856 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
858 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
860 } elsif ($u) { # unlikely
861 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
863 $rv .= "$next $prev$parent ";
864 } else { # unindexed inboxes w/o over
865 $skel = qq( <a\nhref="$upfx">latest</a>);
867 $rv .= qq(<a\nhref="#R">reply</a>);
870 $rv .= msg_reply($ctx, $hdr);
873 sub linkify_ref_no_over {
875 my $href = mid_href($mid);
876 my $html = ascii_html($mid);
877 "<<a\nhref=\"../$href/\">$html</a>>";
881 my ($upfx, $mid) = @_;
883 my $href = mid_href($mid);
884 my $html = ascii_html($mid);
885 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
890 $level ? INDENT x ($level - 1) : '';
894 my ($ctx, $level, $node, $idx) = @_;
895 ++$ctx->{root_idx} if $level == 0;
896 if ($node->{mid} eq $ctx->{mid}) {
897 $ctx->{found_mid_at} = $ctx->{root_idx};
898 return 0; # stop iterating
903 sub strict_loose_note ($) {
906 " -- strict thread matches above, loose matches on Subject: below --\n";
908 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
910 " -- use mbox.gz link to download all $nr messages --\n";
916 my ($ctx, $msgs) = @_;
917 require PublicInbox::SearchThread;
918 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
920 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
921 # and preserved in the future. This bug is hidden by `sid' matches
922 # in get_thread, so we never noticed it until now. And even when
923 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
924 # bump since reindexing is expensive and users may not do it
926 # loose threading could've returned too many results,
927 # put the root the message we care about at the top:
928 my $mid = $ctx->{mid};
929 if (defined($mid) && scalar(@$rootset) > 1) {
930 $ctx->{root_idx} = -1;
931 my $nr = scalar @$msgs;
932 walk_thread($rootset, $ctx, \&find_mid_root);
933 my $idx = $ctx->{found_mid_at};
934 if (defined($idx) && $idx != 0) {
935 my $tip = splice(@$rootset, $idx, 1);
936 @$rootset = reverse @$rootset;
937 unshift @$rootset, $tip;
938 $ctx->{sl_note} = strict_loose_note($nr);
946 require PublicInbox::ExtMsg;
947 PublicInbox::ExtMsg::ext_msg($ctx);
951 my ($prev_subj, $subj, $val) = @_;
953 my $omit; # '"' denotes identical text omitted
954 my (@prev_pop, @curr_pop);
955 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
956 push(@prev_pop, pop(@$prev_subj));
957 push(@curr_pop, pop(@$subj));
960 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
961 if (scalar(@curr_pop) == 1) {
963 push @$prev_subj, @prev_pop;
964 push @$subj, @curr_pop;
969 sub skel_dump { # walk_thread callback
970 my ($ctx, $level, $smsg) = @_;
971 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
973 my $skel = $ctx->{skel};
974 my $cur = $ctx->{cur};
975 my $mid = $smsg->{mid};
977 if ($level == 0 && $ctx->{skel_dump_roots}++) {
978 $$skel .= delete($ctx->{sl_note}) || '';
981 my $f = ascii_html(delete $smsg->{from_name});
982 my $obfs_ibx = $ctx->{-obfs_ibx};
983 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
985 my $d = fmt_ts($smsg->{ds});
986 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
987 if (exists $ctx->{searchview}) {
988 if (defined(my $pct = $smsg->{pct})) {
989 $d .= (sprintf(' % 2u', $pct) . '%');
995 $d .= ' ' . indent_for($level) . th_pfx($level);
997 $ctx->{first_level} ||= $level;
999 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
1000 $ctx->{prev_attr} = $attr;
1002 $ctx->{prev_level} = $level;
1007 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
1008 "$attr [this message]</a></b>\n";
1011 $ctx->{prev_msg} = $mid;
1014 $ctx->{next_msg} ||= $mid;
1017 # Subject is never undef, this mail was loaded from
1018 # our Xapian which would've resulted in '' if it were
1019 # really missing (and Filter rejects empty subjects)
1020 my @subj = split(/ /, subject_normalized($smsg->{subject}));
1021 # remove common suffixes from the subject if it matches the previous,
1022 # so we do not show redundant text at the end.
1023 my $prev_subj = $ctx->{prev_subj} || [];
1024 $ctx->{prev_subj} = [ @subj ];
1025 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
1028 my $subj = join(' ', @subj);
1029 $subj = ascii_html($subj);
1030 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1031 $end = "$subj</a> $omit$f\n"
1037 my $mapping = $unmatched ? undef : $ctx->{mapping};
1039 my $map = $mapping->{$mid};
1040 $id = id_compress($mid, 1);
1042 $map->[0] = "$d<a\nhref=\"$m\">$end";
1045 $m = $ctx->{-upfx}.mid_href($mid).'/';
1047 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1052 my ($ctx, $level, $node) = @_;
1054 my $mid = $node->{mid};
1055 my $d = ' [not found] ';
1056 $d .= ' ' if exists $ctx->{searchview};
1057 $d .= indent_for($level) . th_pfx($level);
1058 my $upfx = $ctx->{-upfx};
1059 my $href = $upfx . mid_href($mid) . '/';
1060 my $html = ascii_html($mid);
1062 my $mapping = $ctx->{mapping};
1063 my $map = $mapping->{$mid} if $mapping;
1065 my $id = id_compress($mid, 1);
1066 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1067 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1069 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1071 ${$ctx->{skel}} .= $d;
1077 (eval { $a->topmost->{ds} } || 0) <=>
1078 (eval { $b->topmost->{ds} } || 0)
1082 # accumulate recent topics if search is supported
1083 # returns 200 if done, 404 if not
1084 sub acc_topic { # walk_thread callback
1085 my ($ctx, $level, $smsg) = @_;
1086 my $mid = $smsg->{mid};
1087 my $has_blob = $smsg->{blob} // do {
1088 if (my $by_mid = $ctx->{ibx}->smsg_by_mid($mid)) {
1089 %$smsg = (%$smsg, %$by_mid);
1094 my $subj = subject_normalized($smsg->{subject});
1095 $subj = '(no subject)' if $subj eq '';
1096 my $ds = $smsg->{ds};
1097 if ($level == 0) { # new, top-level topic
1098 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1099 $ctx->{-cur_topic} = $topic;
1100 push @{$ctx->{order}}, $topic;
1104 # continue existing topic
1105 my $topic = $ctx->{-cur_topic}; # should never be undef
1106 $topic->[0] = $ds if $ds > $topic->[0];
1107 $topic->[1]++; # bump N+ message counter
1108 my $seen = $topic->[2];
1109 if (scalar(@$topic) == 3) { # parent was a ghost
1110 push @$topic, $subj;
1111 } elsif (!defined($seen->{$subj})) {
1112 push @$topic, $level, $subj; # @extra messages
1114 $seen->{$subj} = $mid; # latest for subject
1115 } else { # ghost message
1116 return 1 if $level != 0; # ignore child ghosts
1117 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1118 push @{$ctx->{order}}, $topic;
1125 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1127 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1132 my $ibx = $ctx->{ibx};
1133 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1135 # sort by recency, this allows new posts to "bump" old topics...
1136 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1137 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1139 next unless defined $top_subj; # ghost topic
1140 my $mid = delete $seen->{$top_subj};
1141 my $href = mid_href($mid);
1142 my $prev_subj = [ split(/ /, $top_subj) ];
1143 $top_subj = ascii_html($top_subj);
1146 # $n isn't the total number of posts on the topic,
1147 # just the number of posts in the current results window
1151 $anchor = '#u'; # top of only message
1153 $n = " ($n+ messages)";
1154 $anchor = '#t'; # thread skeleton
1157 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1159 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1160 my $level = $extra[$i];
1161 my $subj = $extra[$i + 1]; # already normalized
1162 $mid = delete $seen->{$subj};
1163 my @subj = split(/ /, $subj);
1164 my @next_prev = @subj; # full copy
1165 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1166 $prev_subj = \@next_prev;
1167 $subj = join(' ', @subj);
1168 $subj = ascii_html($subj);
1169 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1170 $href = mid_href($mid);
1171 $s .= indent_for($level) . TCHILD;
1172 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1176 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1181 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1182 timegm($ss || 0, $mm || 0, $hh || 0, $dd, $mon - 1, $yyyy);
1185 sub pagination_footer ($$) {
1186 my ($ctx, $latest) = @_;
1187 my $next = $ctx->{next_page} || '';
1188 my $prev = $ctx->{prev_page} || '';
1189 if ($prev) { # aligned padding for: 'next (older) | '
1190 $next = $next ? "$next | " : ' | ';
1191 $prev .= qq[ | <a\nhref="$latest">latest</a>];
1193 "<hr><pre>page: $next$prev</pre>";
1196 sub paginate_recent ($$) {
1197 my ($ctx, $lim) = @_;
1198 my $t = $ctx->{qp}->{t} || '';
1199 my $opts = { limit => $lim };
1200 my ($after, $before);
1202 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1203 # if only $after exists "YYYYMMDD.." because "." could be skipped
1204 # if interpreted as an end-of-sentence
1205 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1206 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1208 my $ibx = $ctx->{ibx};
1209 my $msgs = $ibx->recent($opts, $after, $before);
1210 my $nr = scalar @$msgs;
1211 if ($nr < $lim && defined($after)) {
1212 $after = $before = undef;
1213 $msgs = $ibx->recent($opts);
1214 $nr = scalar @$msgs;
1216 my $more = $nr == $lim;
1217 my ($newest, $oldest);
1219 $newest = $msgs->[0]->{ts};
1220 $oldest = $msgs->[-1]->{ts};
1221 # if we only had $after, our SQL query in ->recent ordered
1222 if ($newest < $oldest) {
1223 ($oldest, $newest) = ($newest, $oldest);
1224 $more = 0 if defined($after) && $after < $oldest;
1227 if (defined($oldest) && $more) {
1228 my $s = ts2str($oldest);
1229 $ctx->{next_page} = qq[<a\nhref="?t=$s"\nrel=next>] .
1232 if (defined($newest) && (defined($before) || defined($after))) {
1233 my $s = ts2str($newest);
1234 $ctx->{prev_page} = qq[<a\nhref="?t=$s-"\nrel=prev>] .
1240 # GET /$INBOX - top-level inbox view for indexed inboxes
1243 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1245 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1247 html_oneshot($ctx, dump_topics($ctx), \pagination_footer($ctx, '.'));
1251 sub thread_adj_level {
1252 my ($ctx, $level) = @_;
1254 my $max = $ctx->{cur_level};
1256 return ('', '') if $max == 0; # flat output
1258 # reset existing lists
1259 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1260 $ctx->{cur_level} = 0;
1262 } elsif ($level == $max) { # continue existing list
1264 } elsif ($level < $max) {
1265 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1266 $ctx->{cur_level} = $level;
1267 ("$beg<li>", '</li>');
1268 } else { # ($level > $max) # start a new level
1269 $ctx->{cur_level} = $level;
1270 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1275 sub ghost_index_entry {
1276 my ($ctx, $level, $node) = @_;
1277 my ($beg, $end) = thread_adj_level($ctx, $level);
1278 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid} // '?')