1 # Copyright (C) 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 Text::Wrap qw(wrap); # stdlib, we need Perl 5.6+ for $huge
11 use PublicInbox::MsgTime qw(msg_datestamp);
12 use PublicInbox::Hval qw(ascii_html obfuscate_addrs prurl mid_href
14 use PublicInbox::Linkify;
15 use PublicInbox::MID qw(id_compress mids mids_for_index references
17 use PublicInbox::MsgIter;
18 use PublicInbox::Address;
19 use PublicInbox::WwwStream qw(html_oneshot);
20 use PublicInbox::Reply;
21 use PublicInbox::ViewDiff qw(flush_diff);
23 use POSIX qw(strftime);
24 use Time::Local qw(timegm);
25 use PublicInbox::Smsg qw(subject_normalized);
26 use PublicInbox::ContentHash qw(content_hash);
27 use constant COLS => 72;
28 use constant INDENT => ' ';
29 use constant TCHILD => '` ';
30 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
34 if ($eml) { # called by WwwStream::async_eml or getline
35 my $smsg = $ctx->{smsg};
36 my $over = $ctx->{ibx}->over;
37 $ctx->{smsg} = $over ? $over->next_by_mid(@{$ctx->{next_arg}})
39 $ctx->{mhref} = ($ctx->{nr} || $ctx->{smsg}) ?
40 "../${\mid_href($smsg->{mid})}/" : '';
41 if (_msg_page_prepare_obuf($eml, $ctx)) {
42 multipart_text_as_html($eml, $ctx);
43 ${$ctx->{obuf}} .= '</pre><hr>';
45 html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
46 delete($ctx->{obuf}) // \'';
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($ctx);
59 if (_msg_page_prepare_obuf($eml, $ctx)) { # sets {-title_html}
60 multipart_text_as_html($eml, $ctx);
61 ${$ctx->{obuf}} .= '</pre><hr>';
63 html_footer($ctx, $eml);
67 # public functions: (unstable)
71 my $ibx = $ctx->{ibx};
72 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
73 my $over = $ibx->over or return no_over_html($ctx);
75 my $next_arg = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
77 my $smsg = $ctx->{smsg} = $over->next_by_mid(@$next_arg) or
78 return; # undef == 404
80 # allow user to easily browse the range around this message if
82 $ctx->{-t_max} = $smsg->{ts};
83 PublicInbox::WwwStream::aresponse($ctx, \&msg_page_i);
86 # /$INBOX/$MESSAGE_ID/#R
90 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
92 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
95 my $ibx = $ctx->{ibx};
96 if (my $url = $ibx->{infourl}) {
97 $url = prurl($ctx->{env}, $url);
98 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
101 my ($arg, $link, $reply_to_all) =
102 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
103 if (ref($arg) eq 'SCALAR') {
104 return '<pre id=R>'.ascii_html($$arg).'</pre>';
107 # mailto: link only works if address obfuscation is disabled
111 * If your mail client supports setting the <b>In-Reply-To</b> header
112 via mailto: links, try the <a
113 href="$link">mailto: link</a>
117 push @$arg, '/path/to/YOUR_REPLY';
118 $arg = ascii_html(join(" \\\n ", '', @$arg));
121 id=R><b>Reply instructions:</b>
123 You may reply publicly to <a
124 href=#t>this message</a> via plain-text email
125 using any one of the following methods:
127 * Save the following mbox file, import it into your mail client,
128 and $reply_to_all from there: <a
131 Avoid top-posting and favor interleaved quoting:
133 href="$p_url">$p_url</a>
135 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
136 switches of git-send-email(1):
141 href="$se_url">$se_url</a>
144 Be sure your reply has a <b>Subject:</b> header at the top and a blank line
145 before the message body.
151 my $refs = references($hdr);
155 sub fold_addresses ($) {
156 return $_[0] if length($_[0]) <= COLS;
157 # try to fold on commas after non-word chars before $lim chars,
158 # Try to get the "," preceded by ">" or ")", but avoid folding
159 # on the comma where somebody uses "Lastname, Firstname".
160 # We also try to keep the last and penultimate addresses in
161 # the list on the same line if possible, hence the extra \z
162 # Fall back to folding on spaces at $lim + 1 chars
163 my $lim = COLS - 8; # 8 = "\t" display width
164 my $too_long = $lim + 1;
165 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
167 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
170 .{$too_long,}?)(?:\s|\z)/xgo));
173 sub _hdr_names_html ($$) {
174 my ($hdr, $field) = @_;
175 my @vals = $hdr->header($field) or return '';
176 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
180 my ($nr, $singular, $plural) = @_;
181 return "0 $plural" if $nr == 0;
182 $nr == 1 ? "$nr $singular" : "$nr $plural";
185 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
186 # this is already inside a <pre>
188 my ($ctx, $eml) = @_;
189 my $smsg = delete $ctx->{smsg};
190 my $subj = delete $smsg->{subject};
191 my $mid_raw = $smsg->{mid};
192 my $id = id_compress($mid_raw, 1);
194 my $root_anchor = $ctx->{root_anchor} || '';
196 my $obfs_ibx = $ctx->{-obfs_ibx};
198 $subj = '(no subject)' if $subj eq '';
199 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
200 $subj = '<b>'.ascii_html($subj).'</b>';
201 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
202 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
204 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
206 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
208 # Deleting these fields saves about 400K as we iterate across 1K msgs
209 delete @$smsg{qw(ts blob)};
211 my $from = _hdr_names_html($eml, 'From');
212 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
213 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
214 my $upfx = $ctx->{-upfx};
215 my $mhref = $upfx . mid_href($mid_raw) . '/';
216 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
217 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
218 my $to = fold_addresses(_hdr_names_html($eml, 'To'));
219 my $cc = fold_addresses(_hdr_names_html($eml, 'Cc'));
220 my ($tlen, $clen) = (length($to), length($cc));
222 if (($tlen + $clen) > COLS) {
223 $to_cc .= ' To: '.$to."\n" if $tlen;
224 $to_cc .= ' Cc: '.$cc."\n" if $clen;
227 $to_cc .= ' To: '.$to;
228 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
230 $to_cc .= ' Cc: '.$cc if $clen;
234 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
237 my $mapping = $ctx->{mapping};
238 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($eml)))) {
239 my $href = $upfx . mid_href($irt) . '/';
240 my $html = ascii_html($irt);
241 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
245 # scan through all parts, looking for displayable text
246 $ctx->{mhref} = $mhref;
247 $ctx->{changed_href} = "#e$id"; # for diffstat "files? changed,"
249 $eml->each_part(\&add_text_body, $ctx, 1);
253 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
254 "<a\nhref=\"$mhref\">permalink</a>" .
255 " <a\nhref=\"${mhref}raw\">raw</a>" .
256 " <a\nhref=\"${mhref}#R\">reply</a>";
258 delete($ctx->{-qry}) and
259 $rv .= qq[ <a\nhref="${mhref}#related">related</a>];
262 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
263 $rv .= "\t[relevance $pct%]";
266 my $nested = 'nested';
270 $flat = "<b>$flat</b>";
272 $nested = "<b>$nested</b>";
274 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
275 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
276 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
281 # do we have more messages? start a new <pre> if so
282 $rv .= scalar(@{$ctx->{msgs}}) ? '</pre><hr><pre>' : '</pre>' if $hr;
286 sub pad_link ($$;$) {
287 my ($mid, $level, $s) = @_;
289 my $href = defined($mid) ?
290 ("<a\nhref=#r".id_compress($mid, 1).">($s)</a>\n") :
292 (' 'x19).indent_for($level).th_pfx($level).$href;
296 # my ($mapping, $mid) = @_;
297 ($_[0]->{$_[1] // \'bogus'} // [ "(?)\n" ])->[0];
301 my ($mid_raw, $irt, $id, $ctx) = @_;
303 my $mapping = $ctx->{mapping} or return $rv;
305 my $mid_map = $mapping->{$mid_raw};
307 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
308 my ($attr, $node, $idx, $level) = @$mid_map;
309 my $children = $node->{children};
310 my $nr_c = scalar @$children;
313 # delete saves about 200KB on a 1K message thread
314 if (my $refs = delete $node->{references}) {
315 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
317 my $irt_map = $mapping->{$$irt} if defined $$irt;
318 if (defined $irt_map) {
319 $siblings = $irt_map->[1]->{children};
320 $nr_s = scalar(@$siblings) - 1;
321 $rv .= $pad . $irt_map->[0];
323 my $prev = $siblings->[$idx - 1];
324 my $pmid = $prev->{mid};
326 my $s = ($idx - 1). ' preceding siblings ...';
327 $rv .= pad_link($pmid, $level, $s);
328 } elsif ($idx == 2) {
329 $rv .= $pad . _skel_hdr($mapping,
331 $siblings->[0]->{mid} : undef);
333 $rv .= $pad . _skel_hdr($mapping, $pmid);
336 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
337 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
338 $attr =~ s!\n\z!</b>\n!s;
339 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
340 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
343 my $cmid = $children->[0] ? $children->[0]->{mid} : undef;
344 $rv .= $pad . _skel_hdr($mapping, $cmid);
346 my $s = ($nr_c - 1). ' more replies';
347 $rv .= pad_link($cmid, $level + 1, $s);
348 } elsif (my $cn = $children->[1]) {
349 $rv .= $pad . _skel_hdr($mapping, $cn->{mid});
353 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
355 my $nmid = $next->{mid};
356 $rv .= $pad . _skel_hdr($mapping, $nmid);
357 my $nnext = $nr_s - $idx;
359 my $s = ($nnext - 1).' subsequent siblings';
360 $rv .= pad_link($nmid, $level, $s);
361 } elsif (my $nn = $siblings->[$idx + 2]) {
362 $rv .= $pad . _skel_hdr($mapping, $nn->{mid});
365 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
368 # non-recursive thread walker
369 sub walk_thread ($$$) {
370 my ($rootset, $ctx, $cb) = @_;
371 my @q = map { (0, $_, -1) } @$rootset;
373 my ($level, $node, $i) = splice(@q, 0, 3);
374 defined $node or next;
375 $cb->($ctx, $level, $node, $i) or return;
378 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
382 sub pre_thread { # walk_thread callback
383 my ($ctx, $level, $node, $idx) = @_;
384 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
385 skel_dump($ctx, $level, $node);
388 sub thread_eml_entry {
389 my ($ctx, $eml) = @_;
390 my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
391 $beg . '<pre>' . eml_entry($ctx, $eml) . '</pre>' . $end;
394 sub next_in_queue ($$) {
395 my ($q, $ghost_ok) = @_;
397 my ($level, $smsg) = splice(@$q, 0, 2);
399 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
400 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
405 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
406 my ($ctx, $eml) = @_;
407 return thread_eml_entry($ctx, $eml) if $eml;
408 return unless exists($ctx->{skel});
409 my $ghost_ok = $ctx->{nr}++;
411 my ($lvl, $smsg) = next_in_queue($ctx->{-queue}, $ghost_ok);
413 if (exists $smsg->{blob}) { # next message for cat-file
414 $ctx->{level} = $lvl;
415 if (!$ghost_ok) { # first non-ghost
416 $ctx->{-title_html} =
417 ascii_html($smsg->{subject});
418 $ctx->zmore($ctx->html_top);
422 # buffer the ghost entry and loop
423 $ctx->zmore(ghost_index_entry($ctx, $lvl, $smsg));
425 $ctx->zmore(join('', thread_adj_level($ctx, 0)));
426 $ctx->zmore(${delete($ctx->{skel})});
432 sub stream_thread ($$) {
433 my ($rootset, $ctx) = @_;
434 @{$ctx->{-queue}} = map { (0, $_) } @$rootset;
435 PublicInbox::WwwStream::aresponse($ctx, \&stream_thread_i);
438 # /$INBOX/$MSGID/t/ and /$INBOX/$MSGID/T/
441 $ctx->{-upfx} = '../../';
442 my $mid = $ctx->{mid};
443 my $ibx = $ctx->{ibx};
444 my ($nr, $msgs) = $ibx->over->get_thread($mid);
445 return missing_thread($ctx) if $nr == 0;
447 # link $INBOX_DIR/description text to "index_topics" view around
448 # the newest message in this thread
449 my $t = ts2str($ctx->{-t_max} = max(map { delete $_->{ts} } @$msgs));
450 my $t_fmt = fmt_ts($ctx->{-t_max});
452 my $skel = '<hr><pre>';
453 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
456 href="../../?t=$t">~$t_fmt UTC</a> | <a
457 href="../../">newest</a>]
460 $skel .= "<b\nid=t>Thread overview:</b> ";
461 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
462 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
463 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
464 $skel .= "-- links below jump to the message on this page --\n";
465 $ctx->{cur_level} = 0;
466 $ctx->{skel} = \$skel;
467 $ctx->{prev_attr} = '';
468 $ctx->{prev_level} = 0;
469 $ctx->{root_anchor} = 'm' . id_compress($mid, 1);
470 $ctx->{mapping} = {}; # mid -> [ header_summary, node, idx, level ]
471 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
474 my $rootset = thread_results($ctx, $msgs);
476 # reduce hash lookups in pre_thread->skel_dump
477 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
478 walk_thread($rootset, $ctx, \&pre_thread);
481 return stream_thread($rootset, $ctx) unless $ctx->{flat};
483 # flat display: lazy load the full message from smsg
484 $ctx->{msgs} = $msgs;
485 $ctx->{-html_tip} = '<pre>';
486 PublicInbox::WwwStream::aresponse($ctx, \&thread_html_i);
489 sub thread_html_i { # PublicInbox::WwwStream::getline callback
490 my ($ctx, $eml) = @_;
492 my $smsg = $ctx->{smsg};
493 if (exists $ctx->{-html_tip}) {
494 $ctx->{-title_html} = ascii_html($smsg->{subject});
495 $ctx->zmore($ctx->html_top);
497 return eml_entry($ctx, $eml);
499 while (my $smsg = shift @{$ctx->{msgs}}) {
500 return $smsg if exists($smsg->{blob});
502 my $skel = delete($ctx->{skel}) or return; # all done
508 sub multipart_text_as_html {
509 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
511 # scan through all parts, looking for displayable text
512 $_[0]->each_part(\&add_text_body, $_[1], 1);
515 sub submsg_hdr ($$) {
516 my ($ctx, $eml) = @_;
517 my $obfs_ibx = $ctx->{-obfs_ibx};
518 my $rv = $ctx->{obuf};
520 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
521 my @v = $eml->header($h);
523 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
530 sub attach_link ($$$$;$) {
531 my ($ctx, $ct, $p, $fn, $err) = @_;
532 my ($part, $depth, $idx) = @$p;
534 # Eml iteration clobbers multipart ->{bdy}, so do not offer
535 # downloads for 0-byte multipart attachments
536 return unless $part->{bdy};
538 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
539 my $size = length($part->body);
540 delete $part->{bdy}; # save memory
542 # hide attributes normally, unless we want to aid users in
543 # spotting MUA problems:
544 $ct =~ s/;.*// unless $err;
545 $ct = ascii_html($ct);
547 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
549 } elsif ($ct eq 'text/plain') {
554 my $rv = $ctx->{obuf};
555 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
558 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
561 $$rv .= "[-- Attachment #$idx: ";
562 my $ts = "Type: $ct, Size: $size bytes";
563 my $desc = $part->header('Content-Description') // $fn // '';
564 $desc = ascii_html($desc);
565 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
568 submsg_hdr($ctx, $part) if $part->{is_submsg};
573 sub add_text_body { # callback for each_part
575 my $upfx = $ctx->{mhref};
576 my $ibx = $ctx->{ibx};
577 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
578 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
579 my ($part, $depth, $idx) = @$p;
580 my $ct = $part->content_type || 'text/plain';
581 my $fn = $part->filename;
582 my ($s, $err) = msg_part_text($part, $ct);
583 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
585 my $rv = $ctx->{obuf};
586 if ($part->{is_submsg}) {
587 submsg_hdr($ctx, $part);
591 # makes no difference to browsers, and don't screw up filename
592 # link generation in diffs with the extra '%0D'
595 # will be escaped to `•' in HTML
596 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
598 # always support diff-highlighting, but we can't linkify hunk
599 # headers for solver unless some coderepo are configured:
601 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
602 # diffstat anchors do not link across attachments or messages,
603 # -apfx is just a stable prefix for making diffstat anchors
604 # linkable to the first diff hunk w/o crossing attachments
605 $idx =~ tr!.!/!; # compatibility with previous versions
606 $ctx->{-apfx} = $upfx . $idx;
608 # do attr => filename mappings for diffstats in git diffs:
609 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
611 delete $ctx->{-long_path};
613 # absolute URL (Atom feeds)
614 if ($ibx->{coderepo}) {
615 if (index($upfx, '//') >= 0) {
617 $spfx =~ s!/([^/]*)/\z!/!;
619 my $n_slash = $upfx =~ tr!/!/!;
622 } elsif ($n_slash == 1) {
624 } else { # nslash == 2
629 $ctx->{-spfx} = $spfx;
632 # split off quoted and unquoted blocks:
633 my @sections = PublicInbox::MsgIter::split_quotes($s);
634 undef $s; # free memory
635 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
636 # badly-encoded message with $err? tell the world about it!
637 attach_link($ctx, $ct, $p, $fn, $err);
640 delete $part->{bdy}; # save memory
641 foreach my $cur (@sections) {
643 # we use a <span> here to allow users to specify
644 # their own color for quoted text
645 $$rv .= qq(<span\nclass="q">);
646 $$rv .= $l->to_html($cur);
649 flush_diff($ctx, \$cur);
652 $$rv .= $l->to_html($cur);
654 undef $cur; # free memory
658 sub _msg_page_prepare_obuf {
659 my ($eml, $ctx) = @_;
660 my $have_over = !!$ctx->{ibx}->over;
661 my $obfs_ibx = $ctx->{-obfs_ibx};
662 $ctx->{obuf} = \(my $rv = '');
663 my $mids = mids_for_index($eml);
664 my $nr = $ctx->{nr}++;
665 if ($nr) { # unlikely
666 if ($ctx->{chash} eq content_hash($eml)) {
667 warn "W: BUG? @$mids not deduplicated properly\n";
671 "<pre>WARNING: multiple messages have this Message-ID\n</pre><pre>";
673 $ctx->{first_hdr} = $eml->header_obj;
674 $ctx->{chash} = content_hash($eml) if $ctx->{smsg}; # reused MID
675 $rv .= "<pre\nid=b>"; # anchor for body start
677 $ctx->{-upfx} = '../';
678 my @title; # (Subject[0], From[0])
679 for my $v ($eml->header('From')) {
680 my @n = PublicInbox::Address::names($v);
682 $title[1] //= ascii_html(join(', ', @n));
684 obfuscate_addrs($obfs_ibx, $v);
685 obfuscate_addrs($obfs_ibx, $title[1]);
687 $rv .= "From: $v\n" if $v ne '';
689 foreach my $h (qw(To Cc)) {
690 for my $v ($eml->header($h)) {
693 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
694 $rv .= "$h: $v\n" if $v ne '';
697 my @subj = $eml->header('Subject');
699 my $v = ascii_html(shift @subj);
700 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
702 $rv .= $have_over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
704 for $v (@subj) { # multi-Subject message :<
706 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
707 $rv .= "Subject: $v\n";
709 } else { # dummy anchor for thread skeleton at bottom of page
710 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $have_over;
711 $title[0] = '(no subject)';
713 for my $v ($eml->header('Date')) {
715 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
716 $rv .= qq{Date: $v\n};
718 # [thread overview] link is typically added after Date,
719 # but added after Subject, or even nothing.
721 chop $rv; # drop "\n", or noop if $rv eq ''
722 $rv .= qq{\t<a\nhref="#r">[thread overview]</a>\n};
724 if (!$nr) { # first (and only) message, common case
725 $ctx->{-title_html} = join(' - ', @title);
726 $rv = $ctx->html_top . $rv;
729 $ctx->{-linkify} //= PublicInbox::Linkify->new;
730 if (scalar(@$mids) == 1) { # common case
731 my $mhtml = ascii_html($mids->[0]);
732 $rv .= qq[Message-ID: <$mhtml> (<a href="raw">raw</a>)\n];
734 # X-Alt-Message-ID can happen if a message is injected from
735 # public-inbox-nntpd because of multiple Message-ID headers.
737 for my $h (qw(Message-ID X-Alt-Message-ID)) {
738 $s .= "$h: $_\n" for ($eml->header_raw($h));
740 $ctx->{-linkify}->linkify_mids('..', \$s, 1);
743 $rv .= _parent_headers($ctx, $eml);
748 qq(expand[<a\nhref="T/#u">flat</a>) .
749 qq(|<a\nhref="t/#u">nested</a>] ) .
750 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
751 qq(<a\nhref="t.atom">Atom feed</a>);
754 sub thread_skel ($$$) {
755 my ($skel, $ctx, $hdr) = @_;
756 my $mid = mids($hdr)->[0];
757 my $ibx = $ctx->{ibx};
758 my ($nr, $msgs) = $ibx->over->get_thread($mid);
759 my $parent = in_reply_to($hdr);
760 $$skel .= "\n<b>Thread overview: </b>";
762 if (defined $parent) {
763 $$skel .= SKEL_EXPAND."\n ";
764 $$skel .= ghost_parent('../', $parent) . "\n";
766 $$skel .= "<a\nid=r>[no followups]</a> ".
769 $ctx->{next_msg} = undef;
770 $ctx->{parent_msg} = $parent;
775 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
777 # nb: mutt only shows the first Subject in the index pane
778 # when multiple Subject: headers are present, so we follow suit:
779 my $subj = $hdr->header('Subject') // '';
780 $subj = '(no subject)' if $subj eq '';
782 $ctx->{prev_attr} = '';
783 $ctx->{prev_level} = 0;
784 $ctx->{skel} = $skel;
786 # reduce hash lookups in skel_dump
787 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
788 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
790 $ctx->{parent_msg} = $parent;
793 sub _parent_headers {
794 my ($ctx, $hdr) = @_;
795 my @irt = $hdr->header_raw('In-Reply-To');
799 $refs = references($hdr);
800 $irt[0] = pop(@$refs) if scalar @$refs;
802 $s .= "In-Reply-To: $_\n" for @irt;
804 # do not display References: if search is present,
805 # we show the thread skeleton at the bottom, instead.
806 if (!$ctx->{ibx}->over) {
807 $refs //= references($hdr);
808 $s .= 'References: <'.join(">\n\t<", @$refs).">\n" if @$refs;
810 $ctx->{-linkify}->linkify_mids('..', \$s); # escapes HTML
816 my ($ctx, $hdr) = @_;
818 my ($related, $skel);
819 my $qry = delete $ctx->{-qry};
820 if ($qry && $ctx->{ibx}->isrch) {
821 my $q = ''; # search for either ancestor or descendent patches
822 for (@{$qry->{dfpre}}, @{$qry->{dfpost}}) {
823 chop if length > 7; # include 1 abbrev "older" patches
826 chop $q; # omit trailing SP
827 local $Text::Wrap::columns = COLS;
828 local $Text::Wrap::huge = 'overflow';
829 $q = wrap('', '', $q);
830 my $rows = ($q =~ tr/\n/\n/) + 1;
835 ><pre>find likely ancestor, descendant, or conflicting patches for <a
836 href=#t>this message</a>:
837 <textarea name=q cols=${\COLS} rows=$rows>$q</textarea>
838 <input type=submit value=search
839 />\t(<a href=${upfx}_/text/help/#search>help</a>)</pre></form>
842 if ($ctx->{ibx}->over) {
843 my $t = ts2str($ctx->{-t_max});
844 my $t_fmt = fmt_ts($ctx->{-t_max});
845 my $fallback = $related ? "\t" : "<a id=related>\t</a>";
847 ${fallback}other threads:[<a
848 href="$upfx?t=$t">~$t_fmt UTC</a>|<a
849 href="$upfx">newest</a>]
852 thread_skel(\$skel, $ctx, $hdr);
857 if (my $n = $ctx->{next_msg}) {
859 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
862 my $par = $ctx->{parent_msg};
867 if (my $p = $ctx->{prev_msg}) {
868 $prev = mid_href($p);
869 if ($p && $par && $p eq $par) {
870 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
871 'rel=prev>prev parent</a>';
874 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
876 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
878 } elsif ($u) { # unlikely
879 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
881 ${$ctx->{obuf}} .= "<pre>$next $prev$parent ";
882 } else { # unindexed inboxes w/o over
883 ${$ctx->{obuf}} .= '<pre>';
884 $skel = qq( <a\nhref="$upfx">latest</a>);
886 ${$ctx->{obuf}} .= qq(<a\nhref="#R">reply</a>);
887 # $skel may be big for big threads, don't append it to obuf
888 $skel .= '</pre>' . ($related // '');
889 $ctx->zmore($skel .= msg_reply($ctx, $hdr)); # flushes obuf
893 my ($upfx, $mid) = @_;
895 my $href = mid_href($mid);
896 my $html = ascii_html($mid);
897 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
902 $level ? INDENT x ($level - 1) : '';
906 my ($ctx, $level, $node, $idx) = @_;
907 ++$ctx->{root_idx} if $level == 0;
908 if ($node->{mid} eq $ctx->{mid}) {
909 $ctx->{found_mid_at} = $ctx->{root_idx};
910 return 0; # stop iterating
915 sub strict_loose_note ($) {
918 " -- strict thread matches above, loose matches on Subject: below --\n";
920 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
922 " -- use mbox.gz link to download all $nr messages --\n";
928 my ($ctx, $msgs) = @_;
929 require PublicInbox::SearchThread;
930 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
932 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
933 # and preserved in the future. This bug is hidden by `sid' matches
934 # in get_thread, so we never noticed it until now. And even when
935 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
936 # bump since reindexing is expensive and users may not do it
938 # loose threading could've returned too many results,
939 # put the root the message we care about at the top:
940 my $mid = $ctx->{mid};
941 if (defined($mid) && scalar(@$rootset) > 1) {
942 $ctx->{root_idx} = -1;
943 my $nr = scalar @$msgs;
944 walk_thread($rootset, $ctx, \&find_mid_root);
945 my $idx = $ctx->{found_mid_at};
946 if (defined($idx) && $idx != 0) {
947 my $tip = splice(@$rootset, $idx, 1);
948 @$rootset = reverse @$rootset;
949 unshift @$rootset, $tip;
950 $ctx->{sl_note} = strict_loose_note($nr);
958 require PublicInbox::ExtMsg;
959 PublicInbox::ExtMsg::ext_msg($ctx);
963 my ($prev_subj, $subj, $val) = @_;
965 my $omit; # '"' denotes identical text omitted
966 my (@prev_pop, @curr_pop);
967 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
968 push(@prev_pop, pop(@$prev_subj));
969 push(@curr_pop, pop(@$subj));
972 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
973 if (scalar(@curr_pop) == 1) {
975 push @$prev_subj, @prev_pop;
976 push @$subj, @curr_pop;
981 sub skel_dump { # walk_thread callback
982 my ($ctx, $level, $smsg) = @_;
983 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
985 my $skel = $ctx->{skel};
986 my $cur = $ctx->{cur};
987 my $mid = $smsg->{mid};
989 if ($level == 0 && $ctx->{skel_dump_roots}++) {
990 $$skel .= delete($ctx->{sl_note}) || '';
993 my $f = ascii_html(delete $smsg->{from_name});
994 my $obfs_ibx = $ctx->{-obfs_ibx};
995 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
997 my $d = fmt_ts($smsg->{ds});
998 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
999 if (exists $ctx->{searchview}) {
1000 if (defined(my $pct = $smsg->{pct})) {
1001 $d .= (sprintf(' % 2u', $pct) . '%');
1007 $d .= ' ' . indent_for($level) . th_pfx($level);
1009 $ctx->{first_level} ||= $level;
1011 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
1012 $ctx->{prev_attr} = $attr;
1014 $ctx->{prev_level} = $level;
1019 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
1020 "$attr [this message]</a></b>\n";
1023 $ctx->{prev_msg} = $mid;
1026 $ctx->{next_msg} ||= $mid;
1029 # Subject is never undef, this mail was loaded from
1030 # our Xapian which would've resulted in '' if it were
1031 # really missing (and Filter rejects empty subjects)
1032 my @subj = split(/ /, subject_normalized($smsg->{subject}));
1033 # remove common suffixes from the subject if it matches the previous,
1034 # so we do not show redundant text at the end.
1035 my $prev_subj = $ctx->{prev_subj} || [];
1036 $ctx->{prev_subj} = [ @subj ];
1037 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
1040 my $subj = join(' ', @subj);
1041 $subj = ascii_html($subj);
1042 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1043 $end = "$subj</a> $omit$f\n"
1049 my $mapping = $unmatched ? undef : $ctx->{mapping};
1051 my $map = $mapping->{$mid};
1052 $id = id_compress($mid, 1);
1054 $map->[0] = "$d<a\nhref=\"$m\">$end";
1057 $m = $ctx->{-upfx}.mid_href($mid).'/';
1059 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1064 my ($ctx, $level, $node) = @_;
1066 my $mid = $node->{mid};
1067 my $d = ' [not found] ';
1068 $d .= ' ' if exists $ctx->{searchview};
1069 $d .= indent_for($level) . th_pfx($level);
1070 my $upfx = $ctx->{-upfx};
1071 my $href = $upfx . mid_href($mid) . '/';
1072 my $html = ascii_html($mid);
1074 my $mapping = $ctx->{mapping};
1075 my $map = $mapping->{$mid} if $mapping;
1077 my $id = id_compress($mid, 1);
1078 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1079 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1081 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1083 ${$ctx->{skel}} .= $d;
1089 (eval { $a->topmost->{ds} } || 0) <=>
1090 (eval { $b->topmost->{ds} } || 0)
1094 # accumulate recent topics if search is supported
1095 # returns 200 if done, 404 if not
1096 sub acc_topic { # walk_thread callback
1097 my ($ctx, $level, $smsg) = @_;
1098 my $mid = $smsg->{mid};
1099 my $has_blob = $smsg->{blob} // do {
1100 if (my $by_mid = $ctx->{ibx}->smsg_by_mid($mid)) {
1101 %$smsg = (%$smsg, %$by_mid);
1106 my $subj = subject_normalized($smsg->{subject});
1107 $subj = '(no subject)' if $subj eq '';
1108 my $ds = $smsg->{ds};
1109 if ($level == 0) { # new, top-level topic
1110 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1111 $ctx->{-cur_topic} = $topic;
1112 push @{$ctx->{order}}, $topic;
1116 # continue existing topic
1117 my $topic = $ctx->{-cur_topic}; # should never be undef
1118 $topic->[0] = $ds if $ds > $topic->[0];
1119 $topic->[1]++; # bump N+ message counter
1120 my $seen = $topic->[2];
1121 if (scalar(@$topic) == 3) { # parent was a ghost
1122 push @$topic, $subj;
1123 } elsif (!defined($seen->{$subj})) {
1124 push @$topic, $level, $subj; # @extra messages
1126 $seen->{$subj} = $mid; # latest for subject
1127 } else { # ghost message
1128 return 1 if $level != 0; # ignore child ghosts
1129 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1130 push @{$ctx->{order}}, $topic;
1137 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1139 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1144 my $obfs_ibx = $ctx->{ibx}->{obfuscate} ? $ctx->{ibx} : undef;
1145 if (my $note = delete $ctx->{t_note}) {
1146 push @out, $note; # "messages from ... to ..."
1148 # sort by recency, this allows new posts to "bump" old topics...
1149 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1150 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1152 next unless defined $top_subj; # ghost topic
1153 my $mid = delete $seen->{$top_subj};
1154 my $href = mid_href($mid);
1155 my $prev_subj = [ split(/ /, $top_subj) ];
1156 $top_subj = ascii_html($top_subj);
1159 # $n isn't the total number of posts on the topic,
1160 # just the number of posts in the current results window
1164 $anchor = '#u'; # top of only message
1166 $n = " ($n+ messages)";
1167 $anchor = '#t'; # thread skeleton
1170 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1173 my $level = shift @extra;
1174 my $subj = shift @extra; # already normalized
1175 $mid = delete $seen->{$subj};
1176 my @subj = split(/ /, $subj);
1177 my @next_prev = @subj; # full copy
1178 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1179 $prev_subj = \@next_prev;
1180 $subj = join(' ', @subj);
1181 $subj = ascii_html($subj);
1182 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1183 $href = mid_href($mid);
1184 $s .= indent_for($level) . TCHILD;
1185 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1189 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1194 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1195 timegm($ss || 0, $mm || 0, $hh || 0, $dd, $mon - 1, $yyyy);
1198 sub pagination_footer ($$) {
1199 my ($ctx, $latest) = @_;
1200 my $next = $ctx->{next_page} || '';
1201 my $prev = $ctx->{prev_page} || '';
1202 if ($prev) { # aligned padding for: 'next (older) | '
1203 $next = $next ? "$next | " : ' | ';
1204 $prev .= qq[ | <a\nhref="$latest">latest</a>];
1206 ($next || $prev) ? "<hr><pre id=nav>page: $next$prev</pre>" : '';
1209 sub paginate_recent ($$) {
1210 my ($ctx, $lim) = @_;
1211 my $t = $ctx->{qp}->{t} || '';
1212 my $opts = { limit => $lim };
1213 my ($after, $before);
1215 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1216 # if only $after exists "YYYYMMDD.." because "." could be skipped
1217 # if interpreted as an end-of-sentence
1218 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1219 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1221 my $msgs = $ctx->{ibx}->over->recent($opts, $after, $before);
1222 if (defined($after) && scalar(@$msgs) < $lim) {
1223 $after = $before = undef;
1224 $msgs = $ctx->{ibx}->over->recent($opts);
1226 my $more = scalar(@$msgs) == $lim;
1227 my ($newest, $oldest);
1229 $newest = $msgs->[0]->{ts};
1230 $oldest = $msgs->[-1]->{ts};
1231 # if we only had $after, our SQL query in ->recent ordered
1232 if ($newest < $oldest) {
1233 ($oldest, $newest) = ($newest, $oldest);
1234 $more = undef if defined($after) && $after < $oldest;
1236 if (defined($after // $before)) {
1237 my $n = strftime('%Y-%m-%d %H:%M:%S', gmtime($newest));
1238 my $o = strftime('%Y-%m-%d %H:%M:%S', gmtime($oldest));
1239 $ctx->{t_note} = <<EOM;
1240 messages from $o to $n UTC [<a href="#nav">more...</a>]
1242 my $s = ts2str($newest);
1243 $ctx->{prev_page} = qq[<a\nhref="?t=$s-"\nrel=prev>] .
1247 if (defined($oldest) && $more) {
1248 my $s = ts2str($oldest);
1249 $ctx->{next_page} = qq[<a\nhref="?t=$s"\nrel=next>] .
1255 # GET /$INBOX - top-level inbox view for indexed inboxes
1258 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1259 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic) if @$msgs;
1260 html_oneshot($ctx, dump_topics($ctx), pagination_footer($ctx, '.'));
1263 sub thread_adj_level {
1264 my ($ctx, $level) = @_;
1266 my $max = $ctx->{cur_level};
1268 return ('', '') if $max == 0; # flat output
1270 # reset existing lists
1271 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1272 $ctx->{cur_level} = 0;
1274 } elsif ($level == $max) { # continue existing list
1276 } elsif ($level < $max) {
1277 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1278 $ctx->{cur_level} = $level;
1279 ("$beg<li>", '</li>');
1280 } else { # ($level > $max) # start a new level
1281 $ctx->{cur_level} = $level;
1282 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1287 sub ghost_index_entry {
1288 my ($ctx, $level, $node) = @_;
1289 my ($beg, $end) = thread_adj_level($ctx, $level);
1290 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid} // '?')