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 $eml->each_part(\&add_text_body, $ctx, 1);
43 $ctx->zmore('</pre><hr>');
45 html_footer($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 $eml->each_part(\&add_text_body, $ctx, 1);
61 $ctx->zmore('</pre><hr>');
63 html_footer($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); # expensive
250 $ctx->zmore; # TODO: remove once add_text_body is updated
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 ($$;$) {
287 my ($mid, $level, $s) = @_;
289 my $href = defined($mid) ?
290 ("<a\nhref=#r".id_compress($mid, 1).">($s)</a>\n") :
292 (' 'x19).indent_for($level).th_pfx($level).$href;
296 # my ($mapping, $mid) = @_;
297 ($_[0]->{$_[1] // \'bogus'} // [ "(?)\n" ])->[0];
301 my ($mid_raw, $irt, $id, $ctx) = @_;
303 my $mapping = $ctx->{mapping} or return $rv;
305 my $mid_map = $mapping->{$mid_raw} //
306 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
307 my ($attr, $node, $idx, $level) = @$mid_map;
308 my $children = $node->{children};
309 my $nr_c = scalar @$children;
312 # delete saves about 200KB on a 1K message thread
313 if (my $refs = delete $node->{references}) {
314 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
316 my $irt_map = $mapping->{$$irt} if defined $$irt;
317 if (defined $irt_map) {
318 $siblings = $irt_map->[1]->{children};
319 $nr_s = scalar(@$siblings) - 1;
320 $rv .= $pad . $irt_map->[0];
322 my $prev = $siblings->[$idx - 1];
323 my $pmid = $prev->{mid};
325 my $s = ($idx - 1). ' preceding siblings ...';
326 $rv .= pad_link($pmid, $level, $s);
327 } elsif ($idx == 2) {
328 $rv .= $pad . _skel_hdr($mapping,
330 $siblings->[0]->{mid} : undef);
332 $rv .= $pad . _skel_hdr($mapping, $pmid);
335 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
336 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
337 chop $attr; # remove "\n"
338 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
339 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
340 $rv .= "<b>@ $attr</b>\n";
342 my $cmid = $children->[0] ? $children->[0]->{mid} : undef;
343 $rv .= $pad . _skel_hdr($mapping, $cmid);
345 my $s = ($nr_c - 1). ' more replies';
346 $rv .= pad_link($cmid, $level + 1, $s);
347 } elsif (my $cn = $children->[1]) {
348 $rv .= $pad . _skel_hdr($mapping, $cn->{mid});
352 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
354 my $nmid = $next->{mid};
355 $rv .= $pad . _skel_hdr($mapping, $nmid);
356 my $nnext = $nr_s - $idx;
358 my $s = ($nnext - 1).' subsequent siblings';
359 $rv .= pad_link($nmid, $level, $s);
360 } elsif (my $nn = $siblings->[$idx + 2]) {
361 $rv .= $pad . _skel_hdr($mapping, $nn->{mid});
364 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
367 # non-recursive thread walker
368 sub walk_thread ($$$) {
369 my ($rootset, $ctx, $cb) = @_;
370 my @q = map { (0, $_, -1) } @$rootset;
372 my ($level, $node, $i) = splice(@q, 0, 3);
373 defined $node or next;
374 $cb->($ctx, $level, $node, $i) or return;
377 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
381 sub pre_thread { # walk_thread callback
382 my ($ctx, $level, $node, $idx) = @_;
383 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
384 skel_dump($ctx, $level, $node);
387 sub thread_eml_entry {
388 my ($ctx, $eml) = @_;
389 my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
390 $ctx->zmore($beg.'<pre>');
391 eml_entry($ctx, $eml) . '</pre>' . $end;
394 sub next_in_queue ($$) {
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 submsg_hdr ($$) {
509 my ($ctx, $eml) = @_;
511 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
512 $s .= "$h: $_\n" for $eml->header($h);
514 obfuscate_addrs($ctx->{-obfs_ibx}, $s) if $ctx->{-obfs_ibx};
518 sub attach_link ($$$$;$) {
519 my ($ctx, $ct, $p, $fn, $err) = @_;
520 my ($part, $depth, $idx) = @$p;
522 # Eml iteration clobbers multipart ->{bdy}, so do not offer
523 # downloads for 0-byte multipart attachments
524 return unless $part->{bdy};
526 my $size = length($part->body);
527 delete $part->{bdy}; # save memory
529 # hide attributes normally, unless we want to aid users in
530 # spotting MUA problems:
531 $ct =~ s/;.*// unless $err;
532 $ct = ascii_html($ct);
534 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
536 } elsif ($ct eq 'text/plain') {
541 my $rv = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
542 $rv .= qq(<a\nhref="$ctx->{mhref}$idx-$sfn">);
543 $rv .= <<EOF if $err;
544 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
546 $rv .= "[-- Attachment #$idx: ";
547 my $desc = $part->header('Content-Description') // $fn // '';
548 $rv .= ascii_html($desc)." --]\n[-- " if $desc ne '';
549 $rv .= "Type: $ct, Size: $size bytes --]</a>\n";
550 $rv .= submsg_hdr($ctx, $part) if $part->{is_submsg};
554 sub add_text_body { # callback for each_part
556 my $upfx = $ctx->{mhref};
557 my $ibx = $ctx->{ibx};
558 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
559 # $p - from each_part: [ Email::MIME-like, depth, $idx ]
560 my ($part, $depth, $idx) = @$p;
561 my $ct = $part->content_type || 'text/plain';
562 my $fn = $part->filename;
563 my $rv = $ctx->{obuf};
564 my ($s, $err) = msg_part_text($part, $ct);
565 $s // return $$rv .= (attach_link($ctx, $ct, $p, $fn) // '');
566 if ($part->{is_submsg}) {
567 $$rv .= submsg_hdr($ctx, $part);
571 # makes no difference to browsers, and don't screw up filename
572 # link generation in diffs with the extra '%0D'
575 # will be escaped to `•' in HTML
576 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
578 # always support diff-highlighting, but we can't linkify hunk
579 # headers for solver unless some coderepo are configured:
581 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
582 # diffstat anchors do not link across attachments or messages,
583 # -apfx is just a stable prefix for making diffstat anchors
584 # linkable to the first diff hunk w/o crossing attachments
585 $idx =~ tr!.!/!; # compatibility with previous versions
586 $ctx->{-apfx} = $upfx . $idx;
588 # do attr => filename mappings for diffstats in git diffs:
589 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
591 delete $ctx->{-long_path};
593 # absolute URL (Atom feeds)
594 if ($ibx->{coderepo}) {
595 if (index($upfx, '//') >= 0) {
597 $spfx =~ s!/([^/]*)/\z!/!;
599 my $n_slash = $upfx =~ tr!/!/!;
602 } elsif ($n_slash == 1) {
604 } else { # nslash == 2
609 $ctx->{-spfx} = $spfx;
612 # split off quoted and unquoted blocks:
613 my @sections = PublicInbox::MsgIter::split_quotes($s);
614 undef $s; # free memory
615 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
616 # badly-encoded message with $err? tell the world about it!
617 $$rv .= attach_link($ctx, $ct, $p, $fn, $err);
620 delete $part->{bdy}; # save memory
621 foreach my $cur (@sections) {
623 # we use a <span> here to allow users to specify
624 # their own color for quoted text
625 $$rv .= qq(<span\nclass="q">);
626 $$rv .= $l->to_html($cur);
629 flush_diff($ctx, \$cur);
632 $$rv .= $l->to_html($cur);
634 undef $cur; # free memory
638 sub _msg_page_prepare_obuf {
639 my ($eml, $ctx) = @_;
640 my $have_over = !!$ctx->{ibx}->over;
641 my $obfs_ibx = $ctx->{-obfs_ibx};
642 $ctx->{obuf} = \(my $rv = '');
643 my $mids = mids_for_index($eml);
644 my $nr = $ctx->{nr}++;
645 if ($nr) { # unlikely
646 if ($ctx->{chash} eq content_hash($eml)) {
647 warn "W: BUG? @$mids not deduplicated properly\n";
651 "<pre>WARNING: multiple messages have this Message-ID\n</pre><pre>";
653 $ctx->{first_hdr} = $eml->header_obj;
654 $ctx->{chash} = content_hash($eml) if $ctx->{smsg}; # reused MID
655 $rv .= "<pre\nid=b>"; # anchor for body start
657 $ctx->{-upfx} = '../';
658 my @title; # (Subject[0], From[0])
659 for my $v ($eml->header('From')) {
660 my @n = PublicInbox::Address::names($v);
662 $title[1] //= ascii_html(join(', ', @n));
664 obfuscate_addrs($obfs_ibx, $v);
665 obfuscate_addrs($obfs_ibx, $title[1]);
667 $rv .= "From: $v\n" if $v ne '';
669 foreach my $h (qw(To Cc)) {
670 for my $v ($eml->header($h)) {
673 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
674 $rv .= "$h: $v\n" if $v ne '';
677 my @subj = $eml->header('Subject');
679 my $v = ascii_html(shift @subj);
680 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
682 $rv .= $have_over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
684 for $v (@subj) { # multi-Subject message :<
686 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
687 $rv .= "Subject: $v\n";
689 } else { # dummy anchor for thread skeleton at bottom of page
690 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $have_over;
691 $title[0] = '(no subject)';
693 for my $v ($eml->header('Date')) {
695 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
696 $rv .= qq{Date: $v\n};
698 # [thread overview] link is typically added after Date,
699 # but added after Subject, or even nothing.
701 chop $rv; # drop "\n", or noop if $rv eq ''
702 $rv .= qq{\t<a\nhref="#r">[thread overview]</a>\n};
704 if (!$nr) { # first (and only) message, common case
705 $ctx->{-title_html} = join(' - ', @title);
706 $rv = $ctx->html_top . $rv;
709 $ctx->{-linkify} //= PublicInbox::Linkify->new;
710 if (scalar(@$mids) == 1) { # common case
711 my $mhtml = ascii_html($mids->[0]);
712 $rv .= qq[Message-ID: <$mhtml> (<a href="raw">raw</a>)\n];
714 # X-Alt-Message-ID can happen if a message is injected from
715 # public-inbox-nntpd because of multiple Message-ID headers.
717 for my $h (qw(Message-ID X-Alt-Message-ID)) {
718 $s .= "$h: $_\n" for ($eml->header_raw($h));
720 $ctx->{-linkify}->linkify_mids('..', \$s, 1);
723 $rv .= _parent_headers($ctx, $eml);
728 qq(expand[<a\nhref="T/#u">flat</a>) .
729 qq(|<a\nhref="t/#u">nested</a>] ) .
730 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
731 qq(<a\nhref="t.atom">Atom feed</a>);
734 sub thread_skel ($$$) {
735 my ($skel, $ctx, $hdr) = @_;
736 my $mid = mids($hdr)->[0];
737 my $ibx = $ctx->{ibx};
738 my ($nr, $msgs) = $ibx->over->get_thread($mid);
739 my $parent = in_reply_to($hdr);
740 $$skel .= "\n<b>Thread overview: </b>";
742 if (defined $parent) {
743 $$skel .= SKEL_EXPAND."\n ";
744 $$skel .= ghost_parent('../', $parent) . "\n";
746 $$skel .= "<a\nid=r>[no followups]</a> ".
749 $ctx->{next_msg} = undef;
750 $ctx->{parent_msg} = $parent;
755 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
757 # nb: mutt only shows the first Subject in the index pane
758 # when multiple Subject: headers are present, so we follow suit:
759 my $subj = $hdr->header('Subject') // '';
760 $subj = '(no subject)' if $subj eq '';
762 $ctx->{prev_attr} = '';
763 $ctx->{prev_level} = 0;
764 $ctx->{skel} = $skel;
766 # reduce hash lookups in skel_dump
767 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
768 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
770 $ctx->{parent_msg} = $parent;
773 sub _parent_headers {
774 my ($ctx, $hdr) = @_;
775 my @irt = $hdr->header_raw('In-Reply-To');
779 $refs = references($hdr);
780 $irt[0] = pop(@$refs) if scalar @$refs;
782 $s .= "In-Reply-To: $_\n" for @irt;
784 # do not display References: if search is present,
785 # we show the thread skeleton at the bottom, instead.
786 if (!$ctx->{ibx}->over) {
787 $refs //= references($hdr);
788 $s .= 'References: <'.join(">\n\t<", @$refs).">\n" if @$refs;
790 $ctx->{-linkify}->linkify_mids('..', \$s); # escapes HTML
796 my ($ctx, $hdr) = @_;
798 my ($related, $skel);
799 my $qry = delete $ctx->{-qry};
800 if ($qry && $ctx->{ibx}->isrch) {
801 my $q = ''; # search for either ancestor or descendent patches
802 for (@{$qry->{dfpre}}, @{$qry->{dfpost}}) {
803 chop if length > 7; # include 1 abbrev "older" patches
806 chop $q; # omit trailing SP
807 local $Text::Wrap::columns = COLS;
808 local $Text::Wrap::huge = 'overflow';
809 $q = wrap('', '', $q);
810 my $rows = ($q =~ tr/\n/\n/) + 1;
815 ><pre>find likely ancestor, descendant, or conflicting patches for <a
816 href=#t>this message</a>:
817 <textarea name=q cols=${\COLS} rows=$rows>$q</textarea>
818 <input type=submit value=search
819 />\t(<a href=${upfx}_/text/help/#search>help</a>)</pre></form>
822 if ($ctx->{ibx}->over) {
823 my $t = ts2str($ctx->{-t_max});
824 my $t_fmt = fmt_ts($ctx->{-t_max});
825 my $fallback = $related ? "\t" : "<a id=related>\t</a>";
827 ${fallback}other threads:[<a
828 href="$upfx?t=$t">~$t_fmt UTC</a>|<a
829 href="$upfx">newest</a>]
832 thread_skel(\$skel, $ctx, $hdr);
837 if (my $n = $ctx->{next_msg}) {
839 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
842 my $par = $ctx->{parent_msg};
847 if (my $p = $ctx->{prev_msg}) {
848 $prev = mid_href($p);
849 if ($p && $par && $p eq $par) {
850 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
851 'rel=prev>prev parent</a>';
854 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
856 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
858 } elsif ($u) { # unlikely
859 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
861 ${$ctx->{obuf}} .= "<pre>$next $prev$parent ";
862 } else { # unindexed inboxes w/o over
863 ${$ctx->{obuf}} .= '<pre>';
864 $skel = qq( <a\nhref="$upfx">latest</a>);
866 ${$ctx->{obuf}} .= qq(<a\nhref="#R">reply</a>);
867 # $skel may be big for big threads, don't append it to obuf
868 $skel .= '</pre>' . ($related // '');
869 $ctx->zmore($skel .= msg_reply($ctx, $hdr)); # flushes obuf
873 my ($upfx, $mid) = @_;
875 my $href = mid_href($mid);
876 my $html = ascii_html($mid);
877 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
882 $level ? INDENT x ($level - 1) : '';
886 my ($ctx, $level, $node, $idx) = @_;
887 ++$ctx->{root_idx} if $level == 0;
888 if ($node->{mid} eq $ctx->{mid}) {
889 $ctx->{found_mid_at} = $ctx->{root_idx};
890 return 0; # stop iterating
895 sub strict_loose_note ($) {
898 " -- strict thread matches above, loose matches on Subject: below --\n";
900 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
902 " -- use mbox.gz link to download all $nr messages --\n";
908 my ($ctx, $msgs) = @_;
909 require PublicInbox::SearchThread;
910 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
912 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
913 # and preserved in the future. This bug is hidden by `sid' matches
914 # in get_thread, so we never noticed it until now. And even when
915 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
916 # bump since reindexing is expensive and users may not do it
918 # loose threading could've returned too many results,
919 # put the root the message we care about at the top:
920 my $mid = $ctx->{mid};
921 if (defined($mid) && scalar(@$rootset) > 1) {
922 $ctx->{root_idx} = -1;
923 my $nr = scalar @$msgs;
924 walk_thread($rootset, $ctx, \&find_mid_root);
925 my $idx = $ctx->{found_mid_at};
926 if (defined($idx) && $idx != 0) {
927 my $tip = splice(@$rootset, $idx, 1);
928 @$rootset = reverse @$rootset;
929 unshift @$rootset, $tip;
930 $ctx->{sl_note} = strict_loose_note($nr);
938 require PublicInbox::ExtMsg;
939 PublicInbox::ExtMsg::ext_msg($ctx);
943 my ($prev_subj, $subj, $val) = @_;
945 my $omit; # '"' denotes identical text omitted
946 my (@prev_pop, @curr_pop);
947 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
948 push(@prev_pop, pop(@$prev_subj));
949 push(@curr_pop, pop(@$subj));
952 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
953 if (scalar(@curr_pop) == 1) {
955 push @$prev_subj, @prev_pop;
956 push @$subj, @curr_pop;
961 sub skel_dump { # walk_thread callback
962 my ($ctx, $level, $smsg) = @_;
963 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
965 my $skel = $ctx->{skel};
966 my $cur = $ctx->{cur};
967 my $mid = $smsg->{mid};
969 if ($level == 0 && $ctx->{skel_dump_roots}++) {
970 $$skel .= delete($ctx->{sl_note}) || '';
973 my $f = ascii_html(delete $smsg->{from_name});
974 my $obfs_ibx = $ctx->{-obfs_ibx};
975 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
977 my $d = fmt_ts($smsg->{ds});
978 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
979 if (exists $ctx->{searchview}) {
980 if (defined(my $pct = $smsg->{pct})) {
981 $d .= (sprintf(' % 2u', $pct) . '%');
987 $d .= ' ' . indent_for($level) . th_pfx($level);
989 $ctx->{first_level} ||= $level;
991 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
992 $ctx->{prev_attr} = $attr;
994 $ctx->{prev_level} = $level;
999 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
1000 "$attr [this message]</a></b>\n";
1003 $ctx->{prev_msg} = $mid;
1006 $ctx->{next_msg} ||= $mid;
1009 # Subject is never undef, this mail was loaded from
1010 # our Xapian which would've resulted in '' if it were
1011 # really missing (and Filter rejects empty subjects)
1012 my @subj = split(/ /, subject_normalized($smsg->{subject}));
1013 # remove common suffixes from the subject if it matches the previous,
1014 # so we do not show redundant text at the end.
1015 my $prev_subj = $ctx->{prev_subj} || [];
1016 $ctx->{prev_subj} = [ @subj ];
1017 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
1020 my $subj = join(' ', @subj);
1021 $subj = ascii_html($subj);
1022 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1023 $end = "$subj</a> $omit$f\n"
1029 my $mapping = $unmatched ? undef : $ctx->{mapping};
1031 my $map = $mapping->{$mid};
1032 $id = id_compress($mid, 1);
1034 $map->[0] = "$d<a\nhref=\"$m\">$end";
1037 $m = $ctx->{-upfx}.mid_href($mid).'/';
1039 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1044 my ($ctx, $level, $node) = @_;
1046 my $mid = $node->{mid};
1047 my $d = ' [not found] ';
1048 $d .= ' ' if exists $ctx->{searchview};
1049 $d .= indent_for($level) . th_pfx($level);
1050 my $upfx = $ctx->{-upfx};
1051 my $href = $upfx . mid_href($mid) . '/';
1052 my $html = ascii_html($mid);
1054 my $mapping = $ctx->{mapping};
1055 my $map = $mapping->{$mid} if $mapping;
1057 my $id = id_compress($mid, 1);
1058 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1059 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1061 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1063 ${$ctx->{skel}} .= $d;
1069 (eval { $a->topmost->{ds} } || 0) <=>
1070 (eval { $b->topmost->{ds} } || 0)
1074 # accumulate recent topics if search is supported
1075 # returns 200 if done, 404 if not
1076 sub acc_topic { # walk_thread callback
1077 my ($ctx, $level, $smsg) = @_;
1078 my $mid = $smsg->{mid};
1079 my $has_blob = $smsg->{blob} // do {
1080 if (my $by_mid = $ctx->{ibx}->smsg_by_mid($mid)) {
1081 %$smsg = (%$smsg, %$by_mid);
1086 my $subj = subject_normalized($smsg->{subject});
1087 $subj = '(no subject)' if $subj eq '';
1088 my $ds = $smsg->{ds};
1089 if ($level == 0) { # new, top-level topic
1090 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1091 $ctx->{-cur_topic} = $topic;
1092 push @{$ctx->{order}}, $topic;
1096 # continue existing topic
1097 my $topic = $ctx->{-cur_topic}; # should never be undef
1098 $topic->[0] = $ds if $ds > $topic->[0];
1099 $topic->[1]++; # bump N+ message counter
1100 my $seen = $topic->[2];
1101 if (scalar(@$topic) == 3) { # parent was a ghost
1102 push @$topic, $subj;
1103 } elsif (!defined($seen->{$subj})) {
1104 push @$topic, $level, $subj; # @extra messages
1106 $seen->{$subj} = $mid; # latest for subject
1107 } else { # ghost message
1108 return 1 if $level != 0; # ignore child ghosts
1109 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1110 push @{$ctx->{order}}, $topic;
1117 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1119 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1124 my $obfs_ibx = $ctx->{ibx}->{obfuscate} ? $ctx->{ibx} : undef;
1125 if (my $note = delete $ctx->{t_note}) {
1126 push @out, $note; # "messages from ... to ..."
1128 # sort by recency, this allows new posts to "bump" old topics...
1129 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1130 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1132 next unless defined $top_subj; # ghost topic
1133 my $mid = delete $seen->{$top_subj};
1134 my $href = mid_href($mid);
1135 my $prev_subj = [ split(/ /, $top_subj) ];
1136 $top_subj = ascii_html($top_subj);
1139 # $n isn't the total number of posts on the topic,
1140 # just the number of posts in the current results window
1144 $anchor = '#u'; # top of only message
1146 $n = " ($n+ messages)";
1147 $anchor = '#t'; # thread skeleton
1150 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1153 my $level = shift @extra;
1154 my $subj = shift @extra; # already normalized
1155 $mid = delete $seen->{$subj};
1156 my @subj = split(/ /, $subj);
1157 my @next_prev = @subj; # full copy
1158 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1159 $prev_subj = \@next_prev;
1160 $subj = join(' ', @subj);
1161 $subj = ascii_html($subj);
1162 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1163 $href = mid_href($mid);
1164 $s .= indent_for($level) . TCHILD;
1165 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1169 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1174 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1175 timegm($ss || 0, $mm || 0, $hh || 0, $dd, $mon - 1, $yyyy);
1178 sub pagination_footer ($$) {
1179 my ($ctx, $latest) = @_;
1180 my $next = $ctx->{next_page} || '';
1181 my $prev = $ctx->{prev_page} || '';
1182 if ($prev) { # aligned padding for: 'next (older) | '
1183 $next = $next ? "$next | " : ' | ';
1184 $prev .= qq[ | <a\nhref="$latest">latest</a>];
1186 ($next || $prev) ? "<hr><pre id=nav>page: $next$prev</pre>" : '';
1189 sub paginate_recent ($$) {
1190 my ($ctx, $lim) = @_;
1191 my $t = $ctx->{qp}->{t} || '';
1192 my $opts = { limit => $lim };
1193 my ($after, $before);
1195 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1196 # if only $after exists "YYYYMMDD.." because "." could be skipped
1197 # if interpreted as an end-of-sentence
1198 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1199 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1201 my $msgs = $ctx->{ibx}->over->recent($opts, $after, $before);
1202 if (defined($after) && scalar(@$msgs) < $lim) {
1203 $after = $before = undef;
1204 $msgs = $ctx->{ibx}->over->recent($opts);
1206 my $more = scalar(@$msgs) == $lim;
1207 my ($newest, $oldest);
1209 $newest = $msgs->[0]->{ts};
1210 $oldest = $msgs->[-1]->{ts};
1211 # if we only had $after, our SQL query in ->recent ordered
1212 if ($newest < $oldest) {
1213 ($oldest, $newest) = ($newest, $oldest);
1214 $more = undef if defined($after) && $after < $oldest;
1216 if (defined($after // $before)) {
1217 my $n = strftime('%Y-%m-%d %H:%M:%S', gmtime($newest));
1218 my $o = strftime('%Y-%m-%d %H:%M:%S', gmtime($oldest));
1219 $ctx->{t_note} = <<EOM;
1220 messages from $o to $n UTC [<a href="#nav">more...</a>]
1222 my $s = ts2str($newest);
1223 $ctx->{prev_page} = qq[<a\nhref="?t=$s-"\nrel=prev>] .
1227 if (defined($oldest) && $more) {
1228 my $s = ts2str($oldest);
1229 $ctx->{next_page} = qq[<a\nhref="?t=$s"\nrel=next>] .
1235 # GET /$INBOX - top-level inbox view for indexed inboxes
1238 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1239 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic) if @$msgs;
1240 html_oneshot($ctx, dump_topics($ctx), pagination_footer($ctx, '.'));
1243 sub thread_adj_level {
1244 my ($ctx, $level) = @_;
1246 my $max = $ctx->{cur_level};
1248 return ('', '') if $max == 0; # flat output
1250 # reset existing lists
1251 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1252 $ctx->{cur_level} = 0;
1254 } elsif ($level == $max) { # continue existing list
1256 } elsif ($level < $max) {
1257 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1258 $ctx->{cur_level} = $level;
1259 ("$beg<li>", '</li>');
1260 } else { # ($level > $max) # start a new level
1261 $ctx->{cur_level} = $level;
1262 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1267 sub ghost_index_entry {
1268 my ($ctx, $level, $node) = @_;
1269 my ($beg, $end) = thread_adj_level($ctx, $level);
1270 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid} // '?')