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($smsg->{mid})}/" : '';
41 if (_msg_page_prepare($eml, $ctx)) {
42 $eml->each_part(\&add_text_body, $ctx, 1);
43 print { $ctx->{zfh} } '</pre><hr>';
45 html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
46 ''; # XXX TODO cleanup
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($ctx);
59 if (_msg_page_prepare($eml, $ctx)) { # sets {-title_html}
60 $eml->each_part(\&add_text_body, $ctx, 1);
61 print { $ctx->{zfh} } '</pre><hr>';
63 html_footer($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 $ctx->{-spfx} = '../' if $ibx->{coderepo};
84 PublicInbox::WwwStream::aresponse($ctx, \&msg_page_i);
87 # /$INBOX/$MESSAGE_ID/#R
91 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
93 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
96 my $ibx = $ctx->{ibx};
97 if (my $url = $ibx->{infourl}) {
98 $url = prurl($ctx->{env}, $url);
99 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
102 my ($arg, $link, $reply_to_all) =
103 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
104 if (ref($arg) eq 'SCALAR') {
105 return '<pre id=R>'.ascii_html($$arg).'</pre>';
108 # mailto: link only works if address obfuscation is disabled
112 * If your mail client supports setting the <b>In-Reply-To</b> header
113 via mailto: links, try the <a
114 href="$link">mailto: link</a>
118 push @$arg, '/path/to/YOUR_REPLY';
119 $arg = ascii_html(join(" \\\n ", '', @$arg));
122 id=R><b>Reply instructions:</b>
124 You may reply publicly to <a
125 href=#t>this message</a> via plain-text email
126 using any one of the following methods:
128 * Save the following mbox file, import it into your mail client,
129 and $reply_to_all from there: <a
132 Avoid top-posting and favor interleaved quoting:
134 href="$p_url">$p_url</a>
136 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
137 switches of git-send-email(1):
142 href="$se_url">$se_url</a>
145 Be sure your reply has a <b>Subject:</b> header at the top and a blank line
146 before the message body.
152 my $refs = references($hdr);
156 sub fold_addresses ($) {
157 return $_[0] if length($_[0]) <= COLS;
158 # try to fold on commas after non-word chars before $lim chars,
159 # Try to get the "," preceded by ">" or ")", but avoid folding
160 # on the comma where somebody uses "Lastname, Firstname".
161 # We also try to keep the last and penultimate addresses in
162 # the list on the same line if possible, hence the extra \z
163 # Fall back to folding on spaces at $lim + 1 chars
164 my $lim = COLS - 8; # 8 = "\t" display width
165 my $too_long = $lim + 1;
166 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
168 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
171 .{$too_long,}?)(?:\s|\z)/xgo));
174 sub _hdr_names_html ($$) {
175 my ($hdr, $field) = @_;
176 my @vals = $hdr->header($field) or return '';
177 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
181 my ($nr, $singular, $plural) = @_;
182 return "0 $plural" if $nr == 0;
183 $nr == 1 ? "$nr $singular" : "$nr $plural";
186 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
187 # this is already inside a <pre>
189 my ($ctx, $eml) = @_;
190 my $smsg = delete $ctx->{smsg};
191 my $subj = delete $smsg->{subject};
192 my $mid_raw = $smsg->{mid};
193 my $id = id_compress($mid_raw, 1);
195 my $root_anchor = $ctx->{root_anchor} || '';
197 my $obfs_ibx = $ctx->{-obfs_ibx};
199 $subj = '(no subject)' if $subj eq '';
200 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
201 $subj = '<b>'.ascii_html($subj).'</b>';
202 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
203 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
205 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
207 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
209 # Deleting these fields saves about 400K as we iterate across 1K msgs
210 delete @$smsg{qw(ts blob)};
212 my $from = _hdr_names_html($eml, 'From');
213 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
214 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
215 my $upfx = $ctx->{-upfx};
216 my $mhref = $upfx . mid_href($mid_raw) . '/';
217 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
218 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
219 my $to = fold_addresses(_hdr_names_html($eml, 'To'));
220 my $cc = fold_addresses(_hdr_names_html($eml, 'Cc'));
221 my ($tlen, $clen) = (length($to), length($cc));
223 if (($tlen + $clen) > COLS) {
224 $to_cc .= ' To: '.$to."\n" if $tlen;
225 $to_cc .= ' Cc: '.$cc."\n" if $clen;
228 $to_cc .= ' To: '.$to;
229 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
231 $to_cc .= ' Cc: '.$cc if $clen;
235 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
238 my $mapping = $ctx->{mapping};
239 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($eml)))) {
240 my $href = $upfx . mid_href($irt) . '/';
241 my $html = ascii_html($irt);
242 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
244 say { $ctx->zfh } $rv;
246 # scan through all parts, looking for displayable text
247 $ctx->{mhref} = $mhref;
248 $ctx->{changed_href} = "#e$id"; # for diffstat "files? changed,"
249 $eml->each_part(\&add_text_body, $ctx, 1); # expensive
252 $rv = "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
253 "<a\nhref=\"$mhref\">permalink</a>" .
254 " <a\nhref=\"${mhref}raw\">raw</a>" .
255 " <a\nhref=\"${mhref}#R\">reply</a>";
257 delete($ctx->{-qry}) and
258 $rv .= qq[ <a\nhref="${mhref}#related">related</a>];
261 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
262 $rv .= "\t[relevance $pct%]";
265 my $nested = 'nested';
269 $flat = "<b>$flat</b>";
271 $nested = "<b>$nested</b>";
273 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
274 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
275 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
280 # do we have more messages? start a new <pre> if so
281 $rv .= scalar(@{$ctx->{msgs}}) ? '</pre><hr><pre>' : '</pre>' if $hr;
285 sub pad_link ($$;$) {
286 my ($mid, $level, $s) = @_;
288 my $href = defined($mid) ?
289 ("<a\nhref=#r".id_compress($mid, 1).">($s)</a>\n") :
291 (' 'x19).indent_for($level).th_pfx($level).$href;
295 # my ($mapping, $mid) = @_;
296 ($_[0]->{$_[1] // \'bogus'} // [ "(?)\n" ])->[0];
300 my ($mid_raw, $irt, $id, $ctx) = @_;
302 my $mapping = $ctx->{mapping} or return $rv;
304 my $mid_map = $mapping->{$mid_raw} //
305 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
306 my ($attr, $node, $idx, $level) = @$mid_map;
307 my $children = $node->{children};
308 my $nr_c = scalar @$children;
311 # delete saves about 200KB on a 1K message thread
312 if (my $refs = delete $node->{references}) {
313 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
315 my $irt_map = $mapping->{$$irt} if defined $$irt;
316 if (defined $irt_map) {
317 $siblings = $irt_map->[1]->{children};
318 $nr_s = scalar(@$siblings) - 1;
319 $rv .= $pad . $irt_map->[0];
321 my $prev = $siblings->[$idx - 1];
322 my $pmid = $prev->{mid};
324 my $s = ($idx - 1). ' preceding siblings ...';
325 $rv .= pad_link($pmid, $level, $s);
326 } elsif ($idx == 2) {
327 $rv .= $pad . _skel_hdr($mapping,
329 $siblings->[0]->{mid} : undef);
331 $rv .= $pad . _skel_hdr($mapping, $pmid);
334 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
335 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
336 chop $attr; # remove "\n"
337 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
338 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
339 $rv .= "<b>@ $attr</b>\n";
341 my $cmid = $children->[0] ? $children->[0]->{mid} : undef;
342 $rv .= $pad . _skel_hdr($mapping, $cmid);
344 my $s = ($nr_c - 1). ' more replies';
345 $rv .= pad_link($cmid, $level + 1, $s);
346 } elsif (my $cn = $children->[1]) {
347 $rv .= $pad . _skel_hdr($mapping, $cn->{mid});
351 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
353 my $nmid = $next->{mid};
354 $rv .= $pad . _skel_hdr($mapping, $nmid);
355 my $nnext = $nr_s - $idx;
357 my $s = ($nnext - 1).' subsequent siblings';
358 $rv .= pad_link($nmid, $level, $s);
359 } elsif (my $nn = $siblings->[$idx + 2]) {
360 $rv .= $pad . _skel_hdr($mapping, $nn->{mid});
363 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
366 # non-recursive thread walker
367 sub walk_thread ($$$) {
368 my ($rootset, $ctx, $cb) = @_;
369 my @q = map { (0, $_, -1) } @$rootset;
371 my ($level, $node, $i) = splice(@q, 0, 3);
372 defined $node or next;
373 $cb->($ctx, $level, $node, $i) or return;
376 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
380 sub pre_thread { # walk_thread callback
381 my ($ctx, $level, $node, $idx) = @_;
382 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
383 skel_dump($ctx, $level, $node);
386 sub thread_eml_entry {
387 my ($ctx, $eml) = @_;
388 my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
389 print { $ctx->zfh } $beg, '<pre>';
390 print { $ctx->{zfh} } eml_entry($ctx, $eml), '</pre>', $end;
393 sub next_in_queue ($$) {
394 my ($q, $ghost_ok) = @_;
396 my ($level, $smsg) = splice(@$q, 0, 2);
398 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
399 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
404 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
405 my ($ctx, $eml) = @_;
406 return thread_eml_entry($ctx, $eml) if $eml;
407 return unless exists($ctx->{skel});
408 my $ghost_ok = $ctx->{nr}++;
410 my ($lvl, $smsg) = next_in_queue($ctx->{-queue}, $ghost_ok);
412 if (exists $smsg->{blob}) { # next message for cat-file
413 $ctx->{level} = $lvl;
414 if (!$ghost_ok) { # first non-ghost
415 $ctx->{-title_html} =
416 ascii_html($smsg->{subject});
417 print { $ctx->zfh } $ctx->html_top;
421 # buffer the ghost entry and loop
422 print { $ctx->zfh } ghost_index_entry($ctx, $lvl, $smsg)
424 print { $ctx->zfh } thread_adj_level($ctx, 0),
425 ${delete($ctx->{skel})};
431 sub stream_thread ($$) {
432 my ($rootset, $ctx) = @_;
433 @{$ctx->{-queue}} = map { (0, $_) } @$rootset;
434 PublicInbox::WwwStream::aresponse($ctx, \&stream_thread_i);
437 # /$INBOX/$MSGID/t/ and /$INBOX/$MSGID/T/
440 $ctx->{-upfx} = '../../';
441 my $mid = $ctx->{mid};
442 my $ibx = $ctx->{ibx};
443 my ($nr, $msgs) = $ibx->over->get_thread($mid);
444 return missing_thread($ctx) if $nr == 0;
445 $ctx->{-spfx} = '../../' if $ibx->{coderepo};
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 print { $ctx->zfh } $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
503 print { $ctx->zfh } $$skel;
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 ($s, $err) = msg_part_text($part, $ct);
565 $s // return print $zfh (attach_link($ctx, $ct, $p, $fn) // '');
566 say $zfh submsg_hdr($ctx, $part) if $part->{is_submsg};
568 # makes no difference to browsers, and don't screw up filename
569 # link generation in diffs with the extra '%0D'
572 # will be escaped to `•' in HTML
573 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
575 # always support diff-highlighting, but we can't linkify hunk
576 # headers for solver unless some coderepo are configured:
578 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
579 # diffstat anchors do not link across attachments or messages,
580 # -apfx is just a stable prefix for making diffstat anchors
581 # linkable to the first diff hunk w/o crossing attachments
582 $idx =~ tr!.!/!; # compatibility with previous versions
583 $ctx->{-apfx} = $upfx . $idx;
585 # do attr => filename mappings for diffstats in git diffs:
586 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
588 delete $ctx->{-long_path};
591 # split off quoted and unquoted blocks:
592 my @sections = PublicInbox::MsgIter::split_quotes($s);
593 undef $s; # free memory
594 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
595 # badly-encoded message with $err? tell the world about it!
596 say $zfh attach_link($ctx, $ct, $p, $fn, $err);
598 delete $part->{bdy}; # save memory
599 for my $cur (@sections) { # $cur may be huge
601 # we use a <span> here to allow users to specify
602 # their own color for quoted text
603 print $zfh qq(<span\nclass="q">),
604 $l->to_html($cur), '</span>';
606 flush_diff($ctx, \$cur);
607 } else { # regular lines, OK
608 print $zfh $l->to_html($cur);
610 undef $cur; # free memory
614 sub _msg_page_prepare {
615 my ($eml, $ctx) = @_;
616 my $have_over = !!$ctx->{ibx}->over;
617 my $mids = mids_for_index($eml);
618 my $nr = $ctx->{nr}++;
619 if ($nr) { # unlikely
620 if ($ctx->{chash} eq content_hash($eml)) {
621 warn "W: BUG? @$mids not deduplicated properly\n";
625 "<pre>WARNING: multiple messages have this Message-ID\n</pre><pre>";
627 $ctx->{first_hdr} = $eml->header_obj;
628 $ctx->{chash} = content_hash($eml) if $ctx->{smsg}; # reused MID
629 $ctx->{-html_tip} = "<pre\nid=b>"; # anchor for body start
631 $ctx->{-upfx} = '../';
632 my @title; # (Subject[0], From[0])
634 for my $v ($eml->header('From')) {
635 my @n = PublicInbox::Address::names($v);
636 $title[1] //= join(', ', @n);
637 $hbuf .= "From: $v\n" if $v ne '';
639 for my $h (qw(To Cc)) {
640 for my $v ($eml->header($h)) {
642 $hbuf .= "$h: $v\n" if $v ne '';
645 my @subj = $eml->header('Subject');
646 $hbuf .= "Subject: $_\n" for @subj;
647 $title[0] = $subj[0] // '(no subject)';
648 $hbuf .= "Date: $_\n" for $eml->header('Date');
649 $hbuf = ascii_html($hbuf);
650 $ctx->{-title_html} = ascii_html(join(' - ', @title));
651 if (my $obfs_ibx = $ctx->{-obfs_ibx}) {
652 obfuscate_addrs($obfs_ibx, $hbuf);
653 obfuscate_addrs($obfs_ibx, $ctx->{-title_html});
656 # [thread overview] link is typically added after Date,
657 # but added after Subject, or even nothing.
659 chop $hbuf; # drop "\n", or noop if $rv eq ''
660 $hbuf .= qq{\t<a\nhref="#r">[thread overview]</a>\n};
661 $hbuf =~ s!^Subject:\x20(.*?)(\n[A-Z]|\z)
662 !Subject: <a\nhref="#r"\nid=t>$1</a>$2!msx or
663 $hbuf .= qq(<a\nhref="#r\nid=t></a>);
665 if (scalar(@$mids) == 1) { # common case
666 my $x = ascii_html($mids->[0]);
667 $hbuf .= qq[Message-ID: <$x> (<a href="raw">raw</a>)\n];
669 if (!$nr) { # first (and only) message, common case
670 print { $ctx->zfh } $ctx->html_top, $hbuf;
672 delete $ctx->{-title_html};
673 print { $ctx->zfh } $ctx->{-html_tip}, $hbuf;
675 $ctx->{-linkify} //= PublicInbox::Linkify->new;
677 if (scalar(@$mids) != 1) { # unlikely, but it happens :<
678 # X-Alt-Message-ID can happen if a message is injected from
679 # public-inbox-nntpd because of multiple Message-ID headers.
680 for my $h (qw(Message-ID X-Alt-Message-ID)) {
681 $hbuf .= "$h: $_\n" for ($eml->header_raw($h));
683 $ctx->{-linkify}->linkify_mids('..', \$hbuf, 1); # escapes HTML
684 print { $ctx->{zfh} } $hbuf;
687 my @irt = $eml->header_raw('In-Reply-To');
690 $refs = references($eml);
691 $irt[0] = pop(@$refs) if scalar @$refs;
693 $hbuf .= "In-Reply-To: $_\n" for @irt;
695 # do not display References: if search is present,
696 # we show the thread skeleton at the bottom, instead.
698 $refs //= references($eml);
699 $hbuf .= 'References: <'.join(">\n\t<", @$refs).">\n" if @$refs;
701 $ctx->{-linkify}->linkify_mids('..', \$hbuf); # escapes HTML
702 say { $ctx->{zfh} } $hbuf;
707 qq(expand[<a\nhref="T/#u">flat</a>) .
708 qq(|<a\nhref="t/#u">nested</a>] ) .
709 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
710 qq(<a\nhref="t.atom">Atom feed</a>);
713 sub thread_skel ($$$) {
714 my ($skel, $ctx, $hdr) = @_;
715 my $mid = mids($hdr)->[0];
716 my $ibx = $ctx->{ibx};
717 my ($nr, $msgs) = $ibx->over->get_thread($mid);
718 my $parent = in_reply_to($hdr);
719 $$skel .= "\n<b>Thread overview: </b>";
721 if (defined $parent) {
722 $$skel .= SKEL_EXPAND."\n ";
723 $$skel .= ghost_parent('../', $parent) . "\n";
725 $$skel .= "<a\nid=r>[no followups]</a> ".
728 $ctx->{next_msg} = undef;
729 $ctx->{parent_msg} = $parent;
734 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
736 # nb: mutt only shows the first Subject in the index pane
737 # when multiple Subject: headers are present, so we follow suit:
738 my $subj = $hdr->header('Subject') // '';
739 $subj = '(no subject)' if $subj eq '';
741 $ctx->{prev_attr} = '';
742 $ctx->{prev_level} = 0;
743 $ctx->{skel} = $skel;
745 # reduce hash lookups in skel_dump
746 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
747 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
749 $ctx->{parent_msg} = $parent;
754 my ($ctx, $hdr) = @_;
756 my (@related, $skel);
758 my $qry = delete $ctx->{-qry};
759 if ($qry && $ctx->{ibx}->isrch) {
760 my $q = ''; # search for either ancestor or descendent patches
761 for (@{$qry->{dfpre}}, @{$qry->{dfpost}}) {
762 chop if length > 7; # include 1 abbrev "older" patches
765 chop $q; # omit trailing SP
766 local $Text::Wrap::columns = COLS;
767 local $Text::Wrap::huge = 'overflow';
768 $q = wrap('', '', $q);
769 my $rows = ($q =~ tr/\n/\n/) + 1;
774 ><pre>find likely ancestor, descendant, or conflicting patches for <a
775 href=#t>this message</a>:
776 <textarea name=q cols=${\COLS} rows=$rows>$q</textarea>
777 <input type=submit value=search
778 />\t(<a href=${upfx}_/text/help/#search>help</a>)</pre></form>
781 if ($ctx->{ibx}->over) {
782 my $t = ts2str($ctx->{-t_max});
783 my $t_fmt = fmt_ts($ctx->{-t_max});
784 my $fallback = @related ? "\t" : "<a id=related>\t</a>";
786 ${fallback}other threads:[<a
787 href="$upfx?t=$t">~$t_fmt UTC</a>|<a
788 href="$upfx">newest</a>]
790 thread_skel(\$skel, $ctx, $hdr);
795 if (my $n = $ctx->{next_msg}) {
797 $next = qq(<a\nhref="$upfx$n/"\nrel=next>next</a>);
799 my $par = $ctx->{parent_msg};
800 my $u = $par ? $upfx.mid_href($par).'/' : undef;
801 if (my $p = $ctx->{prev_msg}) {
802 $prev = mid_href($p);
803 if ($p && $par && $p eq $par) {
804 $prev = qq(<a\nhref="$upfx$prev/"\n) .
805 'rel=prev>prev parent</a>';
808 $prev = qq(<a\nhref="$upfx$prev/"\n) .
810 $parent = qq( <a\nhref="$u">parent</a>) if $u;
812 } elsif ($u) { # unlikely
813 $parent = qq( <a\nhref="$u"\nrel=prev>parent</a>);
815 $foot .= "$next $prev$parent ";
816 } else { # unindexed inboxes w/o over
817 $skel = qq( <a\nhref="$upfx">latest</a>);
819 # $skel may be big for big threads, don't append it to $foot
820 print { $ctx->zfh } $foot, qq(<a\nhref="#R">reply</a>),
821 $skel, '</pre>', @related,
822 msg_reply($ctx, $hdr);
826 my ($upfx, $mid) = @_;
828 my $href = mid_href($mid);
829 my $html = ascii_html($mid);
830 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
835 $level ? INDENT x ($level - 1) : '';
839 my ($ctx, $level, $node, $idx) = @_;
840 ++$ctx->{root_idx} if $level == 0;
841 if ($node->{mid} eq $ctx->{mid}) {
842 $ctx->{found_mid_at} = $ctx->{root_idx};
843 return 0; # stop iterating
848 sub strict_loose_note ($) {
851 " -- strict thread matches above, loose matches on Subject: below --\n";
853 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
855 " -- use mbox.gz link to download all $nr messages --\n";
861 my ($ctx, $msgs) = @_;
862 require PublicInbox::SearchThread;
863 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
865 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
866 # and preserved in the future. This bug is hidden by `sid' matches
867 # in get_thread, so we never noticed it until now. And even when
868 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
869 # bump since reindexing is expensive and users may not do it
871 # loose threading could've returned too many results,
872 # put the root the message we care about at the top:
873 my $mid = $ctx->{mid};
874 if (defined($mid) && scalar(@$rootset) > 1) {
875 $ctx->{root_idx} = -1;
876 my $nr = scalar @$msgs;
877 walk_thread($rootset, $ctx, \&find_mid_root);
878 my $idx = $ctx->{found_mid_at};
879 if (defined($idx) && $idx != 0) {
880 my $tip = splice(@$rootset, $idx, 1);
881 @$rootset = reverse @$rootset;
882 unshift @$rootset, $tip;
883 $ctx->{sl_note} = strict_loose_note($nr);
891 require PublicInbox::ExtMsg;
892 PublicInbox::ExtMsg::ext_msg($ctx);
896 my ($prev_subj, $subj, $val) = @_;
898 my $omit; # '"' denotes identical text omitted
899 my (@prev_pop, @curr_pop);
900 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
901 push(@prev_pop, pop(@$prev_subj));
902 push(@curr_pop, pop(@$subj));
905 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
906 if (scalar(@curr_pop) == 1) {
908 push @$prev_subj, @prev_pop;
909 push @$subj, @curr_pop;
914 sub skel_dump { # walk_thread callback
915 my ($ctx, $level, $smsg) = @_;
916 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
918 my $skel = $ctx->{skel};
919 my $cur = $ctx->{cur};
920 my $mid = $smsg->{mid};
922 if ($level == 0 && $ctx->{skel_dump_roots}++) {
923 $$skel .= delete($ctx->{sl_note}) || '';
926 my $f = ascii_html(delete $smsg->{from_name});
927 my $obfs_ibx = $ctx->{-obfs_ibx};
928 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
930 my $d = fmt_ts($smsg->{ds});
931 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
932 if (exists $ctx->{searchview}) {
933 if (defined(my $pct = $smsg->{pct})) {
934 $d .= (sprintf(' % 2u', $pct) . '%');
940 $d .= ' ' . indent_for($level) . th_pfx($level);
942 $ctx->{first_level} ||= $level;
944 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
945 $ctx->{prev_attr} = $attr;
947 $ctx->{prev_level} = $level;
952 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
953 "$attr [this message]</a></b>\n";
956 $ctx->{prev_msg} = $mid;
959 $ctx->{next_msg} ||= $mid;
962 # Subject is never undef, this mail was loaded from
963 # our Xapian which would've resulted in '' if it were
964 # really missing (and Filter rejects empty subjects)
965 my @subj = split(/ /, subject_normalized($smsg->{subject}));
966 # remove common suffixes from the subject if it matches the previous,
967 # so we do not show redundant text at the end.
968 my $prev_subj = $ctx->{prev_subj} || [];
969 $ctx->{prev_subj} = [ @subj ];
970 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
973 my $subj = join(' ', @subj);
974 $subj = ascii_html($subj);
975 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
976 $end = "$subj</a> $omit$f\n"
982 my $mapping = $unmatched ? undef : $ctx->{mapping};
984 my $map = $mapping->{$mid};
985 $id = id_compress($mid, 1);
987 $map->[0] = "$d<a\nhref=\"$m\">$end";
990 $m = $ctx->{-upfx}.mid_href($mid).'/';
992 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
997 my ($ctx, $level, $node) = @_;
999 my $mid = $node->{mid};
1000 my $d = ' [not found] ';
1001 $d .= ' ' if exists $ctx->{searchview};
1002 $d .= indent_for($level) . th_pfx($level);
1003 my $upfx = $ctx->{-upfx};
1004 my $href = $upfx . mid_href($mid) . '/';
1005 my $html = ascii_html($mid);
1007 my $mapping = $ctx->{mapping};
1008 my $map = $mapping->{$mid} if $mapping;
1010 my $id = id_compress($mid, 1);
1011 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1012 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1014 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1016 ${$ctx->{skel}} .= $d;
1022 (eval { $a->topmost->{ds} } || 0) <=>
1023 (eval { $b->topmost->{ds} } || 0)
1027 # accumulate recent topics if search is supported
1028 # returns 200 if done, 404 if not
1029 sub acc_topic { # walk_thread callback
1030 my ($ctx, $level, $smsg) = @_;
1031 my $mid = $smsg->{mid};
1032 my $has_blob = $smsg->{blob} // do {
1033 if (my $by_mid = $ctx->{ibx}->smsg_by_mid($mid)) {
1034 %$smsg = (%$smsg, %$by_mid);
1039 my $subj = subject_normalized($smsg->{subject});
1040 $subj = '(no subject)' if $subj eq '';
1041 my $ds = $smsg->{ds};
1042 if ($level == 0) { # new, top-level topic
1043 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1044 $ctx->{-cur_topic} = $topic;
1045 push @{$ctx->{order}}, $topic;
1049 # continue existing topic
1050 my $topic = $ctx->{-cur_topic}; # should never be undef
1051 $topic->[0] = $ds if $ds > $topic->[0];
1052 $topic->[1]++; # bump N+ message counter
1053 my $seen = $topic->[2];
1054 if (scalar(@$topic) == 3) { # parent was a ghost
1055 push @$topic, $subj;
1056 } elsif (!defined($seen->{$subj})) {
1057 push @$topic, $level, $subj; # @extra messages
1059 $seen->{$subj} = $mid; # latest for subject
1060 } else { # ghost message
1061 return 1 if $level != 0; # ignore child ghosts
1062 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1063 push @{$ctx->{order}}, $topic;
1070 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1072 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1077 my $obfs_ibx = $ctx->{ibx}->{obfuscate} ? $ctx->{ibx} : undef;
1078 if (my $note = delete $ctx->{t_note}) {
1079 push @out, $note; # "messages from ... to ..."
1081 # sort by recency, this allows new posts to "bump" old topics...
1082 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1083 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1085 next unless defined $top_subj; # ghost topic
1086 my $mid = delete $seen->{$top_subj};
1087 my $href = mid_href($mid);
1088 my $prev_subj = [ split(/ /, $top_subj) ];
1089 $top_subj = ascii_html($top_subj);
1092 # $n isn't the total number of posts on the topic,
1093 # just the number of posts in the current results window
1097 $anchor = '#u'; # top of only message
1099 $n = " ($n+ messages)";
1100 $anchor = '#t'; # thread skeleton
1103 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1106 my $level = shift @extra;
1107 my $subj = shift @extra; # already normalized
1108 $mid = delete $seen->{$subj};
1109 my @subj = split(/ /, $subj);
1110 my @next_prev = @subj; # full copy
1111 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1112 $prev_subj = \@next_prev;
1113 $subj = join(' ', @subj);
1114 $subj = ascii_html($subj);
1115 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1116 $href = mid_href($mid);
1117 $s .= indent_for($level) . TCHILD;
1118 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1122 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1127 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1128 timegm($ss || 0, $mm || 0, $hh || 0, $dd, $mon - 1, $yyyy);
1131 sub pagination_footer ($$) {
1132 my ($ctx, $latest) = @_;
1133 my $next = $ctx->{next_page} || '';
1134 my $prev = $ctx->{prev_page} || '';
1135 if ($prev) { # aligned padding for: 'next (older) | '
1136 $next = $next ? "$next | " : ' | ';
1137 $prev .= qq[ | <a\nhref="$latest">latest</a>];
1139 ($next || $prev) ? "<hr><pre id=nav>page: $next$prev</pre>" : '';
1142 sub paginate_recent ($$) {
1143 my ($ctx, $lim) = @_;
1144 my $t = $ctx->{qp}->{t} || '';
1145 my $opts = { limit => $lim };
1146 my ($after, $before);
1148 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1149 # if only $after exists "YYYYMMDD.." because "." could be skipped
1150 # if interpreted as an end-of-sentence
1151 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1152 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1154 my $msgs = $ctx->{ibx}->over->recent($opts, $after, $before);
1155 if (defined($after) && scalar(@$msgs) < $lim) {
1156 $after = $before = undef;
1157 $msgs = $ctx->{ibx}->over->recent($opts);
1159 my $more = scalar(@$msgs) == $lim;
1160 my ($newest, $oldest);
1162 $newest = $msgs->[0]->{ts};
1163 $oldest = $msgs->[-1]->{ts};
1164 # if we only had $after, our SQL query in ->recent ordered
1165 if ($newest < $oldest) {
1166 ($oldest, $newest) = ($newest, $oldest);
1167 $more = undef if defined($after) && $after < $oldest;
1169 if (defined($after // $before)) {
1170 my $n = strftime('%Y-%m-%d %H:%M:%S', gmtime($newest));
1171 my $o = strftime('%Y-%m-%d %H:%M:%S', gmtime($oldest));
1172 $ctx->{t_note} = <<EOM;
1173 messages from $o to $n UTC [<a href="#nav">more...</a>]
1175 my $s = ts2str($newest);
1176 $ctx->{prev_page} = qq[<a\nhref="?t=$s-"\nrel=prev>] .
1180 if (defined($oldest) && $more) {
1181 my $s = ts2str($oldest);
1182 $ctx->{next_page} = qq[<a\nhref="?t=$s"\nrel=next>] .
1188 # GET /$INBOX - top-level inbox view for indexed inboxes
1191 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1192 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic) if @$msgs;
1193 html_oneshot($ctx, dump_topics($ctx), pagination_footer($ctx, '.'));
1196 sub thread_adj_level {
1197 my ($ctx, $level) = @_;
1199 my $max = $ctx->{cur_level};
1201 return ('', '') if $max == 0; # flat output
1203 # reset existing lists
1204 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1205 $ctx->{cur_level} = 0;
1207 } elsif ($level == $max) { # continue existing list
1209 } elsif ($level < $max) {
1210 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1211 $ctx->{cur_level} = $level;
1212 ("$beg<li>", '</li>');
1213 } else { # ($level > $max) # start a new level
1214 $ctx->{cur_level} = $level;
1215 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1220 sub ghost_index_entry {
1221 my ($ctx, $level, $node) = @_;
1222 my ($beg, $end) = thread_adj_level($ctx, $level);
1223 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid} // '?')