1 # Copyright (C) 2014-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Used for displaying the HTML web interface.
5 # See Documentation/design_www.txt for this.
6 package PublicInbox::View;
9 use bytes (); # only for bytes::length
10 use List::Util qw(max);
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 Time::Local qw(timegm);
24 use PublicInbox::Smsg qw(subject_normalized);
25 use constant COLS => 72;
26 use constant INDENT => ' ';
27 use constant TCHILD => '` ';
28 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
32 if ($eml) { # called by WwwStream::async_eml or getline
33 my $smsg = $ctx->{smsg};
34 $ctx->{smsg} = $ctx->{over}->next_by_mid(@{$ctx->{next_arg}});
35 $ctx->{mhref} = ($ctx->{nr} || $ctx->{smsg}) ?
36 "../${\mid_href($smsg->{mid})}/" : '';
37 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($eml, $ctx);
38 multipart_text_as_html($eml, $ctx);
40 $$obuf .= '</pre><hr>';
41 $$obuf .= html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
43 } else { # called by WwwStream::async_next or getline
44 $ctx->{smsg}; # may be undef
48 # /$INBOX/$MSGID/ for unindexed v1 inboxes
49 sub no_over_html ($) {
51 my $bref = $ctx->{-inbox}->msg_by_mid($ctx->{mid}) or return; # 404
52 my $eml = PublicInbox::Eml->new($bref);
54 PublicInbox::WwwStream::init($ctx);
55 my $obuf = $ctx->{obuf} = _msg_page_prepare_obuf($eml, $ctx);
56 multipart_text_as_html($eml, $ctx);
58 $$obuf .= '</pre><hr>';
59 eval { $$obuf .= html_footer($ctx, $eml) };
60 html_oneshot($ctx, 200, $obuf);
63 # public functions: (unstable)
67 my $ibx = $ctx->{-inbox};
68 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
69 my $over = $ctx->{over} = $ibx->over or return no_over_html($ctx);
71 my $next_arg = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
73 my $smsg = $ctx->{smsg} = $over->next_by_mid(@$next_arg) or
74 return; # undef == 404
76 # allow user to easily browse the range around this message if
78 $ctx->{-t_max} = $smsg->{ts};
79 PublicInbox::WwwStream::aresponse($ctx, 200, \&msg_page_i);
82 # /$INBOX/$MESSAGE_ID/#R
86 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
88 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
91 my $ibx = $ctx->{-inbox};
92 if (my $url = $ibx->{infourl}) {
93 $url = prurl($ctx->{env}, $url);
94 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
97 my ($arg, $link, $reply_to_all) =
98 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
99 if (ref($arg) eq 'SCALAR') {
100 return '<pre id=R>'.ascii_html($$arg).'</pre>';
103 # mailto: link only works if address obfuscation is disabled
107 * If your mail client supports setting the <b>In-Reply-To</b> header
108 via mailto: links, try the <a
109 href="$link">mailto: link</a>
113 push @$arg, '/path/to/YOUR_REPLY';
114 $arg = ascii_html(join(" \\\n ", '', @$arg));
117 id=R><b>Reply instructions:</b>
119 You may reply publicly to <a
120 href=#t>this message</a> via plain-text email
121 using any one of the following methods:
123 * Save the following mbox file, import it into your mail client,
124 and $reply_to_all from there: <a
127 Avoid top-posting and favor interleaved quoting:
129 href="$p_url">$p_url</a>
131 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
132 switches of git-send-email(1):
137 href="$se_url">$se_url</a>
144 my $refs = references($hdr);
148 sub fold_addresses ($) {
149 return $_[0] if length($_[0]) <= COLS;
150 # try to fold on commas after non-word chars before $lim chars,
151 # Try to get the "," preceded by ">" or ")", but avoid folding
152 # on the comma where somebody uses "Lastname, Firstname".
153 # We also try to keep the last and penultimate addresses in
154 # the list on the same line if possible, hence the extra \z
155 # Fall back to folding on spaces at $lim + 1 chars
156 my $lim = COLS - 8; # 8 = "\t" display width
157 my $too_long = $lim + 1;
158 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
160 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
163 .{$too_long,}?)(?:\s|\z)/xgo));
166 sub _hdr_names_html ($$) {
167 my ($hdr, $field) = @_;
168 my @vals = $hdr->header($field) or return '';
169 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
173 my ($nr, $singular, $plural) = @_;
174 return "0 $plural" if $nr == 0;
175 $nr == 1 ? "$nr $singular" : "$nr $plural";
178 # Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
179 # this is already inside a <pre>
181 my ($ctx, $eml) = @_;
182 my $smsg = delete $ctx->{smsg};
183 my $subj = delete $smsg->{subject};
184 my $mid_raw = $smsg->{mid};
185 my $id = id_compress($mid_raw, 1);
187 my $root_anchor = $ctx->{root_anchor} || '';
189 my $obfs_ibx = $ctx->{-obfs_ibx};
191 $subj = '(no subject)' if $subj eq '';
192 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
193 $subj = '<b>'.ascii_html($subj).'</b>';
194 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
195 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
197 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
199 my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
201 # Deleting these fields saves about 400K as we iterate across 1K msgs
202 delete @$smsg{qw(ts blob)};
204 my $from = _hdr_names_html($eml, 'From');
205 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
206 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
207 my $upfx = $ctx->{-upfx};
208 my $mhref = $upfx . mid_href($mid_raw) . '/';
209 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
210 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
211 my $to = fold_addresses(_hdr_names_html($eml, 'To'));
212 my $cc = fold_addresses(_hdr_names_html($eml, 'Cc'));
213 my ($tlen, $clen) = (length($to), length($cc));
215 if (($tlen + $clen) > COLS) {
216 $to_cc .= ' To: '.$to."\n" if $tlen;
217 $to_cc .= ' Cc: '.$cc."\n" if $clen;
220 $to_cc .= ' To: '.$to;
221 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
223 $to_cc .= ' Cc: '.$cc if $clen;
227 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
230 my $mapping = $ctx->{mapping};
231 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($eml)))) {
232 my $href = $upfx . mid_href($irt) . '/';
233 my $html = ascii_html($irt);
234 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
238 # scan through all parts, looking for displayable text
239 $ctx->{mhref} = $mhref;
241 $eml->each_part(\&add_text_body, $ctx, 1);
245 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
246 "<a\nhref=\"$mhref\">permalink</a>" .
247 " <a\nhref=\"${mhref}raw\">raw</a>" .
248 " <a\nhref=\"${mhref}#R\">reply</a>";
251 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
252 $rv .= "\t[relevance $pct%]";
255 my $nested = 'nested';
260 $flat = "<b>$flat</b>";
262 $nested = "<b>$nested</b>";
264 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
265 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
266 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
271 # do we have more messages? start a new <pre> if so
272 $rv .= scalar(@{$ctx->{msgs}}) ? '</pre><hr><pre>' : '</pre>' if $hr;
276 sub pad_link ($$;$) {
277 my ($mid, $level, $s) = @_;
279 my $id = id_compress($mid, 1);
280 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
284 my ($mid_raw, $irt, $id, $ctx) = @_;
286 my $mapping = $ctx->{mapping} or return $rv;
288 my $mid_map = $mapping->{$mid_raw};
290 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
291 my ($attr, $node, $idx, $level) = @$mid_map;
292 my $children = $node->{children};
293 my $nr_c = scalar @$children;
296 # delete saves about 200KB on a 1K message thread
297 if (my $refs = delete $node->{references}) {
298 ($$irt) = ($refs =~ m/$MID_EXTRACT\z/o);
300 my $irt_map = $mapping->{$$irt} if defined $$irt;
301 if (defined $irt_map) {
302 $siblings = $irt_map->[1]->{children};
303 $nr_s = scalar(@$siblings) - 1;
304 $rv .= $pad . $irt_map->[0];
306 my $prev = $siblings->[$idx - 1];
307 my $pmid = $prev->{mid};
309 my $s = ($idx - 1). ' preceding siblings ...';
310 $rv .= pad_link($pmid, $level, $s);
311 } elsif ($idx == 2) {
312 my $ppmid = $siblings->[0]->{mid};
313 $rv .= $pad . $mapping->{$ppmid}->[0];
315 $rv .= $pad . $mapping->{$pmid}->[0];
318 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
319 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
320 $attr =~ s!\n\z!</b>\n!s;
321 $attr =~ s!<a\nhref.*</a> (?:" )?!!s; # no point in dup subject
322 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
325 my $cmid = $children->[0]->{mid};
326 $rv .= $pad . $mapping->{$cmid}->[0];
328 my $s = ($nr_c - 1). ' more replies';
329 $rv .= pad_link($cmid, $level + 1, $s);
330 } elsif (my $cn = $children->[1]) {
331 $rv .= $pad . $mapping->{$cn->{mid}}->[0];
335 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
337 my $nmid = $next->{mid};
338 $rv .= $pad . $mapping->{$nmid}->[0];
339 my $nnext = $nr_s - $idx;
341 my $s = ($nnext - 1).' subsequent siblings';
342 $rv .= pad_link($nmid, $level, $s);
343 } elsif (my $nn = $siblings->[$idx + 2]) {
344 $rv .= $pad . $mapping->{$nn->{mid}}->[0];
347 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
350 # non-recursive thread walker
351 sub walk_thread ($$$) {
352 my ($rootset, $ctx, $cb) = @_;
353 my @q = map { (0, $_, -1) } @$rootset;
355 my ($level, $node, $i) = splice(@q, 0, 3);
356 defined $node or next;
357 $cb->($ctx, $level, $node, $i) or return;
360 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
364 sub pre_thread { # walk_thread callback
365 my ($ctx, $level, $node, $idx) = @_;
366 $ctx->{mapping}->{$node->{mid}} = [ '', $node, $idx, $level ];
367 skel_dump($ctx, $level, $node);
370 sub thread_eml_entry {
371 my ($ctx, $eml) = @_;
372 my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
373 $beg . '<pre>' . eml_entry($ctx, $eml) . '</pre>' . $end;
376 sub next_in_queue ($$) {
377 my ($q, $ghost_ok) = @_;
379 my ($level, $smsg) = splice(@$q, 0, 2);
381 unshift @$q, map { ($cl, $_) } @{$smsg->{children}};
382 return ($level, $smsg) if $ghost_ok || exists($smsg->{blob});
387 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
388 my ($ctx, $eml) = @_;
389 return thread_eml_entry($ctx, $eml) if $eml;
390 return unless exists($ctx->{skel});
391 my $ghost_ok = $ctx->{nr}++;
393 my ($lvl, $smsg) = next_in_queue($ctx->{-queue}, $ghost_ok);
395 if (exists $smsg->{blob}) { # next message for cat-file
396 $ctx->{level} = $lvl;
397 if (!$ghost_ok) { # first non-ghost
398 $ctx->{-title_html} =
399 ascii_html($smsg->{subject});
400 $ctx->zmore($ctx->html_top);
404 # buffer the ghost entry and loop
405 $ctx->zmore(ghost_index_entry($ctx, $lvl, $smsg));
407 $ctx->zmore(join('', thread_adj_level($ctx, 0)));
408 $ctx->zmore(${delete($ctx->{skel})});
414 sub stream_thread ($$) {
415 my ($rootset, $ctx) = @_;
416 $ctx->{-queue} = [ map { (0, $_) } @$rootset ];
417 PublicInbox::WwwStream::aresponse($ctx, 200, \&stream_thread_i);
420 # /$INBOX/$MSGID/t/ and /$INBOX/$MSGID/T/
423 my $mid = $ctx->{mid};
424 my $ibx = $ctx->{-inbox};
425 my ($nr, $msgs) = $ibx->over->get_thread($mid);
426 return missing_thread($ctx) if $nr == 0;
428 # link $INBOX_DIR/description text to "index_topics" view around
429 # the newest message in this thread
430 my $t = ts2str($ctx->{-t_max} = max(map { delete $_->{ts} } @$msgs));
431 my $t_fmt = fmt_ts($ctx->{-t_max});
433 my $skel = '<hr><pre>';
434 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
437 href="../../?t=$t">~$t_fmt UTC</a> | <a
438 href="../../">newest</a>]
441 $skel .= "<b\nid=t>Thread overview:</b> ";
442 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
443 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
444 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
445 $skel .= "-- links below jump to the message on this page --\n";
446 $ctx->{-upfx} = '../../';
447 $ctx->{cur_level} = 0;
448 $ctx->{skel} = \$skel;
449 $ctx->{prev_attr} = '';
450 $ctx->{prev_level} = 0;
451 $ctx->{root_anchor} = anchor_for($mid);
452 $ctx->{mapping} = {};
453 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
456 my $rootset = thread_results($ctx, $msgs);
458 # reduce hash lookups in pre_thread->skel_dump
459 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
460 walk_thread($rootset, $ctx, \&pre_thread);
463 return stream_thread($rootset, $ctx) unless $ctx->{flat};
465 # flat display: lazy load the full message from smsg
466 $ctx->{msgs} = $msgs;
467 $ctx->{-html_tip} = '<pre>';
468 PublicInbox::WwwStream::aresponse($ctx, 200, \&thread_html_i);
471 sub thread_html_i { # PublicInbox::WwwStream::getline callback
472 my ($ctx, $eml) = @_;
474 my $smsg = $ctx->{smsg};
475 if (exists $ctx->{-html_tip}) {
476 $ctx->{-title_html} = ascii_html($smsg->{subject});
477 $ctx->zmore($ctx->html_top);
479 return eml_entry($ctx, $eml);
481 while (my $smsg = shift @{$ctx->{msgs}}) {
482 return $smsg if exists($smsg->{blob});
484 my $skel = delete($ctx->{skel}) or return; # all done
490 sub multipart_text_as_html {
491 # ($mime, $ctx) = @_; # each_part may do "$_[0] = undef"
493 # scan through all parts, looking for displayable text
494 $_[0]->each_part(\&add_text_body, $_[1], 1);
497 sub submsg_hdr ($$) {
498 my ($ctx, $eml) = @_;
499 my $obfs_ibx = $ctx->{-obfs_ibx};
500 my $rv = $ctx->{obuf};
502 for my $h (qw(From To Cc Subject Date Message-ID X-Alt-Message-ID)) {
503 my @v = $eml->header($h);
505 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
512 sub attach_link ($$$$;$) {
513 my ($ctx, $ct, $p, $fn, $err) = @_;
514 my ($part, $depth, $idx) = @$p;
516 # Eml iteration clobbers multipart ->{bdy}, so do not offer
517 # downloads for 0-byte multipart attachments
518 return unless $part->{bdy};
520 my $nl = $idx eq '1' ? '' : "\n"; # like join("\n", ...)
521 my $size = bytes::length($part->body);
523 # hide attributes normally, unless we want to aid users in
524 # spotting MUA problems:
525 $ct =~ s/;.*// unless $err;
526 $ct = ascii_html($ct);
528 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
530 } elsif ($ct eq 'text/plain') {
535 my $rv = $ctx->{obuf};
536 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
539 [-- Warning: decoded text below may be mangled, UTF-8 assumed --]
542 $$rv .= "[-- Attachment #$idx: ";
543 my $ts = "Type: $ct, Size: $size bytes";
544 my $desc = $part->header('Content-Description') // $fn // '';
545 $desc = ascii_html($desc);
546 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
549 submsg_hdr($ctx, $part) if $part->{is_submsg};
554 sub add_text_body { # callback for each_part
556 my $upfx = $ctx->{mhref};
557 my $ibx = $ctx->{-inbox};
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, $ct);
564 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
566 my $rv = $ctx->{obuf};
567 if ($part->{is_submsg}) {
568 submsg_hdr($ctx, $part);
572 # makes no difference to browsers, and don't screw up filename
573 # link generation in diffs with the extra '%0D'
576 # will be escaped to `•' in HTML
577 obfuscate_addrs($ibx, $s, "\x{2022}") if $ibx->{obfuscate};
579 # always support diff-highlighting, but we can't linkify hunk
580 # headers for solver unless some coderepo are configured:
582 if ($s =~ /^--- [^\n]+\n\+{3} [^\n]+\n@@ /ms) {
583 # diffstat anchors do not link across attachments or messages,
584 # -apfx is just a stable prefix for making diffstat anchors
585 # linkable to the first diff hunk w/o crossing attachments
586 $idx =~ tr!.!/!; # compatibility with previous versions
587 $ctx->{-apfx} = $upfx . $idx;
589 # do attr => filename mappings for diffstats in git diffs:
590 $ctx->{-anchors} = {} if $s =~ /^diff --git /sm;
592 delete $ctx->{-long_path};
594 if ($ibx->{-repo_objs}) {
595 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
597 $spfx =~ s!/([^/]*)/\z!/!;
599 my $n_slash = $upfx =~ tr!/!/!;
602 } elsif ($n_slash == 1) {
604 } else { # nslash == 2
609 $ctx->{-spfx} = $spfx;
612 # some editors don't put trailing newlines at the end:
613 $s .= "\n" unless $s =~ /\n\z/s;
615 # split off quoted and unquoted blocks:
616 my @sections = PublicInbox::MsgIter::split_quotes($s);
617 undef $s; # free memory
618 if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
619 # badly-encoded message with $err? tell the world about it!
620 attach_link($ctx, $ct, $p, $fn, $err);
623 foreach my $cur (@sections) {
625 # we use a <span> here to allow users to specify
626 # their own color for quoted text
627 $$rv .= qq(<span\nclass="q">);
628 $$rv .= $l->to_html($cur);
631 flush_diff($ctx, \$cur);
634 $$rv .= $l->to_html($cur);
636 undef $cur; # free memory
640 sub _msg_page_prepare_obuf {
641 my ($eml, $ctx) = @_;
642 my $over = $ctx->{-inbox}->over;
643 my $obfs_ibx = $ctx->{-obfs_ibx};
645 my $mids = mids_for_index($eml);
646 my $nr = $ctx->{nr}++;
647 if ($nr) { # unlikely
650 $ctx->{first_hdr} = $eml->header_obj;
653 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
655 $rv .= "<pre\nid=b>"; # anchor for body start
657 $ctx->{-upfx} = '../' if $over;
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 .= $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 $over;
691 $title[0] = '(no subject)';
693 for my $v ($eml->header('Date')) {
695 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
698 if (!$nr) { # first (and only) message, common case
699 $ctx->{-title_html} = join(' - ', @title);
700 $rv = $ctx->html_top . $rv;
702 if (scalar(@$mids) == 1) { # common case
703 my $mhtml = ascii_html($mids->[0]);
704 $rv .= "Message-ID: <$mhtml> ";
705 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
707 # X-Alt-Message-ID can happen if a message is injected from
708 # public-inbox-nntpd because of multiple Message-ID headers.
709 my $lnk = PublicInbox::Linkify->new;
711 for my $h (qw(Message-ID X-Alt-Message-ID)) {
712 $s .= "$h: $_\n" for ($eml->header_raw($h));
714 $lnk->linkify_mids('..', \$s, 1);
717 $rv .= _parent_headers($eml, $over);
723 qq(expand[<a\nhref="T/#u">flat</a>) .
724 qq(|<a\nhref="t/#u">nested</a>] ) .
725 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
726 qq(<a\nhref="t.atom">Atom feed</a>);
729 sub thread_skel ($$$) {
730 my ($skel, $ctx, $hdr) = @_;
731 my $mid = mids($hdr)->[0];
732 my $ibx = $ctx->{-inbox};
733 my ($nr, $msgs) = $ibx->over->get_thread($mid);
734 my $parent = in_reply_to($hdr);
735 $$skel .= "\n<b>Thread overview: </b>";
737 if (defined $parent) {
738 $$skel .= SKEL_EXPAND."\n ";
739 $$skel .= ghost_parent('../', $parent) . "\n";
741 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
743 $ctx->{next_msg} = undef;
744 $ctx->{parent_msg} = $parent;
749 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
751 # nb: mutt only shows the first Subject in the index pane
752 # when multiple Subject: headers are present, so we follow suit:
753 my $subj = $hdr->header('Subject') // '';
754 $subj = '(no subject)' if $subj eq '';
755 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
757 $ctx->{prev_attr} = '';
758 $ctx->{prev_level} = 0;
759 $ctx->{skel} = $skel;
761 # reduce hash lookups in skel_dump
762 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
763 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
765 $ctx->{parent_msg} = $parent;
768 sub _parent_headers {
769 my ($hdr, $over) = @_;
771 my @irt = $hdr->header_raw('In-Reply-To');
774 my $lnk = PublicInbox::Linkify->new;
775 $rv .= "In-Reply-To: $_\n" for @irt;
776 $lnk->linkify_mids('..', \$rv);
778 $refs = references($hdr);
779 my $irt = pop @$refs;
781 my $html = ascii_html($irt);
782 my $href = mid_href($irt);
783 $rv .= "In-Reply-To: <";
784 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
788 # do not display References: if search is present,
789 # we show the thread skeleton at the bottom, instead.
792 $refs //= references($hdr);
794 @$refs = map { linkify_ref_no_over($_) } @$refs;
795 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
800 # returns a string buffer
802 my ($ctx, $hdr) = @_;
803 my $ibx = $ctx->{-inbox};
808 my $t = ts2str($ctx->{-t_max});
809 my $t_fmt = fmt_ts($ctx->{-t_max});
812 href="$upfx?t=$t">~$t_fmt UTC</a>|<a
813 href="$upfx">newest</a>]
816 thread_skel(\$skel, $ctx, $hdr);
821 if (my $n = $ctx->{next_msg}) {
823 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
826 my $par = $ctx->{parent_msg};
831 if (my $p = $ctx->{prev_msg}) {
832 $prev = mid_href($p);
833 if ($p && $par && $p eq $par) {
834 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
835 'rel=prev>prev parent</a>';
838 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
840 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
842 } elsif ($u) { # unlikely
843 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
845 $rv .= "$next $prev$parent ";
846 } else { # unindexed inboxes w/o over
847 $skel = qq( <a\nhref="$upfx">latest</a>);
849 $rv .= qq(<a\nhref="#R">reply</a>);
852 $rv .= msg_reply($ctx, $hdr);
855 sub linkify_ref_no_over {
857 my $href = mid_href($mid);
858 my $html = ascii_html($mid);
859 "<<a\nhref=\"../$href/\">$html</a>>";
864 'm' . id_compress($msgid, 1);
868 my ($upfx, $mid) = @_;
870 my $href = mid_href($mid);
871 my $html = ascii_html($mid);
872 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
877 $level ? INDENT x ($level - 1) : '';
881 my ($ctx, $level, $node, $idx) = @_;
882 ++$ctx->{root_idx} if $level == 0;
883 if ($node->{mid} eq $ctx->{mid}) {
884 $ctx->{found_mid_at} = $ctx->{root_idx};
885 return 0; # stop iterating
890 sub strict_loose_note ($) {
893 " -- strict thread matches above, loose matches on Subject: below --\n";
895 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
897 " -- use mbox.gz link to download all $nr messages --\n";
903 my ($ctx, $msgs) = @_;
904 require PublicInbox::SearchThread;
905 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
907 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
908 # and preserved in the future. This bug is hidden by `sid' matches
909 # in get_thread, so we never noticed it until now. And even when
910 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
911 # bump since reindexing is expensive and users may not do it
913 # loose threading could've returned too many results,
914 # put the root the message we care about at the top:
915 my $mid = $ctx->{mid};
916 if (defined($mid) && scalar(@$rootset) > 1) {
917 $ctx->{root_idx} = -1;
918 my $nr = scalar @$msgs;
919 walk_thread($rootset, $ctx, \&find_mid_root);
920 my $idx = $ctx->{found_mid_at};
921 if (defined($idx) && $idx != 0) {
922 my $tip = splice(@$rootset, $idx, 1);
923 @$rootset = reverse @$rootset;
924 unshift @$rootset, $tip;
925 $ctx->{sl_note} = strict_loose_note($nr);
933 require PublicInbox::ExtMsg;
934 PublicInbox::ExtMsg::ext_msg($ctx);
938 my ($prev_subj, $subj, $val) = @_;
940 my $omit; # '"' denotes identical text omitted
941 my (@prev_pop, @curr_pop);
942 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
943 push(@prev_pop, pop(@$prev_subj));
944 push(@curr_pop, pop(@$subj));
947 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
948 if (scalar(@curr_pop) == 1) {
950 push @$prev_subj, @prev_pop;
951 push @$subj, @curr_pop;
956 sub skel_dump { # walk_thread callback
957 my ($ctx, $level, $smsg) = @_;
958 $smsg->{blob} or return _skel_ghost($ctx, $level, $smsg);
960 my $skel = $ctx->{skel};
961 my $cur = $ctx->{cur};
962 my $mid = $smsg->{mid};
964 if ($level == 0 && $ctx->{skel_dump_roots}++) {
965 $$skel .= delete($ctx->{sl_note}) || '';
968 my $f = ascii_html($smsg->{from_name});
969 my $obfs_ibx = $ctx->{-obfs_ibx};
970 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
972 my $d = fmt_ts($smsg->{ds});
973 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
974 if (exists $ctx->{searchview}) {
975 if (defined(my $pct = $smsg->{pct})) {
976 $d .= (sprintf(' % 2u', $pct) . '%');
982 $d .= ' ' . indent_for($level) . th_pfx($level);
984 $ctx->{first_level} ||= $level;
986 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
987 $ctx->{prev_attr} = $attr;
989 $ctx->{prev_level} = $level;
994 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
995 "$attr [this message]</a></b>\n";
998 $ctx->{prev_msg} = $mid;
1001 $ctx->{next_msg} ||= $mid;
1004 # Subject is never undef, this mail was loaded from
1005 # our Xapian which would've resulted in '' if it were
1006 # really missing (and Filter rejects empty subjects)
1007 my @subj = split(/ /, subject_normalized($smsg->{subject}));
1008 # remove common suffixes from the subject if it matches the previous,
1009 # so we do not show redundant text at the end.
1010 my $prev_subj = $ctx->{prev_subj} || [];
1011 $ctx->{prev_subj} = [ @subj ];
1012 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
1015 my $subj = join(' ', @subj);
1016 $subj = ascii_html($subj);
1017 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1018 $end = "$subj</a> $omit$f\n"
1024 my $mapping = $unmatched ? undef : $ctx->{mapping};
1026 my $map = $mapping->{$mid};
1027 $id = id_compress($mid, 1);
1029 $map->[0] = "$d<a\nhref=\"$m\">$end";
1032 $m = $ctx->{-upfx}.mid_href($mid).'/';
1034 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1039 my ($ctx, $level, $node) = @_;
1041 my $mid = $node->{mid};
1042 my $d = ' [not found] ';
1043 $d .= ' ' if exists $ctx->{searchview};
1044 $d .= indent_for($level) . th_pfx($level);
1045 my $upfx = $ctx->{-upfx};
1046 my $href = $upfx . mid_href($mid) . '/';
1047 my $html = ascii_html($mid);
1049 my $mapping = $ctx->{mapping};
1050 my $map = $mapping->{$mid} if $mapping;
1052 my $id = id_compress($mid, 1);
1053 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1054 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1056 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1058 ${$ctx->{skel}} .= $d;
1064 (eval { $a->topmost->{ds} } || 0) <=>
1065 (eval { $b->topmost->{ds} } || 0)
1069 # accumulate recent topics if search is supported
1070 # returns 200 if done, 404 if not
1071 sub acc_topic { # walk_thread callback
1072 my ($ctx, $level, $smsg) = @_;
1073 my $mid = $smsg->{mid};
1074 my $has_blob = $smsg->{blob} // do {
1075 if (my $by_mid = $ctx->{-inbox}->smsg_by_mid($mid)) {
1076 %$smsg = (%$smsg, %$by_mid);
1081 my $subj = subject_normalized($smsg->{subject});
1082 $subj = '(no subject)' if $subj eq '';
1083 my $ds = $smsg->{ds};
1084 if ($level == 0) { # new, top-level topic
1085 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1086 $ctx->{-cur_topic} = $topic;
1087 push @{$ctx->{order}}, $topic;
1091 # continue existing topic
1092 my $topic = $ctx->{-cur_topic}; # should never be undef
1093 $topic->[0] = $ds if $ds > $topic->[0];
1094 $topic->[1]++; # bump N+ message counter
1095 my $seen = $topic->[2];
1096 if (scalar(@$topic) == 3) { # parent was a ghost
1097 push @$topic, $subj;
1098 } elsif (!defined($seen->{$subj})) {
1099 push @$topic, $level, $subj; # @extra messages
1101 $seen->{$subj} = $mid; # latest for subject
1102 } else { # ghost message
1103 return 1 if $level != 0; # ignore child ghosts
1104 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1105 push @{$ctx->{order}}, $topic;
1112 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1114 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1119 my $ibx = $ctx->{-inbox};
1120 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1122 # sort by recency, this allows new posts to "bump" old topics...
1123 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1124 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1126 next unless defined $top_subj; # ghost topic
1127 my $mid = delete $seen->{$top_subj};
1128 my $href = mid_href($mid);
1129 my $prev_subj = [ split(/ /, $top_subj) ];
1130 $top_subj = ascii_html($top_subj);
1133 # $n isn't the total number of posts on the topic,
1134 # just the number of posts in the current results window
1138 $anchor = '#u'; # top of only message
1140 $n = " ($n+ messages)";
1141 $anchor = '#t'; # thread skeleton
1144 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1145 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1146 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1147 " $ds UTC $n - $mbox / $atom\n";
1148 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1149 my $level = $extra[$i];
1150 my $subj = $extra[$i + 1]; # already normalized
1151 $mid = delete $seen->{$subj};
1152 my @subj = split(/ /, $subj);
1153 my @next_prev = @subj; # full copy
1154 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1155 $prev_subj = \@next_prev;
1156 $subj = join(' ', @subj);
1157 $subj = ascii_html($subj);
1158 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1159 $href = mid_href($mid);
1160 $s .= indent_for($level) . TCHILD;
1161 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1165 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1170 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1171 timegm($ss || 0, $mm || 0, $hh || 0, $dd, $mon - 1, $yyyy);
1174 sub pagination_footer ($$) {
1175 my ($ctx, $latest) = @_;
1176 my $next = $ctx->{next_page} || '';
1177 my $prev = $ctx->{prev_page} || '';
1178 if ($prev) { # aligned padding for: 'next (older) | '
1179 $next = $next ? "$next | " : ' | ';
1180 $prev .= qq[ | <a\nhref="$latest">latest</a>];
1182 "<hr><pre>page: $next$prev</pre>";
1185 sub paginate_recent ($$) {
1186 my ($ctx, $lim) = @_;
1187 my $t = $ctx->{qp}->{t} || '';
1188 my $opts = { limit => $lim };
1189 my ($after, $before);
1191 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1192 # if only $after exists "YYYYMMDD.." because "." could be skipped
1193 # if interpreted as an end-of-sentence
1194 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1195 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1197 my $ibx = $ctx->{-inbox};
1198 my $msgs = $ibx->recent($opts, $after, $before);
1199 my $nr = scalar @$msgs;
1200 if ($nr < $lim && defined($after)) {
1201 $after = $before = undef;
1202 $msgs = $ibx->recent($opts);
1203 $nr = scalar @$msgs;
1205 my $more = $nr == $lim;
1206 my ($newest, $oldest);
1208 $newest = $msgs->[0]->{ts};
1209 $oldest = $msgs->[-1]->{ts};
1210 # if we only had $after, our SQL query in ->recent ordered
1211 if ($newest < $oldest) {
1212 ($oldest, $newest) = ($newest, $oldest);
1213 $more = 0 if defined($after) && $after < $oldest;
1216 if (defined($oldest) && $more) {
1217 my $s = ts2str($oldest);
1218 $ctx->{next_page} = qq[<a\nhref="?t=$s"\nrel=next>] .
1221 if (defined($newest) && (defined($before) || defined($after))) {
1222 my $s = ts2str($newest);
1223 $ctx->{prev_page} = qq[<a\nhref="?t=$s-"\nrel=prev>] .
1229 # GET /$INBOX - top-level inbox view for indexed inboxes
1232 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1234 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1236 html_oneshot($ctx, dump_topics($ctx), \pagination_footer($ctx, '.'));
1240 sub thread_adj_level {
1241 my ($ctx, $level) = @_;
1243 my $max = $ctx->{cur_level};
1245 return ('', '') if $max == 0; # flat output
1247 # reset existing lists
1248 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1249 $ctx->{cur_level} = 0;
1251 } elsif ($level == $max) { # continue existing list
1253 } elsif ($level < $max) {
1254 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1255 $ctx->{cur_level} = $level;
1256 ("$beg<li>", '</li>');
1257 } else { # ($level > $max) # start a new level
1258 $ctx->{cur_level} = $level;
1259 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1264 sub ghost_index_entry {
1265 my ($ctx, $level, $node) = @_;
1266 my ($beg, $end) = thread_adj_level($ctx, $level);
1267 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{mid})