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 PublicInbox::MsgTime qw(msg_datestamp);
11 use PublicInbox::Hval qw(ascii_html obfuscate_addrs prurl);
12 use PublicInbox::Linkify;
13 use PublicInbox::MID qw/id_compress mid_escape mids mids_for_index references/;
14 use PublicInbox::MsgIter;
15 use PublicInbox::Address;
16 use PublicInbox::WwwStream;
17 use PublicInbox::Reply;
18 use PublicInbox::ViewDiff qw(flush_diff);
19 use POSIX qw(strftime);
20 use Time::Local qw(timegm);
21 use PublicInbox::SearchMsg qw(subject_normalized);
22 use constant COLS => 72;
23 use constant INDENT => ' ';
24 use constant TCHILD => '` ';
25 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
29 my $more = $ctx->{more};
31 # $more cannot be true w/o $smsg being defined:
32 $ctx->{mhref} = $more ? '../'.mid_escape($ctx->{smsg}->mid).'/'
34 multipart_text_as_html(delete $ctx->{mime}, $ctx);
35 ${delete $ctx->{obuf}} .= '</pre><hr>';
38 # fake an EOF if {more} retrieval fails fails;
39 eval { msg_page_more($ctx, $nr) };
40 } elsif ($nr == $ctx->{end_nr}) {
41 # fake an EOF if generating the footer fails;
42 # we want to at least show the message if something
44 eval { html_footer($ctx) };
50 # public functions: (unstable)
54 my $mid = $ctx->{mid};
55 my $ibx = $ctx->{-inbox};
58 if (my $over = $ibx->over) {
60 $smsg = $over->next_by_mid($mid, \$id, \$prev);
61 $first = $ibx->msg_by_smsg($smsg) if $smsg;
63 my $next = $over->next_by_mid($mid, \$id, \$prev);
64 $ctx->{more} = [ $id, $prev, $next ] if $next;
68 $first = $ibx->msg_by_mid($mid) or return;
70 my $mime = $ctx->{mime} = PublicInbox::MIME->new($first);
71 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
72 my $hdr = $ctx->{hdr} = $mime->header_obj;
73 _msg_page_prepare_obuf($hdr, $ctx, 0);
76 PublicInbox::WwwStream->response($ctx, 200, \&msg_page_i);
81 my ($id, $prev, $smsg) = @{$ctx->{more}};
82 my $ibx = $ctx->{-inbox};
83 $smsg = $ibx->smsg_mime($smsg);
84 my $next = $ibx->over->next_by_mid($ctx->{mid}, \$id, \$prev);
85 $ctx->{more} = $next ? [ $id, $prev, $next ] : undef;
86 return '' unless $smsg;
87 $ctx->{mhref} = '../' . mid_escape($smsg->{mid}) . '/';
88 my $mime = delete $smsg->{mime};
89 _msg_page_prepare_obuf($mime->header_obj, $ctx, $nr);
90 multipart_text_as_html($mime, $ctx);
91 ${delete $ctx->{obuf}} .= '</pre><hr>';
94 # /$INBOX/$MESSAGE_ID/#R
98 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
100 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
103 my $ibx = $ctx->{-inbox};
104 if (my $url = $ibx->{infourl}) {
105 $url = prurl($ctx->{env}, $url);
106 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
109 my ($arg, $link, $reply_to_all) =
110 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
111 if (ref($arg) eq 'SCALAR') {
112 return '<pre id=R>'.ascii_html($$arg).'</pre>';
115 # mailto: link only works if address obfuscation is disabled
119 * If your mail client supports setting the <b>In-Reply-To</b> header
120 via mailto: links, try the <a
121 href="$link">mailto: link</a>
125 push @$arg, '/path/to/YOUR_REPLY';
126 $arg = ascii_html(join(" \\\n ", '', @$arg));
129 id=R><b>Reply instructions:</b>
131 You may reply publicly to <a
132 href=#t>this message</a> via plain-text email
133 using any one of the following methods:
135 * Save the following mbox file, import it into your mail client,
136 and $reply_to_all from there: <a
139 Avoid top-posting and favor interleaved quoting:
141 href="$p_url">$p_url</a>
143 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
144 switches of git-send-email(1):
149 href="$se_url">$se_url</a>
156 my $refs = references($hdr);
160 sub fold_addresses ($) {
161 return $_[0] if length($_[0]) <= COLS;
162 # try to fold on commas after non-word chars before $lim chars,
163 # Try to get the "," preceeded by ">" or ")", but avoid folding
164 # on the comma where somebody uses "Lastname, Firstname".
165 # We also try to keep the last and penultimate addresses in
166 # the list on the same line if possible, hence the extra \z
167 # Fall back to folding on spaces at $lim + 1 chars
168 my $lim = COLS - 8; # 8 = "\t" display width
169 my $too_long = $lim + 1;
170 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
172 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
175 .{$too_long,}?)(?:\s|\z)/xgo));
178 sub _hdr_names_html ($$) {
179 my ($hdr, $field) = @_;
180 my @vals = $hdr->header($field) or return '';
181 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
185 my ($nr, $singular, $plural) = @_;
186 return "0 $plural" if $nr == 0;
187 $nr == 1 ? "$nr $singular" : "$nr $plural";
190 # human-friendly format
191 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
193 # this is already inside a <pre>
195 my ($smsg, $ctx, $more) = @_;
196 my $subj = $smsg->subject;
197 my $mid_raw = $smsg->mid;
198 my $id = id_compress($mid_raw, 1);
201 my $root_anchor = $ctx->{root_anchor} || '';
203 my $obfs_ibx = $ctx->{-obfs_ibx};
205 $subj = '(no subject)' if $subj eq '';
206 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
207 $subj = '<b>'.ascii_html($subj).'</b>';
208 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
209 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
211 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
213 my $ds = $smsg->ds; # for v1 non-Xapian/SQLite users
214 # deleting {mime} is critical to memory use,
215 # the rest of the fields saves about 400K as we iterate across 1K msgs
216 my ($mime) = delete @$smsg{qw(mime ds ts blob subject)};
218 my $hdr = $mime->header_obj;
219 my $from = _hdr_names_html($hdr, 'From');
220 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
221 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
222 my $upfx = $ctx->{-upfx};
223 my $mhref = $upfx . mid_escape($mid_raw) . '/';
224 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
225 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
226 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
227 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
228 my ($tlen, $clen) = (length($to), length($cc));
230 if (($tlen + $clen) > COLS) {
231 $to_cc .= ' To: '.$to."\n" if $tlen;
232 $to_cc .= ' Cc: '.$cc."\n" if $clen;
235 $to_cc .= ' To: '.$to;
236 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
238 $to_cc .= ' Cc: '.$cc if $clen;
242 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
245 my $mapping = $ctx->{mapping};
246 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
247 my $mirt = PublicInbox::Hval->new_msgid($irt);
248 my $href = $upfx . $mirt->{href}. '/';
249 my $html = $mirt->as_html;
250 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
254 # scan through all parts, looking for displayable text
255 $ctx->{mhref} = $mhref;
257 msg_iter($mime, \&add_text_body, $ctx, 1);
261 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
262 "<a\nhref=\"$mhref\">permalink</a>" .
263 " <a\nhref=\"${mhref}raw\">raw</a>" .
264 " <a\nhref=\"${mhref}#R\">reply</a>";
267 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
268 $rv .= "\t[relevance $pct%]";
271 my $nested = 'nested';
276 $flat = "<b>$flat</b>";
278 $nested = "<b>$nested</b>";
280 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
281 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
282 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
287 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
291 sub pad_link ($$;$) {
292 my ($mid, $level, $s) = @_;
294 my $id = id_compress($mid, 1);
295 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
299 my ($mid_raw, $irt, $id, $ctx) = @_;
301 my $mapping = $ctx->{mapping} or return $rv;
303 my $mid_map = $mapping->{$mid_raw};
305 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
306 my ($attr, $node, $idx, $level) = @$mid_map;
307 my $children = $node->{children};
308 my $nr_c = scalar @$children;
311 if (my $smsg = $node->{smsg}) {
312 # delete saves about 200KB on a 1K message thread
313 if (my $refs = delete $smsg->{references}) {
314 ($$irt) = ($refs =~ m/<([^>]+)>\z/);
317 my $irt_map = $mapping->{$$irt} if defined $$irt;
318 if (defined $irt_map) {
319 $siblings = $irt_map->[1]->{children};
320 $nr_s = scalar(@$siblings) - 1;
321 $rv .= $pad . $irt_map->[0];
323 my $prev = $siblings->[$idx - 1];
324 my $pmid = $prev->{id};
326 my $s = ($idx - 1). ' preceding siblings ...';
327 $rv .= pad_link($pmid, $level, $s);
328 } elsif ($idx == 2) {
329 my $ppmid = $siblings->[0]->{id};
330 $rv .= $pad . $mapping->{$ppmid}->[0];
332 $rv .= $pad . $mapping->{$pmid}->[0];
335 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
336 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
337 $attr =~ s!\n\z!</b>\n!s;
338 $attr =~ s!<a\nhref.*</a> !!s; # no point in duplicating subject
339 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
342 my $cmid = $children->[0]->{id};
343 $rv .= $pad . $mapping->{$cmid}->[0];
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 . $mapping->{$cn->{id}}->[0];
352 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
354 my $nmid = $next->{id};
355 $rv .= $pad . $mapping->{$nmid}->[0];
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 . $mapping->{$nn->{id}}->[0];
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 ($$$) {
369 my ($rootset, $ctx, $cb) = @_;
370 my @q = map { (0, $_, -1) } @$rootset;
372 my ($level, $node, $i) = splice(@q, 0, 3);
373 defined $node or next;
374 $cb->($ctx, $level, $node, $i) or return;
377 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
381 sub pre_thread { # walk_thread callback
382 my ($ctx, $level, $node, $idx) = @_;
383 $ctx->{mapping}->{$node->{id}} = [ '', $node, $idx, $level ];
384 skel_dump($ctx, $level, $node);
387 sub thread_index_entry {
388 my ($ctx, $level, $smsg) = @_;
389 my ($beg, $end) = thread_adj_level($ctx, $level);
390 $beg . '<pre>' . index_entry($smsg, $ctx, 0) . '</pre>' . $end;
393 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
395 return unless exists($ctx->{skel});
396 my $q = $ctx->{-queue};
398 my $level = shift @$q;
399 my $node = shift @$q or next;
401 unshift @$q, map { ($cl, $_) } @{$node->{children}};
402 if (my $smsg = $ctx->{-inbox}->smsg_mime($node->{smsg})) {
403 return thread_index_entry($ctx, $level, $smsg);
405 return ghost_index_entry($ctx, $level, $node);
408 join('', thread_adj_level($ctx, 0)) . ${delete $ctx->{skel}};
411 sub stream_thread ($$) {
412 my ($rootset, $ctx) = @_;
413 my $ibx = $ctx->{-inbox};
414 my @q = map { (0, $_) } @$rootset;
418 my $node = shift @q or next;
420 unshift @q, map { ($cl, $_) } @{$node->{children}};
421 $smsg = $ibx->smsg_mime($node->{smsg}) and last;
423 return missing_thread($ctx) unless $smsg;
425 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
426 $ctx->{-title_html} = ascii_html($smsg->{subject});
427 $ctx->{-html_tip} = thread_index_entry($ctx, $level, $smsg);
428 $ctx->{-queue} = \@q;
429 PublicInbox::WwwStream->response($ctx, 200, \&stream_thread_i);
432 # /$INBOX/$MESSAGE_ID/t/
435 my $mid = $ctx->{mid};
436 my $ibx = $ctx->{-inbox};
437 my ($nr, $msgs) = $ibx->over->get_thread($mid);
438 return missing_thread($ctx) if $nr == 0;
439 my $skel = '<hr><pre>';
440 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
441 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
442 $skel .= "<b\nid=t>Thread overview:</b> ";
443 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
444 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
445 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
446 $skel .= "-- links below jump to the message on this page --\n";
447 $ctx->{-upfx} = '../../';
448 $ctx->{cur_level} = 0;
449 $ctx->{skel} = \$skel;
450 $ctx->{prev_attr} = '';
451 $ctx->{prev_level} = 0;
452 $ctx->{root_anchor} = anchor_for($mid);
453 $ctx->{mapping} = {};
454 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
457 my $rootset = thread_results($ctx, $msgs);
459 # reduce hash lookups in pre_thread->skel_dump
460 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
461 walk_thread($rootset, $ctx, \&pre_thread);
464 return stream_thread($rootset, $ctx) unless $ctx->{flat};
466 # flat display: lazy load the full message from smsg
468 while (my $m = shift @$msgs) {
469 $smsg = $ibx->smsg_mime($m) and last;
471 return missing_thread($ctx) unless $smsg;
472 $ctx->{-title_html} = ascii_html($smsg->{subject});
473 $ctx->{-html_tip} = '<pre>'.index_entry($smsg, $ctx, scalar @$msgs);
474 $ctx->{msgs} = $msgs;
475 PublicInbox::WwwStream->response($ctx, 200, \&thread_html_i);
478 sub thread_html_i { # PublicInbox::WwwStream::getline callback
480 my $msgs = $ctx->{msgs} or return;
481 while (my $smsg = shift @$msgs) {
482 $ctx->{-inbox}->smsg_mime($smsg) or next;
483 return index_entry($smsg, $ctx, scalar @$msgs);
485 my ($skel) = delete @$ctx{qw(skel msgs)};
489 sub multipart_text_as_html {
490 # ($mime, $ctx) = @_; # msg_iter will do "$_[0] = undef"
492 # scan through all parts, looking for displayable text
493 msg_iter($_[0], \&add_text_body, $_[1], 1);
497 my ($s, $l, $quot) = @_;
499 my $rv = $l->to_html($$quot);
501 # we use a <span> here to allow users to specify their own
502 # color for quoted text
504 $$s .= qq(<span\nclass="q">) . $rv . '</span>'
507 sub attach_link ($$$$;$) {
508 my ($ctx, $ct, $p, $fn, $err) = @_;
509 my ($part, $depth, @idx) = @$p;
510 my $nl = $idx[-1] > 1 ? "\n" : '';
511 my $idx = join('.', @idx);
512 my $size = bytes::length($part->body);
514 # hide attributes normally, unless we want to aid users in
515 # spotting MUA problems:
516 $ct =~ s/;.*// unless $err;
517 $ct = ascii_html($ct);
518 my $desc = $part->header('Content-Description');
519 $desc = $fn unless defined $desc;
520 $desc = '' unless defined $desc;
522 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
524 } elsif ($ct eq 'text/plain') {
529 my $rv = $ctx->{obuf};
530 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
532 $$rv .= "[-- Warning: decoded text below may be mangled --]\n";
534 $$rv .= "[-- Attachment #$idx: ";
535 my $ts = "Type: $ct, Size: $size bytes";
536 $desc = ascii_html($desc);
537 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
542 sub add_text_body { # callback for msg_iter
544 my $upfx = $ctx->{mhref};
545 my $ibx = $ctx->{-inbox};
546 # $p - from msg_iter: [ Email::MIME, depth, @idx ]
547 my ($part, $depth, @idx) = @$p;
548 my $ct = $part->content_type || 'text/plain';
549 my $fn = $part->filename;
550 my ($s, $err) = msg_part_text($part, $ct);
551 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
553 # makes no difference to browsers, and don't screw up filename
554 # link generation in diffs with the extra '%0D'
557 # always support diff-highlighting, but we can't linkify hunk
558 # headers for solver unless some coderepo are configured:
560 if ($s =~ /^(?:diff|---|\+{3}) /ms) {
561 # diffstat anchors do not link across attachments or messages:
562 $idx[0] = $upfx . $idx[0] if $upfx ne '';
563 $ctx->{-apfx} = join('/', @idx);
564 $ctx->{-anchors} = {}; # attr => filename
566 delete $ctx->{-long_path};
568 if ($ibx->{-repo_objs}) {
569 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
571 $spfx =~ s!/([^/]*)/\z!/!;
573 my $n_slash = $upfx =~ tr!/!/!;
576 } elsif ($n_slash == 1) {
578 } else { # nslash == 2
583 $ctx->{-spfx} = $spfx;
586 # some editors don't put trailing newlines at the end:
587 $s .= "\n" unless $s =~ /\n\z/s;
589 # split off quoted and unquoted blocks:
590 my @sections = split(/((?:^>[^\n]*\n)+)/sm, $s);
592 my $rv = $ctx->{obuf};
593 if (defined($fn) || $depth > 0 || $err) {
594 # badly-encoded message with $err? tell the world about it!
595 attach_link($ctx, $ct, $p, $fn, $err);
598 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
599 foreach my $cur (@sections) {
601 flush_quote($rv, $l, \$cur);
603 flush_diff($rv, $ctx, \$cur);
606 $$rv .= $l->to_html($cur);
611 obfuscate_addrs($ibx, $$rv) if $ibx->{obfuscate};
614 sub _msg_page_prepare_obuf {
615 my ($hdr, $ctx, $nr) = @_;
616 my $over = $ctx->{-inbox}->over;
617 my $obfs_ibx = $ctx->{-obfs_ibx};
619 my $mids = mids_for_index($hdr);
623 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
625 $rv .= "<pre\nid=b>"; # anchor for body start
630 $ctx->{-upfx} = '../';
632 my @title; # (Subject[0], From[0])
633 for my $v ($hdr->header('From')) {
634 $v = PublicInbox::Hval->new($v);
635 my @n = PublicInbox::Address::names($v->raw);
636 $title[1] //= ascii_html(join(', ', @n));
639 obfuscate_addrs($obfs_ibx, $v);
640 obfuscate_addrs($obfs_ibx, $title[1]);
642 $rv .= "From: $v\n" if $v ne '';
644 foreach my $h (qw(To Cc)) {
645 for my $v ($hdr->header($h)) {
648 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
649 $rv .= "$h: $v\n" if $v ne '';
652 my @subj = $hdr->header('Subject');
654 my $v = ascii_html(shift @subj);
655 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
657 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
659 for $v (@subj) { # multi-Subject message :<
661 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
662 $rv .= "Subject: $v\n";
664 } else { # dummy anchor for thread skeleton at bottom of page
665 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
666 $title[0] = '(no subject)';
668 for my $v ($hdr->header('Date')) {
670 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
673 $ctx->{-title_html} = join(' - ', @title);
674 if (scalar(@$mids) == 1) { # common case
675 my $mid = PublicInbox::Hval->new_msgid($mids->[0]);
676 my $mhtml = $mid->as_html;
677 $rv .= "Message-ID: <$mhtml> ";
678 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
680 # X-Alt-Message-ID can happen if a message is injected from
681 # public-inbox-nntpd because of multiple Message-ID headers.
682 my $lnk = PublicInbox::Linkify->new;
684 for my $h (qw(Message-ID X-Alt-Message-ID)) {
685 $s .= "$h: $_\n" for ($hdr->header_raw($h));
687 $lnk->linkify_mids('..', \$s, 1);
690 $rv .= _parent_headers($hdr, $over);
696 qq(expand[<a\nhref="T/#u">flat</a>) .
697 qq(|<a\nhref="t/#u">nested</a>] ) .
698 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
699 qq(<a\nhref="t.atom">Atom feed</a>);
702 sub thread_skel ($$$) {
703 my ($skel, $ctx, $hdr) = @_;
704 my $mid = mids($hdr)->[0];
705 my $ibx = $ctx->{-inbox};
706 my ($nr, $msgs) = $ibx->over->get_thread($mid);
707 my $parent = in_reply_to($hdr);
708 $$skel .= "\n<b>Thread overview: </b>";
710 if (defined $parent) {
711 $$skel .= SKEL_EXPAND."\n ";
712 $$skel .= ghost_parent('../', $parent) . "\n";
714 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
716 $ctx->{next_msg} = undef;
717 $ctx->{parent_msg} = $parent;
722 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
724 # nb: mutt only shows the first Subject in the index pane
725 # when multiple Subject: headers are present, so we follow suit:
726 my $subj = $hdr->header('Subject') // '';
727 $subj = '(no subject)' if $subj eq '';
728 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
730 $ctx->{prev_attr} = '';
731 $ctx->{prev_level} = 0;
732 $ctx->{skel} = $skel;
734 # reduce hash lookups in skel_dump
735 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
736 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
738 $ctx->{parent_msg} = $parent;
741 sub _parent_headers {
742 my ($hdr, $over) = @_;
744 my @irt = $hdr->header_raw('In-Reply-To');
747 my $lnk = PublicInbox::Linkify->new;
748 $rv .= "In-Reply-To: $_\n" for @irt;
749 $lnk->linkify_mids('..', \$rv);
751 $refs = references($hdr);
752 my $irt = pop @$refs;
754 my $v = PublicInbox::Hval->new_msgid($irt);
755 my $html = $v->as_html;
756 my $href = $v->{href};
757 $rv .= "In-Reply-To: <";
758 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
762 # do not display References: if search is present,
763 # we show the thread skeleton at the bottom, instead.
766 $refs //= references($hdr);
768 @$refs = map { linkify_ref_no_over($_) } @$refs;
769 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
774 # returns a string buffer via ->getline
777 my $ibx = $ctx->{-inbox};
778 my $hdr = delete $ctx->{hdr};
780 my $skel = " <a\nhref=\"$upfx\">index</a>";
784 thread_skel(\$skel, $ctx, $hdr);
789 if (my $n = $ctx->{next_msg}) {
790 $n = PublicInbox::Hval->new_msgid($n)->{href};
791 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
794 my $par = $ctx->{parent_msg};
796 $u = PublicInbox::Hval->new_msgid($par)->{href};
799 if (my $p = $ctx->{prev_msg}) {
800 $prev = PublicInbox::Hval->new_msgid($p)->{href};
801 if ($p && $par && $p eq $par) {
802 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
803 'rel=prev>prev parent</a>';
806 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
808 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
810 } elsif ($u) { # unlikely
811 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
813 $rv .= "$next $prev$parent ";
815 $rv .= qq(<a\nhref="#R">reply</a>);
818 $rv .= msg_reply($ctx, $hdr);
821 sub linkify_ref_no_over {
822 my $v = PublicInbox::Hval->new_msgid($_[0]);
823 my $html = $v->as_html;
824 my $href = $v->{href};
825 "<<a\nhref=\"../$href/\">$html</a>>";
830 'm' . id_compress($msgid, 1);
834 my ($upfx, $mid) = @_;
836 $mid = PublicInbox::Hval->new_msgid($mid);
837 my $href = $mid->{href};
838 my $html = $mid->as_html;
839 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
844 $level ? INDENT x ($level - 1) : '';
848 my ($ctx, $level, $node, $idx) = @_;
849 ++$ctx->{root_idx} if $level == 0;
850 if ($node->{id} eq $ctx->{mid}) {
851 $ctx->{found_mid_at} = $ctx->{root_idx};
857 sub strict_loose_note ($) {
860 " -- strict thread matches above, loose matches on Subject: below --\n";
862 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
864 " -- use mbox.gz link to download all $nr messages --\n";
870 my ($ctx, $msgs) = @_;
871 require PublicInbox::SearchThread;
872 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
874 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
875 # and preserved in the future. This bug is hidden by `sid' matches
876 # in get_thread, so we never noticed it until now. And even when
877 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
878 # bump since reindexing is expensive and users may not do it
880 # loose threading could've returned too many results,
881 # put the root the message we care about at the top:
882 my $mid = $ctx->{mid};
883 if (defined($mid) && scalar(@$rootset) > 1) {
884 $ctx->{root_idx} = -1;
885 my $nr = scalar @$msgs;
886 walk_thread($rootset, $ctx, \&find_mid_root);
887 my $idx = $ctx->{found_mid_at};
888 if (defined($idx) && $idx != 0) {
889 my $tip = splice(@$rootset, $idx, 1);
890 @$rootset = reverse @$rootset;
891 unshift @$rootset, $tip;
892 $ctx->{sl_note} = strict_loose_note($nr);
900 require PublicInbox::ExtMsg;
901 PublicInbox::ExtMsg::ext_msg($ctx);
905 my ($prev_subj, $subj, $val) = @_;
907 my $omit = ''; # '"' denotes identical text omitted
908 my (@prev_pop, @curr_pop);
909 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
910 push(@prev_pop, pop(@$prev_subj));
911 push(@curr_pop, pop(@$subj));
914 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
915 if (scalar(@curr_pop) == 1) {
917 push @$prev_subj, @prev_pop;
918 push @$subj, @curr_pop;
923 sub skel_dump { # walk_thread callback
924 my ($ctx, $level, $node) = @_;
925 my $smsg = $node->{smsg} or return _skel_ghost($ctx, $level, $node);
927 my $skel = $ctx->{skel};
928 my $cur = $ctx->{cur};
929 my $mid = $smsg->{mid};
931 if ($level == 0 && $ctx->{skel_dump_roots}++) {
932 $$skel .= delete($ctx->{sl_note}) || '';
935 my $f = ascii_html($smsg->{from_name});
936 my $obfs_ibx = $ctx->{-obfs_ibx};
937 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
939 my $d = fmt_ts($smsg->{ds});
940 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
941 if (exists $ctx->{searchview}) {
942 if (defined(my $pct = $smsg->{pct})) {
943 $d .= (sprintf(' % 2u', $pct) . '%');
949 $d .= ' ' . indent_for($level) . th_pfx($level);
951 $ctx->{first_level} ||= $level;
953 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
954 $ctx->{prev_attr} = $attr;
956 $ctx->{prev_level} = $level;
961 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
962 "$attr [this message]</a></b>\n";
965 $ctx->{prev_msg} = $mid;
968 $ctx->{next_msg} ||= $mid;
971 # Subject is never undef, this mail was loaded from
972 # our Xapian which would've resulted in '' if it were
973 # really missing (and Filter rejects empty subjects)
974 my @subj = split(/ /, subject_normalized($smsg->{subject}));
975 # remove common suffixes from the subject if it matches the previous,
976 # so we do not show redundant text at the end.
977 my $prev_subj = $ctx->{prev_subj} || [];
978 $ctx->{prev_subj} = [ @subj ];
979 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
982 my $subj = join(' ', @subj);
983 $subj = ascii_html($subj);
984 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
985 $end = "$subj</a> $omit$f\n"
991 my $mapping = $unmatched ? undef : $ctx->{mapping};
993 my $map = $mapping->{$mid};
994 $id = id_compress($mid, 1);
996 $map->[0] = "$d<a\nhref=\"$m\">$end";
999 $m = $ctx->{-upfx}.mid_escape($mid).'/';
1001 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
1006 my ($ctx, $level, $node) = @_;
1008 my $mid = $node->{id};
1009 my $d = ' [not found] ';
1010 $d .= ' ' if exists $ctx->{searchview};
1011 $d .= indent_for($level) . th_pfx($level);
1012 my $upfx = $ctx->{-upfx};
1013 my $m = PublicInbox::Hval->new_msgid($mid);
1014 my $href = $upfx . $m->{href} . '/';
1015 my $html = $m->as_html;
1017 my $mapping = $ctx->{mapping};
1018 my $map = $mapping->{$mid} if $mapping;
1020 my $id = id_compress($mid, 1);
1021 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1022 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1024 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1026 ${$ctx->{skel}} .= $d;
1032 (eval { $a->topmost->{smsg}->{ds} } || 0) <=>
1033 (eval { $b->topmost->{smsg}->{ds} } || 0)
1037 # accumulate recent topics if search is supported
1038 # returns 200 if done, 404 if not
1039 sub acc_topic { # walk_thread callback
1040 my ($ctx, $level, $node) = @_;
1041 my $mid = $node->{id};
1042 my $smsg = $node->{smsg} // $ctx->{-inbox}->smsg_by_mid($mid);
1044 my $subj = subject_normalized($smsg->{subject});
1045 $subj = '(no subject)' if $subj eq '';
1046 my $ds = $smsg->{ds};
1047 if ($level == 0) { # new, top-level topic
1048 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1049 $ctx->{-cur_topic} = $topic;
1050 push @{$ctx->{order}}, $topic;
1054 # continue existing topic
1055 my $topic = $ctx->{-cur_topic}; # should never be undef
1056 $topic->[0] = $ds if $ds > $topic->[0];
1057 $topic->[1]++; # bump N+ message counter
1058 my $seen = $topic->[2];
1059 if (scalar(@$topic) == 3) { # parent was a ghost
1060 push @$topic, $subj;
1061 } elsif (!defined($seen->{$subj})) {
1062 push @$topic, $level, $subj; # @extra messages
1064 $seen->{$subj} = $mid; # latest for subject
1065 } else { # ghost message
1066 return 1 if $level != 0; # ignore child ghosts
1067 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1068 push @{$ctx->{order}}, $topic;
1075 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1077 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1082 my $ibx = $ctx->{-inbox};
1083 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1085 # sort by recency, this allows new posts to "bump" old topics...
1086 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1087 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1089 next unless defined $top_subj; # ghost topic
1090 my $mid = delete $seen->{$top_subj};
1091 my $href = mid_escape($mid);
1092 my $prev_subj = [ split(/ /, $top_subj) ];
1093 $top_subj = PublicInbox::Hval->new($top_subj)->as_html;
1096 # $n isn't the total number of posts on the topic,
1097 # just the number of posts in the current results window
1101 $anchor = '#u'; # top of only message
1103 $n = " ($n+ messages)";
1104 $anchor = '#t'; # thread skeleton
1107 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1108 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1109 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1110 " $ds UTC $n - $mbox / $atom\n";
1111 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1112 my $level = $extra[$i];
1113 my $subj = $extra[$i + 1]; # already normalized
1114 $mid = delete $seen->{$subj};
1115 my @subj = split(/ /, $subj);
1116 my @next_prev = @subj; # full copy
1117 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1118 $prev_subj = \@next_prev;
1119 $subj = ascii_html($subj);
1120 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1121 $href = mid_escape($mid);
1122 $s .= indent_for($level) . TCHILD;
1123 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1127 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1131 # only for the t= query parameter passed to overview DB
1132 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1135 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1136 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1139 sub pagination_footer ($$) {
1140 my ($ctx, $latest) = @_;
1141 delete $ctx->{qp} or return;
1142 my $next = $ctx->{next_page} || '';
1143 my $prev = $ctx->{prev_page} || '';
1145 $next = $next ? "$next " : ' ';
1146 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1148 "<hr><pre>page: $next$prev</pre>";
1151 sub index_nav { # callback for WwwStream
1152 my (undef, $ctx) = @_;
1153 pagination_footer($ctx, '.')
1156 sub paginate_recent ($$) {
1157 my ($ctx, $lim) = @_;
1158 my $t = $ctx->{qp}->{t} || '';
1159 my $opts = { limit => $lim };
1160 my ($after, $before);
1162 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1163 # if only $after exists "YYYYMMDD.." because "." could be skipped
1164 # if interpreted as an end-of-sentence
1165 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1166 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1168 my $ibx = $ctx->{-inbox};
1169 my $msgs = $ibx->recent($opts, $after, $before);
1170 my $nr = scalar @$msgs;
1171 if ($nr < $lim && defined($after)) {
1172 $after = $before = undef;
1173 $msgs = $ibx->recent($opts);
1174 $nr = scalar @$msgs;
1176 my $more = $nr == $lim;
1177 my ($newest, $oldest);
1179 $newest = $msgs->[0]->{ts};
1180 $oldest = $msgs->[-1]->{ts};
1181 # if we only had $after, our SQL query in ->recent ordered
1182 if ($newest < $oldest) {
1183 ($oldest, $newest) = ($newest, $oldest);
1184 $more = 0 if defined($after) && $after < $oldest;
1187 if (defined($oldest) && $more) {
1188 my $s = ts2str($oldest);
1189 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1191 if (defined($newest) && (defined($before) || defined($after))) {
1192 my $s = ts2str($newest);
1193 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1200 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1202 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1204 PublicInbox::WwwStream->response($ctx, dump_topics($ctx), \&index_nav);
1207 sub thread_adj_level {
1208 my ($ctx, $level) = @_;
1210 my $max = $ctx->{cur_level};
1212 return ('', '') if $max == 0; # flat output
1214 # reset existing lists
1215 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1216 $ctx->{cur_level} = 0;
1218 } elsif ($level == $max) { # continue existing list
1220 } elsif ($level < $max) {
1221 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1222 $ctx->{cur_level} = $level;
1223 ("$beg<li>", '</li>');
1224 } else { # ($level > $max) # start a new level
1225 $ctx->{cur_level} = $level;
1226 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1231 sub ghost_index_entry {
1232 my ($ctx, $level, $node) = @_;
1233 my ($beg, $end) = thread_adj_level($ctx, $level);
1234 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{id})