1 # Copyright (C) 2014-2018 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/;
12 use PublicInbox::Linkify;
13 use PublicInbox::MID qw/id_compress mid_escape mids references/;
14 use PublicInbox::MsgIter;
15 use PublicInbox::Address;
16 use PublicInbox::WwwStream;
17 use PublicInbox::Reply;
18 use PublicInbox::ViewDiff qw(flush_diff);
20 use Time::Local qw(timegm);
22 use constant COLS => 72;
23 use constant INDENT => ' ';
24 use constant TCHILD => '` ';
25 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
27 # public functions: (unstable)
30 my ($ctx, $mime, $more, $smsg) = @_;
31 my $hdr = $mime->header_obj;
32 my $ibx = $ctx->{-inbox};
33 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
34 my $tip = _msg_html_prepare($hdr, $ctx, $more, 0);
36 PublicInbox::WwwStream->response($ctx, 200, sub {
39 # $more cannot be true w/o $smsg being defined:
40 my $upfx = $more ? '../'.mid_escape($smsg->mid).'/' : '';
41 $tip . multipart_text_as_html($mime, $upfx, $ctx) .
43 } elsif ($more && @$more) {
45 msg_html_more($ctx, $more, $nr);
46 } elsif ($nr == $end) {
47 # fake an EOF if generating the footer fails;
48 # we want to at least show the message if something
51 '<pre>' . html_footer($hdr, 1, $ctx) .
52 '</pre>' . msg_reply($ctx, $hdr)
62 my $mid = $ctx->{mid};
63 my $ibx = $ctx->{-inbox};
66 if (my $srch = $ibx->search) {
68 $smsg = $srch->next_by_mid($mid, \$id, \$prev);
69 $first = $ibx->msg_by_smsg($smsg) if $smsg;
71 my $next = $srch->next_by_mid($mid, \$id, \$prev);
72 $more = [ $id, $prev, $next ] if $next;
76 $first = $ibx->msg_by_mid($mid) or return;
78 msg_html($ctx, PublicInbox::MIME->new($first), $more, $smsg);
82 my ($ctx, $more, $nr) = @_;
84 my ($id, $prev, $smsg) = @$more;
85 my $mid = $ctx->{mid};
86 my $ibx = $ctx->{-inbox};
87 $smsg = $ibx->smsg_mime($smsg);
88 my $next = $ctx->{srch}->next_by_mid($mid, \$id, \$prev);
89 @$more = $next ? ($id, $prev, $next) : ();
91 my $mime = $smsg->{mime};
92 my $upfx = '../' . mid_escape($smsg->mid) . '/';
93 _msg_html_prepare($mime->header_obj, $ctx, $more, $nr) .
94 multipart_text_as_html($mime, $upfx, $ctx) .
101 warn "Error lookup up additional messages: $@\n";
102 $str = '<pre>Error looking up additional messages</pre>';
107 # /$INBOX/$MESSAGE_ID/#R
109 my ($ctx, $hdr) = @_;
111 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
113 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
116 my $ibx = $ctx->{-inbox};
117 if (my $url = $ibx->{infourl}) {
118 $url = PublicInbox::Hval::prurl($ctx->{env}, $url);
119 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
122 my ($arg, $link, $reply_to_all) =
123 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
124 if (ref($arg) eq 'SCALAR') {
125 return '<pre id=R>'.ascii_html($$arg).'</pre>';
128 # mailto: link only works if address obfuscation is disabled
132 * If your mail client supports setting the <b>In-Reply-To</b> header
133 via mailto: links, try the <a
134 href="$link">mailto: link</a>
138 push @$arg, '/path/to/YOUR_REPLY';
139 $arg = ascii_html(join(" \\\n ", '', @$arg));
142 id=R><b>Reply instructions:</b>
144 You may reply publically to <a
145 href=#t>this message</a> via plain-text email
146 using any one of the following methods:
148 * Save the following mbox file, import it into your mail client,
149 and $reply_to_all from there: <a
152 Avoid top-posting and favor interleaved quoting:
154 href="$p_url">$p_url</a>
156 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
157 switches of git-send-email(1):
162 href="$se_url">$se_url</a>
169 my $refs = references($hdr);
173 sub fold_addresses ($) {
174 return $_[0] if length($_[0]) <= COLS;
175 # try to fold on commas after non-word chars before $lim chars,
176 # Try to get the "," preceeded by ">" or ")", but avoid folding
177 # on the comma where somebody uses "Lastname, Firstname".
178 # We also try to keep the last and penultimate addresses in
179 # the list on the same line if possible, hence the extra \z
180 # Fall back to folding on spaces at $lim + 1 chars
181 my $lim = COLS - 8; # 8 = "\t" display width
182 my $too_long = $lim + 1;
183 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
185 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
188 .{$too_long,}?)(?:\s|\z)/xgo));
191 sub _hdr_names_html ($$) {
192 my ($hdr, $field) = @_;
193 my $val = $hdr->header($field) or return '';
194 ascii_html(join(', ', PublicInbox::Address::names($val)));
198 my ($nr, $singular, $plural) = @_;
199 return "0 $plural" if $nr == 0;
200 $nr == 1 ? "$nr $singular" : "$nr $plural";
203 # this is already inside a <pre>
205 my ($smsg, $ctx, $more) = @_;
206 my $srch = $ctx->{srch};
207 my $subj = $smsg->subject;
208 my $mid_raw = $smsg->mid;
209 my $id = id_compress($mid_raw, 1);
212 my $root_anchor = $ctx->{root_anchor} || '';
214 my $obfs_ibx = $ctx->{-obfs_ibx};
216 $subj = '(no subject)' if $subj eq '';
217 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
218 $subj = '<b>'.ascii_html($subj).'</b>';
219 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
220 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
222 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
224 my $ds = $smsg->ds; # for v1 non-Xapian/SQLite users
225 # deleting {mime} is critical to memory use,
226 # the rest of the fields saves about 400K as we iterate across 1K msgs
227 my ($mime) = delete @$smsg{qw(mime ds ts blob subject)};
229 my $hdr = $mime->header_obj;
230 my $from = _hdr_names_html($hdr, 'From');
231 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
232 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
233 my $upfx = $ctx->{-upfx};
234 my $mhref = $upfx . mid_escape($mid_raw) . '/';
235 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
236 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
237 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
238 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
239 my ($tlen, $clen) = (length($to), length($cc));
241 if (($tlen + $clen) > COLS) {
242 $to_cc .= ' To: '.$to."\n" if $tlen;
243 $to_cc .= ' Cc: '.$cc."\n" if $clen;
246 $to_cc .= ' To: '.$to;
247 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
249 $to_cc .= ' Cc: '.$cc if $clen;
253 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
256 my $mapping = $ctx->{mapping};
257 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
258 my $mirt = PublicInbox::Hval->new_msgid($irt);
259 my $href = $upfx . $mirt->{href}. '/';
260 my $html = $mirt->as_html;
261 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
265 # scan through all parts, looking for displayable text
266 my $ibx = $ctx->{-inbox};
267 msg_iter($mime, sub { $rv .= add_text_body($mhref, $ctx, $_[0]) });
270 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
271 "<a\nhref=\"$mhref\">permalink</a>" .
272 " <a\nhref=\"${mhref}raw\">raw</a>" .
273 " <a\nhref=\"${mhref}#R\">reply</a>";
276 if (my $pct = $ctx->{pct}) { # used by SearchView.pm
277 $rv .= "\t[relevance $pct->{$mid_raw}%]";
280 my $nested = 'nested';
285 $flat = "<b>$flat</b>";
287 $nested = "<b>$nested</b>";
289 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
290 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
291 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
296 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
300 sub pad_link ($$;$) {
301 my ($mid, $level, $s) = @_;
303 my $id = id_compress($mid, 1);
304 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
308 my ($mid_raw, $irt, $id, $ctx) = @_;
310 my $mapping = $ctx->{mapping} or return $rv;
312 my $mid_map = $mapping->{$mid_raw};
314 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
315 my ($attr, $node, $idx, $level) = @$mid_map;
316 my $children = $node->{children};
317 my $nr_c = scalar @$children;
320 if (my $smsg = $node->{smsg}) {
321 # delete saves about 200KB on a 1K message thread
322 if (my $refs = delete $smsg->{references}) {
323 ($$irt) = ($refs =~ m/<([^>]+)>\z/);
326 my $irt_map = $mapping->{$$irt} if defined $$irt;
327 if (defined $irt_map) {
328 $siblings = $irt_map->[1]->{children};
329 $nr_s = scalar(@$siblings) - 1;
330 $rv .= $pad . $irt_map->[0];
332 my $prev = $siblings->[$idx - 1];
333 my $pmid = $prev->{id};
335 my $s = ($idx - 1). ' preceding siblings ...';
336 $rv .= pad_link($pmid, $level, $s);
337 } elsif ($idx == 2) {
338 my $ppmid = $siblings->[0]->{id};
339 $rv .= $pad . $mapping->{$ppmid}->[0];
341 $rv .= $pad . $mapping->{$pmid}->[0];
344 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
345 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
346 $attr =~ s!\n\z!</b>\n!s;
347 $attr =~ s!<a\nhref.*</a> !!s; # no point in duplicating subject
348 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
351 my $cmid = $children->[0]->{id};
352 $rv .= $pad . $mapping->{$cmid}->[0];
354 my $s = ($nr_c - 1). ' more replies';
355 $rv .= pad_link($cmid, $level + 1, $s);
356 } elsif (my $cn = $children->[1]) {
357 $rv .= $pad . $mapping->{$cn->{id}}->[0];
361 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
363 my $nmid = $next->{id};
364 $rv .= $pad . $mapping->{$nmid}->[0];
365 my $nnext = $nr_s - $idx;
367 my $s = ($nnext - 1).' subsequent siblings';
368 $rv .= pad_link($nmid, $level, $s);
369 } elsif (my $nn = $siblings->[$idx + 2]) {
370 $rv .= $pad . $mapping->{$nn->{id}}->[0];
373 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
377 my ($rootset, $ctx, $cb) = @_;
378 my @q = map { (0, $_, -1) } @$rootset;
380 my ($level, $node, $i) = splice(@q, 0, 3);
381 defined $node or next;
382 $cb->($ctx, $level, $node, $i) or return;
385 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
390 my ($ctx, $level, $node, $idx) = @_;
391 $ctx->{mapping}->{$node->{id}} = [ '', $node, $idx, $level ];
392 skel_dump($ctx, $level, $node);
395 sub thread_index_entry {
396 my ($ctx, $level, $smsg) = @_;
397 my ($beg, $end) = thread_adj_level($ctx, $level);
398 $beg . '<pre>' . index_entry($smsg, $ctx, 0) . '</pre>' . $end;
401 sub stream_thread ($$) {
402 my ($rootset, $ctx) = @_;
403 my $ibx = $ctx->{-inbox};
404 my @q = map { (0, $_) } @$rootset;
409 my $node = shift @q or next;
411 unshift @q, map { ($cl, $_) } @{$node->{children}};
412 $smsg = $ibx->smsg_mime($node->{smsg}) and last;
414 return missing_thread($ctx) unless $smsg;
416 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
417 $ctx->{-title_html} = ascii_html($smsg->subject);
418 $ctx->{-html_tip} = thread_index_entry($ctx, $level, $smsg);
420 PublicInbox::WwwStream->response($ctx, 200, sub {
424 my $node = shift @q or next;
426 unshift @q, map { ($cl, $_) } @{$node->{children}};
427 if ($smsg = $ibx->smsg_mime($node->{smsg})) {
428 return thread_index_entry($ctx, $level, $smsg);
430 return ghost_index_entry($ctx, $level, $node);
433 my $ret = join('', thread_adj_level($ctx, 0));
434 $ret .= ${$ctx->{dst}}; # skel
442 my $mid = $ctx->{mid};
443 my $srch = $ctx->{srch};
444 my ($nr, $msgs) = $srch->get_thread($mid);
445 return missing_thread($ctx) if $nr == 0;
446 my $skel = '<hr><pre>';
447 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
448 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
449 $skel .= "<b\nid=t>Thread overview:</b> ";
450 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
451 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
452 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
453 $skel .= "-- links below jump to the message on this page --\n";
454 $ctx->{-upfx} = '../../';
455 $ctx->{cur_level} = 0;
456 $ctx->{dst} = \$skel;
457 $ctx->{prev_attr} = '';
458 $ctx->{prev_level} = 0;
459 $ctx->{root_anchor} = anchor_for($mid);
460 $ctx->{mapping} = {};
461 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
464 my $rootset = thread_results($ctx, $msgs);
466 # reduce hash lookups in pre_thread->skel_dump
467 my $ibx = $ctx->{-inbox};
468 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
469 walk_thread($rootset, $ctx, *pre_thread);
472 return stream_thread($rootset, $ctx) unless $ctx->{flat};
474 # flat display: lazy load the full message from smsg
476 while (my $m = shift @$msgs) {
477 $smsg = $ibx->smsg_mime($m) and last;
479 return missing_thread($ctx) unless $smsg;
480 $ctx->{-title_html} = ascii_html($smsg->subject);
481 $ctx->{-html_tip} = '<pre>'.index_entry($smsg, $ctx, scalar @$msgs);
483 PublicInbox::WwwStream->response($ctx, 200, sub {
486 while (my $m = shift @$msgs) {
487 $smsg = $ibx->smsg_mime($m) and last;
489 return index_entry($smsg, $ctx, scalar @$msgs) if $smsg;
495 sub multipart_text_as_html {
496 my ($mime, $upfx, $ctx) = @_;
499 # scan through all parts, looking for displayable text
500 msg_iter($mime, sub { $rv .= add_text_body($upfx, $ctx, $_[0]) });
505 my ($s, $l, $quot) = @_;
507 # show everything in the full version with anchor from
508 # short version (see above)
509 my $rv = $l->linkify_1($$quot);
511 # we use a <span> here to allow users to specify their own
512 # color for quoted text
513 $rv = $l->linkify_2(ascii_html($rv));
515 $$s .= qq(<span\nclass="q">) . $rv . '</span>'
518 sub attach_link ($$$$;$) {
519 my ($upfx, $ct, $p, $fn, $err) = @_;
520 my ($part, $depth, @idx) = @$p;
521 my $nl = $idx[-1] > 1 ? "\n" : '';
522 my $idx = join('.', @idx);
523 my $size = bytes::length($part->body);
525 # hide attributes normally, unless we want to aid users in
526 # spotting MUA problems:
527 $ct =~ s/;.*// unless $err;
528 $ct = ascii_html($ct);
529 my $desc = $part->header('Content-Description');
530 $desc = $fn unless defined $desc;
531 $desc = '' unless defined $desc;
533 if (defined $fn && $fn =~ /\A[[:alnum:]][\w\.-]+[[:alnum:]]\z/) {
535 } elsif ($ct eq 'text/plain') {
540 my $ret = qq($nl<a\nhref="$upfx$idx-$sfn">);
543 "[-- Warning: decoded text below may be mangled --]\n";
545 $ret .= "[-- Attachment #$idx: ";
546 my $ts = "Type: $ct, Size: $size bytes";
547 $desc = ascii_html($desc);
548 $ret .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
553 my ($upfx, $ctx, $p) = @_;
554 my $ibx = $ctx->{-inbox};
555 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
556 # $p - from msg_iter: [ Email::MIME, depth, @idx ]
557 my ($part, $depth, @idx) = @$p;
558 my $ct = $part->content_type || 'text/plain';
559 my $fn = $part->filename;
560 my ($s, $err) = msg_part_text($part, $ct);
562 return attach_link($upfx, $ct, $p, $fn) unless defined $s;
564 # makes no difference to browsers, and don't screw up filename
565 # link generation in diffs with the extra '%0D'
568 # always support diff-highlighting, but we can't linkify hunk
569 # headers for solver unless some coderepo are configured:
571 if ($s =~ /^(?:diff|---|\+{3}) /ms) {
572 # diffstat anchors do not link across attachments or messages:
573 $idx[0] = $upfx . $idx[0] if $upfx ne '';
574 $ctx->{-apfx} = join('/', @idx);
575 $ctx->{-anchors} = {}; # attr => filename
576 $ctx->{-diff} = $diff = [];
577 delete $ctx->{-long_path};
579 if ($ibx->{-repo_objs}) {
580 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
582 $spfx =~ s!/([^/]*)/\z!/!;
584 my $n_slash = $upfx =~ tr!/!/!;
587 } elsif ($n_slash == 1) {
589 } else { # nslash == 2
594 $ctx->{-spfx} = $spfx;
597 # some editors don't put trailing newlines at the end:
598 $s .= "\n" unless $s =~ /\n\z/s;
600 # split off quoted and unquoted blocks:
601 my @sections = split(/((?:^>[^\n]*\n)+)/sm, $s);
603 if (defined($fn) || $depth > 0 || $err) {
604 # badly-encoded message with $err? tell the world about it!
605 $s .= attach_link($upfx, $ct, $p, $fn, $err);
608 my $l = PublicInbox::Linkify->new;
609 foreach my $cur (@sections) {
611 flush_quote(\$s, $l, \$cur);
613 @$diff = split(/^/m, $cur);
615 flush_diff(\$s, $ctx, $l);
619 $s .= $l->linkify_2(ascii_html($cur));
624 obfuscate_addrs($obfs_ibx, $s) if $obfs_ibx;
628 sub _msg_html_prepare {
629 my ($hdr, $ctx, $more, $nr) = @_;
630 my $srch = $ctx->{srch} if $ctx;
632 my $obfs_ibx = $ctx->{-obfs_ibx};
634 my $mids = mids($hdr);
635 my $multiple = scalar(@$mids) > 1; # zero, one, infinity
639 "<pre>WARNING: multiple messages refer to this Message-ID\n</pre>";
641 $rv .= "<pre\nid=b>"; # anchor for body start
646 $ctx->{-upfx} = '../';
650 if (defined($v = $hdr->header('From'))) {
651 $v = PublicInbox::Hval->new($v);
652 my @n = PublicInbox::Address::names($v->raw);
653 $title[1] = ascii_html(join(', ', @n));
656 obfuscate_addrs($obfs_ibx, $v);
657 obfuscate_addrs($obfs_ibx, $title[1]);
659 $rv .= "From: $v\n" if $v ne '';
661 foreach my $h (qw(To Cc)) {
662 defined($v = $hdr->header($h)) or next;
665 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
666 $rv .= "$h: $v\n" if $v ne '';
668 if (defined($v = $hdr->header('Subject')) && ($v ne '')) {
670 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
672 $rv .= qq(Subject: <a\nhref="#r"\nid=t>$v</a>\n);
674 $rv .= "Subject: $v\n";
677 } else { # dummy anchor for thread skeleton at bottom of page
678 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $srch;
679 $title[0] = '(no subject)';
681 if (defined($v = $hdr->header('Date'))) {
683 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
686 $ctx->{-title_html} = join(' - ', @title);
688 my $mid = PublicInbox::Hval->new_msgid($_) ;
689 my $mhtml = $mid->as_html;
691 my $href = $mid->{href};
692 $rv .= "Message-ID: ";
693 $rv .= "<a\nhref=\"../$href/\">";
694 $rv .= "<$mhtml></a> ";
695 $rv .= "(<a\nhref=\"../$href/raw\">raw</a>)\n";
697 $rv .= "Message-ID: <$mhtml> ";
698 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
701 $rv .= _parent_headers($hdr, $srch);
706 my ($dst, $ctx, $hdr, $tpfx) = @_;
707 my $srch = $ctx->{srch};
708 my $mid = mids($hdr)->[0];
709 my ($nr, $msgs) = $srch->get_thread($mid);
710 my $expand = qq(expand[<a\nhref="${tpfx}T/#u">flat</a>) .
711 qq(|<a\nhref="${tpfx}t/#u">nested</a>] ) .
712 qq(<a\nhref="${tpfx}t.mbox.gz">mbox.gz</a> ) .
713 qq(<a\nhref="${tpfx}t.atom">Atom feed</a>);
715 my $parent = in_reply_to($hdr);
716 $$dst .= "\n<b>Thread overview: </b>";
718 if (defined $parent) {
719 $$dst .= "$expand\n ";
720 $$dst .= ghost_parent("$tpfx../", $parent) . "\n";
722 $$dst .= "[no followups] $expand\n";
724 $ctx->{next_msg} = undef;
725 $ctx->{parent_msg} = $parent;
729 $$dst .= "$nr+ messages / $expand";
730 $$dst .= qq! <a\nhref="#b">top</a>\n!;
732 my $subj = $hdr->header('Subject');
733 defined $subj or $subj = '';
734 $subj = '(no subject)' if $subj eq '';
735 $ctx->{prev_subj} = [ split(/ /, $srch->subject_normalized($subj)) ];
737 $ctx->{prev_attr} = '';
738 $ctx->{prev_level} = 0;
741 # reduce hash lookups in skel_dump
742 my $ibx = $ctx->{-inbox};
743 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
744 walk_thread(thread_results($ctx, $msgs), $ctx, *skel_dump);
746 $ctx->{parent_msg} = $parent;
749 sub _parent_headers {
750 my ($hdr, $srch) = @_;
753 my $refs = references($hdr);
754 my $irt = pop @$refs;
756 my $v = PublicInbox::Hval->new_msgid($irt);
757 my $html = $v->as_html;
758 my $href = $v->{href};
759 $rv .= "In-Reply-To: <";
760 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
763 # do not display References: if search is present,
764 # we show the thread skeleton at the bottom, instead.
768 @$refs = map { linkify_ref_nosrch($_) } @$refs;
769 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
775 my ($hdr, $standalone, $ctx, $rhref) = @_;
777 my $srch = $ctx->{srch} if $ctx;
780 my $idx = $standalone ? " <a\nhref=\"$upfx\">index</a>" : '';
784 thread_skel(\$idx, $ctx, $hdr, $tpfx);
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 $irt = "$next $prev$parent ";
818 $irt .= qq(<a\nhref="$rhref">reply</a>);
822 sub linkify_ref_nosrch {
823 my $v = PublicInbox::Hval->new_msgid($_[0]);
824 my $html = $v->as_html;
825 my $href = $v->{href};
826 "<<a\nhref=\"../$href/\">$html</a>>";
831 'm' . id_compress($msgid, 1);
835 my ($upfx, $mid) = @_;
837 $mid = PublicInbox::Hval->new_msgid($mid);
838 my $href = $mid->{href};
839 my $html = $mid->as_html;
840 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
845 $level ? INDENT x ($level - 1) : '';
849 my ($ctx, $level, $node, $idx) = @_;
850 ++$ctx->{root_idx} if $level == 0;
851 if ($node->{id} eq $ctx->{mid}) {
852 $ctx->{found_mid_at} = $ctx->{root_idx};
858 sub strict_loose_note ($) {
861 " -- strict thread matches above, loose matches on Subject: below --\n";
863 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
865 " -- use mbox.gz link to download all $nr messages --\n";
871 my ($ctx, $msgs) = @_;
872 require PublicInbox::SearchThread;
873 my $ibx = $ctx->{-inbox};
874 my $rootset = PublicInbox::SearchThread::thread($msgs, *sort_ds, $ibx);
876 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
877 # and preserved in the future. This bug is hidden by `sid' matches
878 # in get_thread, so we never noticed it until now. And even when
879 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
880 # bump since reindexing is expensive and users may not do it
882 # loose threading could've returned too many results,
883 # put the root the message we care about at the top:
884 my $mid = $ctx->{mid};
885 if (defined($mid) && scalar(@$rootset) > 1) {
886 $ctx->{root_idx} = -1;
887 my $nr = scalar @$msgs;
888 walk_thread($rootset, $ctx, *find_mid_root);
889 my $idx = $ctx->{found_mid_at};
890 if (defined($idx) && $idx != 0) {
891 my $tip = splice(@$rootset, $idx, 1);
892 @$rootset = reverse @$rootset;
893 unshift @$rootset, $tip;
894 $ctx->{sl_note} = strict_loose_note($nr);
902 require PublicInbox::ExtMsg;
903 PublicInbox::ExtMsg::ext_msg($ctx);
906 sub fmt_ts { POSIX::strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
909 my ($prev_subj, $subj, $val) = @_;
911 my $omit = ''; # '"' denotes identical text omitted
912 my (@prev_pop, @curr_pop);
913 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
914 push(@prev_pop, pop(@$prev_subj));
915 push(@curr_pop, pop(@$subj));
918 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
919 if (scalar(@curr_pop) == 1) {
921 push @$prev_subj, @prev_pop;
922 push @$subj, @curr_pop;
928 my ($ctx, $level, $node) = @_;
929 my $smsg = $node->{smsg} or return _skel_ghost($ctx, $level, $node);
931 my $dst = $ctx->{dst};
932 my $cur = $ctx->{cur};
933 my $mid = $smsg->{mid};
935 if ($level == 0 && $ctx->{skel_dump_roots}++) {
936 $$dst .= delete $ctx->{sl_note} || '';
939 my $f = ascii_html($smsg->from_name);
940 my $obfs_ibx = $ctx->{-obfs_ibx};
941 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
943 my $d = fmt_ts($smsg->{ds}) . ' ' . indent_for($level) . th_pfx($level);
945 $ctx->{first_level} ||= $level;
947 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
948 $ctx->{prev_attr} = $attr;
950 $ctx->{prev_level} = $level;
955 $$dst .= "<b>$d<a\nid=r\nhref=\"#t\">".
956 "$attr [this message]</a></b>\n";
959 $ctx->{prev_msg} = $mid;
962 $ctx->{next_msg} ||= $mid;
965 # Subject is never undef, this mail was loaded from
966 # our Xapian which would've resulted in '' if it were
967 # really missing (and Filter rejects empty subjects)
968 my @subj = split(/ /, $ctx->{srch}->subject_normalized($smsg->subject));
970 # remove common suffixes from the subject if it matches the previous,
971 # so we do not show redundant text at the end.
972 my $prev_subj = $ctx->{prev_subj} || [];
973 $ctx->{prev_subj} = [ @subj ];
974 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
977 my $subj = join(' ', @subj);
978 $subj = ascii_html($subj);
979 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
980 $end = "$subj</a> $omit$f\n"
986 my $mapping = $ctx->{mapping};
988 my $map = $mapping->{$mid};
989 $id = id_compress($mid, 1);
991 $map->[0] = "$d<a\nhref=\"$m\">$end";
994 $m = $ctx->{-upfx}.mid_escape($mid).'/';
996 $$dst .= $d . "<a\nhref=\"$m\"$id>" . $end;
1001 my ($ctx, $level, $node) = @_;
1003 my $mid = $node->{id};
1004 my $d = $ctx->{pct} ? ' [irrelevant] ' # search result
1006 $d .= indent_for($level) . th_pfx($level);
1007 my $upfx = $ctx->{-upfx};
1008 my $m = PublicInbox::Hval->new_msgid($mid);
1009 my $href = $upfx . $m->{href} . '/';
1010 my $html = $m->as_html;
1012 my $mapping = $ctx->{mapping};
1013 my $map = $mapping->{$mid} if $mapping;
1015 my $id = id_compress($mid, 1);
1016 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1017 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1019 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1021 my $dst = $ctx->{dst};
1028 (eval { $a->topmost->{smsg}->ds } || 0) <=>
1029 (eval { $b->topmost->{smsg}->ds } || 0)
1033 # accumulate recent topics if search is supported
1034 # returns 200 if done, 404 if not
1036 my ($ctx, $level, $node) = @_;
1037 my $srch = $ctx->{srch};
1038 my $mid = $node->{id};
1039 my $x = $node->{smsg} || $ctx->{-inbox}->smsg_by_mid($mid);
1043 $subj = $x->subject;
1044 $subj = $srch->subject_normalized($subj);
1045 $subj = '(no subject)' if $subj eq '';
1048 $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1049 $ctx->{-cur_topic} = $topic;
1050 push @{$ctx->{order}}, $topic;
1054 $topic = $ctx->{-cur_topic}; # should never be undef
1055 $topic->[0] = $ds if $ds > $topic->[0];
1057 my $seen = $topic->[2];
1058 if (scalar(@$topic) == 3) { # parent was a ghost
1059 push @$topic, $subj;
1060 } elsif (!$seen->{$subj}) {
1061 push @$topic, $level, $subj;
1063 $seen->{$subj} = $mid; # latest for subject
1064 } else { # ghost message
1065 return 1 if $level != 0; # ignore child ghosts
1066 $topic = [ -666, 0, {} ];
1067 $ctx->{-cur_topic} = $topic;
1068 push @{$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;
1084 my $srch = $ctx->{srch};
1086 # sort by recency, this allows new posts to "bump" old topics...
1087 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1088 my ($ds, $n, $seen, $top, @ex) = @$topic;
1090 next unless defined $top; # ghost topic
1091 my $mid = delete $seen->{$top};
1092 my $href = mid_escape($mid);
1093 my $prev_subj = [ split(/ /, $top) ];
1094 $top = PublicInbox::Hval->new($top)->as_html;
1097 # $n isn't the total number of posts on the topic,
1098 # just the number of posts in the current results window
1102 $anchor = '#u'; # top of only message
1104 $n = " ($n+ messages)";
1105 $anchor = '#t'; # thread skeleton
1108 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1109 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1110 my $s = "<a\nhref=\"$href/T/$anchor\">$top</a>\n" .
1111 " $ds UTC $n - $mbox / $atom\n";
1112 for (my $i = 0; $i < scalar(@ex); $i += 2) {
1113 my $level = $ex[$i];
1114 my $subj = $ex[$i + 1];
1115 $mid = delete $seen->{$subj};
1116 my @subj = split(/ /, $srch->subject_normalized($subj));
1117 my @next_prev = @subj; # full copy
1118 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1119 $prev_subj = \@next_prev;
1120 $subj = ascii_html(join(' ', @subj));
1121 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1122 $href = mid_escape($mid);
1123 $s .= indent_for($level) . TCHILD;
1124 $s .= "<a\nhref=\"$href/T/#u\">$subj</a>$omit\n";
1128 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1134 POSIX::strftime('%Y%m%d%H%M%S', gmtime($ts));
1138 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1139 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1142 sub pagination_footer ($$) {
1143 my ($ctx, $latest) = @_;
1144 delete $ctx->{qp} or return;
1145 my $next = $ctx->{next_page} || '';
1146 my $prev = $ctx->{prev_page} || '';
1148 $next = $next ? "$next " : ' ';
1149 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1151 "<hr><pre>page: $next$prev</pre>";
1154 sub index_nav { # callback for WwwStream
1155 my (undef, $ctx) = @_;
1156 pagination_footer($ctx, '.')
1159 sub paginate_recent ($$) {
1160 my ($ctx, $lim) = @_;
1161 my $t = $ctx->{qp}->{t} || '';
1162 my $opts = { limit => $lim };
1163 my ($after, $before);
1165 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1166 # if only $after exists "YYYYMMDD.." because "." could be skipped
1167 # if interpreted as an end-of-sentence
1168 $t =~ s/\A(\d{8,14})-// and $after = str2ts($1);
1169 $t =~ /\A(\d{8,14})\z/ and $before = str2ts($1);
1171 my $ibx = $ctx->{-inbox};
1172 my $msgs = $ibx->recent($opts, $after, $before);
1173 my $nr = scalar @$msgs;
1174 if ($nr < $lim && defined($after)) {
1175 $after = $before = undef;
1176 $msgs = $ibx->recent($opts);
1177 $nr = scalar @$msgs;
1179 my $more = $nr == $lim;
1180 my ($newest, $oldest);
1182 $newest = $msgs->[0]->{ts};
1183 $oldest = $msgs->[-1]->{ts};
1184 # if we only had $after, our SQL query in ->recent ordered
1185 if ($newest < $oldest) {
1186 ($oldest, $newest) = ($newest, $oldest);
1187 $more = 0 if defined($after) && $after < $oldest;
1190 if (defined($oldest) && $more) {
1191 my $s = ts2str($oldest);
1192 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1194 if (defined($newest) && (defined($before) || defined($after))) {
1195 my $s = ts2str($newest);
1196 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1203 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1205 walk_thread(thread_results($ctx, $msgs), $ctx, *acc_topic);
1207 PublicInbox::WwwStream->response($ctx, dump_topics($ctx), *index_nav);
1210 sub thread_adj_level {
1211 my ($ctx, $level) = @_;
1213 my $max = $ctx->{cur_level};
1215 return ('', '') if $max == 0; # flat output
1217 # reset existing lists
1218 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1219 $ctx->{cur_level} = 0;
1221 } elsif ($level == $max) { # continue existing list
1223 } elsif ($level < $max) {
1224 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1225 $ctx->{cur_level} = $level;
1226 ("$beg<li>", '</li>');
1227 } else { # ($level > $max) # start a new level
1228 $ctx->{cur_level} = $level;
1229 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1234 sub ghost_index_entry {
1235 my ($ctx, $level, $node) = @_;
1236 my ($beg, $end) = thread_adj_level($ctx, $level);
1237 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{id})