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 mid_href);
12 use PublicInbox::Linkify;
13 use PublicInbox::MID qw/id_compress 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::Smsg qw(subject_normalized);
22 use constant COLS => 72;
23 use constant INDENT => ' ';
24 use constant TCHILD => '` ';
25 sub th_pfx ($) { $_[0] == 0 ? '' : TCHILD };
29 if (my $more = delete $ctx->{more}) { # unlikely
30 # fake an EOF if $more retrieval fails;
31 eval { msg_page_more($ctx, $nr, @$more) };
32 } elsif (my $hdr = delete $ctx->{hdr}) {
33 # fake an EOF if generating the footer fails;
34 # we want to at least show the message if something
36 eval { html_footer($ctx, $hdr) };
42 # public functions: (unstable)
46 my $mid = $ctx->{mid};
47 my $ibx = $ctx->{-inbox};
48 my ($smsg, $first, $next);
49 if (my $over = $ibx->over) {
51 $smsg = $over->next_by_mid($mid, \$id, \$prev) or return;
52 $first = $ibx->msg_by_smsg($smsg) or return;
53 $next = $over->next_by_mid($mid, \$id, \$prev);
54 $ctx->{more} = [ $id, $prev, $next ] if $next;
56 $first = $ibx->msg_by_mid($mid) or return;
58 my $mime = PublicInbox::MIME->new($first);
59 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
60 my $hdr = $ctx->{hdr} = $mime->header_obj;
61 $ctx->{obuf} = _msg_page_prepare_obuf($hdr, $ctx, 0);
63 # $next cannot be true w/o $smsg being defined:
64 $ctx->{mhref} = $next ? '../'.mid_href($smsg->{mid}).'/' : '';
65 multipart_text_as_html($mime, $ctx);
66 $ctx->{-html_tip} = (${delete $ctx->{obuf}} .= '</pre><hr>');
67 PublicInbox::WwwStream->response($ctx, 200, \&msg_page_i);
70 sub msg_page_more { # cold
71 my ($ctx, $nr, $id, $prev, $smsg) = @_;
72 my $ibx = $ctx->{-inbox};
73 my $next = $ibx->over->next_by_mid($ctx->{mid}, \$id, \$prev);
74 $ctx->{more} = [ $id, $prev, $next ] if $next;
75 $smsg = $ibx->smsg_mime($smsg) or return '';
76 $ctx->{mhref} = '../' . mid_href($smsg->{mid}) . '/';
77 my $mime = delete $smsg->{mime};
78 $ctx->{obuf} = _msg_page_prepare_obuf($mime->header_obj, $ctx, $nr);
79 multipart_text_as_html($mime, $ctx);
80 ${delete $ctx->{obuf}} .= '</pre><hr>';
83 # /$INBOX/$MESSAGE_ID/#R
87 'https://kernel.org/pub/software/scm/git/docs/git-send-email.html';
89 'https://en.wikipedia.org/wiki/Posting_style#Interleaved_style';
92 my $ibx = $ctx->{-inbox};
93 if (my $url = $ibx->{infourl}) {
94 $url = prurl($ctx->{env}, $url);
95 $info = qq(\n List information: <a\nhref="$url">$url</a>\n);
98 my ($arg, $link, $reply_to_all) =
99 PublicInbox::Reply::mailto_arg_link($ibx, $hdr);
100 if (ref($arg) eq 'SCALAR') {
101 return '<pre id=R>'.ascii_html($$arg).'</pre>';
104 # mailto: link only works if address obfuscation is disabled
108 * If your mail client supports setting the <b>In-Reply-To</b> header
109 via mailto: links, try the <a
110 href="$link">mailto: link</a>
114 push @$arg, '/path/to/YOUR_REPLY';
115 $arg = ascii_html(join(" \\\n ", '', @$arg));
118 id=R><b>Reply instructions:</b>
120 You may reply publicly to <a
121 href=#t>this message</a> via plain-text email
122 using any one of the following methods:
124 * Save the following mbox file, import it into your mail client,
125 and $reply_to_all from there: <a
128 Avoid top-posting and favor interleaved quoting:
130 href="$p_url">$p_url</a>
132 * Reply using the <b>--to</b>, <b>--cc</b>, and <b>--in-reply-to</b>
133 switches of git-send-email(1):
138 href="$se_url">$se_url</a>
145 my $refs = references($hdr);
149 sub fold_addresses ($) {
150 return $_[0] if length($_[0]) <= COLS;
151 # try to fold on commas after non-word chars before $lim chars,
152 # Try to get the "," preceeded by ">" or ")", but avoid folding
153 # on the comma where somebody uses "Lastname, Firstname".
154 # We also try to keep the last and penultimate addresses in
155 # the list on the same line if possible, hence the extra \z
156 # Fall back to folding on spaces at $lim + 1 chars
157 my $lim = COLS - 8; # 8 = "\t" display width
158 my $too_long = $lim + 1;
159 $_[0] =~ s/\s*\z//s; # Email::Simple doesn't strip trailing spaces
161 ($_[0] =~ /(.{0,$lim}\W(?:,|\z)|
164 .{$too_long,}?)(?:\s|\z)/xgo));
167 sub _hdr_names_html ($$) {
168 my ($hdr, $field) = @_;
169 my @vals = $hdr->header($field) or return '';
170 ascii_html(join(', ', PublicInbox::Address::names(join(',', @vals))));
174 my ($nr, $singular, $plural) = @_;
175 return "0 $plural" if $nr == 0;
176 $nr == 1 ? "$nr $singular" : "$nr $plural";
179 # human-friendly format
180 sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
182 # this is already inside a <pre>
184 my ($smsg, $ctx, $more) = @_;
185 my $subj = $smsg->subject;
186 my $mid_raw = $smsg->mid;
187 my $id = id_compress($mid_raw, 1);
190 my $root_anchor = $ctx->{root_anchor} || '';
192 my $obfs_ibx = $ctx->{-obfs_ibx};
194 $subj = '(no subject)' if $subj eq '';
195 my $rv = "<a\nhref=#e$id\nid=m$id>*</a> ";
196 $subj = '<b>'.ascii_html($subj).'</b>';
197 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
198 $subj = "<u\nid=u>$subj</u>" if $root_anchor eq $id_m;
200 $rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
202 my $ds = $smsg->ds; # for v1 non-Xapian/SQLite users
203 # deleting {mime} is critical to memory use,
204 # the rest of the fields saves about 400K as we iterate across 1K msgs
205 my ($mime) = delete @$smsg{qw(mime ds ts blob subject)};
207 my $hdr = $mime->header_obj;
208 my $from = _hdr_names_html($hdr, 'From');
209 obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
210 $rv .= "From: $from @ ".fmt_ts($ds)." UTC";
211 my $upfx = $ctx->{-upfx};
212 my $mhref = $upfx . mid_href($mid_raw) . '/';
213 $rv .= qq{ (<a\nhref="$mhref">permalink</a> / };
214 $rv .= qq{<a\nhref="${mhref}raw">raw</a>)\n};
215 my $to = fold_addresses(_hdr_names_html($hdr, 'To'));
216 my $cc = fold_addresses(_hdr_names_html($hdr, 'Cc'));
217 my ($tlen, $clen) = (length($to), length($cc));
219 if (($tlen + $clen) > COLS) {
220 $to_cc .= ' To: '.$to."\n" if $tlen;
221 $to_cc .= ' Cc: '.$cc."\n" if $clen;
224 $to_cc .= ' To: '.$to;
225 $to_cc .= '; <b>+Cc:</b> '.$cc if $clen;
227 $to_cc .= ' Cc: '.$cc if $clen;
231 obfuscate_addrs($obfs_ibx, $to_cc) if $obfs_ibx;
234 my $mapping = $ctx->{mapping};
235 if (!$mapping && (defined($irt) || defined($irt = in_reply_to($hdr)))) {
236 my $href = $upfx . mid_href($irt) . '/';
237 my $html = ascii_html($irt);
238 $rv .= qq(In-Reply-To: <<a\nhref="$href">$html</a>>\n)
242 # scan through all parts, looking for displayable text
243 $ctx->{mhref} = $mhref;
245 msg_iter($mime, \&add_text_body, $ctx, 1);
249 $rv .= "\n<a\nhref=#$id_m\nid=e$id>^</a> ".
250 "<a\nhref=\"$mhref\">permalink</a>" .
251 " <a\nhref=\"${mhref}raw\">raw</a>" .
252 " <a\nhref=\"${mhref}#R\">reply</a>";
255 if (defined(my $pct = $smsg->{pct})) { # used by SearchView.pm
256 $rv .= "\t[relevance $pct%]";
259 my $nested = 'nested';
264 $flat = "<b>$flat</b>";
266 $nested = "<b>$nested</b>";
268 $rv .= "\t[<a\nhref=\"${mhref}T/#u\">$flat</a>";
269 $rv .= "|<a\nhref=\"${mhref}t/#u\">$nested</a>]";
270 $rv .= " <a\nhref=#r$id>$ctx->{s_nr}</a>";
275 $rv .= $more ? '</pre><hr><pre>' : '</pre>' if $hr;
279 sub pad_link ($$;$) {
280 my ($mid, $level, $s) = @_;
282 my $id = id_compress($mid, 1);
283 (' 'x19).indent_for($level).th_pfx($level)."<a\nhref=#r$id>($s)</a>\n";
287 my ($mid_raw, $irt, $id, $ctx) = @_;
289 my $mapping = $ctx->{mapping} or return $rv;
291 my $mid_map = $mapping->{$mid_raw};
293 return 'public-inbox BUG: '.ascii_html($mid_raw).' not mapped';
294 my ($attr, $node, $idx, $level) = @$mid_map;
295 my $children = $node->{children};
296 my $nr_c = scalar @$children;
299 if (my $smsg = $node->{smsg}) {
300 # delete saves about 200KB on a 1K message thread
301 if (my $refs = delete $smsg->{references}) {
302 ($$irt) = ($refs =~ m/<([^>]+)>\z/);
305 my $irt_map = $mapping->{$$irt} if defined $$irt;
306 if (defined $irt_map) {
307 $siblings = $irt_map->[1]->{children};
308 $nr_s = scalar(@$siblings) - 1;
309 $rv .= $pad . $irt_map->[0];
311 my $prev = $siblings->[$idx - 1];
312 my $pmid = $prev->{id};
314 my $s = ($idx - 1). ' preceding siblings ...';
315 $rv .= pad_link($pmid, $level, $s);
316 } elsif ($idx == 2) {
317 my $ppmid = $siblings->[0]->{id};
318 $rv .= $pad . $mapping->{$ppmid}->[0];
320 $rv .= $pad . $mapping->{$pmid}->[0];
323 my $s_s = nr_to_s($nr_s, 'sibling', 'siblings');
324 my $s_c = nr_to_s($nr_c, 'reply', 'replies');
325 $attr =~ s!\n\z!</b>\n!s;
326 $attr =~ s!<a\nhref.*</a> !!s; # no point in duplicating subject
327 $attr =~ s!<a\nhref=[^>]+>([^<]+)</a>!$1!s; # no point linking to self
330 my $cmid = $children->[0]->{id};
331 $rv .= $pad . $mapping->{$cmid}->[0];
333 my $s = ($nr_c - 1). ' more replies';
334 $rv .= pad_link($cmid, $level + 1, $s);
335 } elsif (my $cn = $children->[1]) {
336 $rv .= $pad . $mapping->{$cn->{id}}->[0];
340 my $next = $siblings->[$idx+1] if $siblings && $idx >= 0;
342 my $nmid = $next->{id};
343 $rv .= $pad . $mapping->{$nmid}->[0];
344 my $nnext = $nr_s - $idx;
346 my $s = ($nnext - 1).' subsequent siblings';
347 $rv .= pad_link($nmid, $level, $s);
348 } elsif (my $nn = $siblings->[$idx + 2]) {
349 $rv .= $pad . $mapping->{$nn->{id}}->[0];
352 $rv .= $pad ."<a\nhref=#r$id>$s_s, $s_c; $ctx->{s_nr}</a>\n";
355 # non-recursive thread walker
356 sub walk_thread ($$$) {
357 my ($rootset, $ctx, $cb) = @_;
358 my @q = map { (0, $_, -1) } @$rootset;
360 my ($level, $node, $i) = splice(@q, 0, 3);
361 defined $node or next;
362 $cb->($ctx, $level, $node, $i) or return;
365 unshift @q, map { ($level, $_, $i++) } @{$node->{children}};
369 sub pre_thread { # walk_thread callback
370 my ($ctx, $level, $node, $idx) = @_;
371 $ctx->{mapping}->{$node->{id}} = [ '', $node, $idx, $level ];
372 skel_dump($ctx, $level, $node);
375 sub thread_index_entry {
376 my ($ctx, $level, $smsg) = @_;
377 my ($beg, $end) = thread_adj_level($ctx, $level);
378 $beg . '<pre>' . index_entry($smsg, $ctx, 0) . '</pre>' . $end;
381 sub stream_thread_i { # PublicInbox::WwwStream::getline callback
383 return unless exists($ctx->{skel});
384 my $q = $ctx->{-queue};
386 my $level = shift @$q;
387 my $node = shift @$q or next;
389 unshift @$q, map { ($cl, $_) } @{$node->{children}};
390 if (my $smsg = $ctx->{-inbox}->smsg_mime($node->{smsg})) {
391 return thread_index_entry($ctx, $level, $smsg);
393 return ghost_index_entry($ctx, $level, $node);
396 join('', thread_adj_level($ctx, 0)) . ${delete $ctx->{skel}};
399 sub stream_thread ($$) {
400 my ($rootset, $ctx) = @_;
401 my $ibx = $ctx->{-inbox};
402 my @q = map { (0, $_) } @$rootset;
406 my $node = shift @q or next;
408 unshift @q, map { ($cl, $_) } @{$node->{children}};
409 $smsg = $ibx->smsg_mime($node->{smsg}) and last;
411 return missing_thread($ctx) unless $smsg;
413 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
414 $ctx->{-title_html} = ascii_html($smsg->{subject});
415 $ctx->{-html_tip} = thread_index_entry($ctx, $level, $smsg);
416 $ctx->{-queue} = \@q;
417 PublicInbox::WwwStream->response($ctx, 200, \&stream_thread_i);
420 # /$INBOX/$MESSAGE_ID/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;
427 my $skel = '<hr><pre>';
428 $skel .= $nr == 1 ? 'only message in thread' : 'end of thread';
429 $skel .= ", back to <a\nhref=\"../../\">index</a>\n\n";
430 $skel .= "<b\nid=t>Thread overview:</b> ";
431 $skel .= $nr == 1 ? '(only message)' : "$nr+ messages";
432 $skel .= " (download: <a\nhref=\"../t.mbox.gz\">mbox.gz</a>";
433 $skel .= " / follow: <a\nhref=\"../t.atom\">Atom feed</a>)\n";
434 $skel .= "-- links below jump to the message on this page --\n";
435 $ctx->{-upfx} = '../../';
436 $ctx->{cur_level} = 0;
437 $ctx->{skel} = \$skel;
438 $ctx->{prev_attr} = '';
439 $ctx->{prev_level} = 0;
440 $ctx->{root_anchor} = anchor_for($mid);
441 $ctx->{mapping} = {};
442 $ctx->{s_nr} = ($nr > 1 ? "$nr+ messages" : 'only message')
445 my $rootset = thread_results($ctx, $msgs);
447 # reduce hash lookups in pre_thread->skel_dump
448 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
449 walk_thread($rootset, $ctx, \&pre_thread);
452 return stream_thread($rootset, $ctx) unless $ctx->{flat};
454 # flat display: lazy load the full message from smsg
456 while (my $m = shift @$msgs) {
457 $smsg = $ibx->smsg_mime($m) and last;
459 return missing_thread($ctx) unless $smsg;
460 $ctx->{-title_html} = ascii_html($smsg->{subject});
461 $ctx->{-html_tip} = '<pre>'.index_entry($smsg, $ctx, scalar @$msgs);
462 $ctx->{msgs} = $msgs;
463 PublicInbox::WwwStream->response($ctx, 200, \&thread_html_i);
466 sub thread_html_i { # PublicInbox::WwwStream::getline callback
468 my $msgs = $ctx->{msgs} or return;
469 while (my $smsg = shift @$msgs) {
470 $ctx->{-inbox}->smsg_mime($smsg) or next;
471 return index_entry($smsg, $ctx, scalar @$msgs);
473 my ($skel) = delete @$ctx{qw(skel msgs)};
477 sub multipart_text_as_html {
478 # ($mime, $ctx) = @_; # msg_iter will do "$_[0] = undef"
480 # scan through all parts, looking for displayable text
481 msg_iter($_[0], \&add_text_body, $_[1], 1);
485 my ($s, $l, $quot) = @_;
487 my $rv = $l->to_html($$quot);
489 # we use a <span> here to allow users to specify their own
490 # color for quoted text
492 $$s .= qq(<span\nclass="q">) . $rv . '</span>'
495 sub attach_link ($$$$;$) {
496 my ($ctx, $ct, $p, $fn, $err) = @_;
497 my ($part, $depth, @idx) = @$p;
498 my $nl = $idx[-1] > 1 ? "\n" : '';
499 my $idx = join('.', @idx);
500 my $size = bytes::length($part->body);
502 # hide attributes normally, unless we want to aid users in
503 # spotting MUA problems:
504 $ct =~ s/;.*// unless $err;
505 $ct = ascii_html($ct);
506 my $desc = $part->header('Content-Description');
507 $desc = $fn unless defined $desc;
508 $desc = '' unless defined $desc;
510 if (defined $fn && $fn =~ /\A$PublicInbox::Hval::FN\z/o) {
512 } elsif ($ct eq 'text/plain') {
517 my $rv = $ctx->{obuf};
518 $$rv .= qq($nl<a\nhref="$ctx->{mhref}$idx-$sfn">);
520 $$rv .= "[-- Warning: decoded text below may be mangled --]\n";
522 $$rv .= "[-- Attachment #$idx: ";
523 my $ts = "Type: $ct, Size: $size bytes";
524 $desc = ascii_html($desc);
525 $$rv .= ($desc eq '') ? "$ts --]" : "$desc --]\n[-- $ts --]";
530 sub add_text_body { # callback for msg_iter
532 my $upfx = $ctx->{mhref};
533 my $ibx = $ctx->{-inbox};
534 # $p - from msg_iter: [ Email::MIME, depth, @idx ]
535 my ($part, $depth, @idx) = @$p;
536 my $ct = $part->content_type || 'text/plain';
537 my $fn = $part->filename;
538 my ($s, $err) = msg_part_text($part, $ct);
539 return attach_link($ctx, $ct, $p, $fn) unless defined $s;
541 # makes no difference to browsers, and don't screw up filename
542 # link generation in diffs with the extra '%0D'
545 # always support diff-highlighting, but we can't linkify hunk
546 # headers for solver unless some coderepo are configured:
548 if ($s =~ /^(?:diff|---|\+{3}) /ms) {
549 # diffstat anchors do not link across attachments or messages:
550 $idx[0] = $upfx . $idx[0] if $upfx ne '';
551 $ctx->{-apfx} = join('/', @idx);
552 $ctx->{-anchors} = {}; # attr => filename
554 delete $ctx->{-long_path};
556 if ($ibx->{-repo_objs}) {
557 if (index($upfx, '//') >= 0) { # absolute URL (Atom feeds)
559 $spfx =~ s!/([^/]*)/\z!/!;
561 my $n_slash = $upfx =~ tr!/!/!;
564 } elsif ($n_slash == 1) {
566 } else { # nslash == 2
571 $ctx->{-spfx} = $spfx;
574 # some editors don't put trailing newlines at the end:
575 $s .= "\n" unless $s =~ /\n\z/s;
577 # split off quoted and unquoted blocks:
578 my @sections = split(/((?:^>[^\n]*\n)+)/sm, $s);
580 my $rv = $ctx->{obuf};
581 if (defined($fn) || $depth > 0 || $err) {
582 # badly-encoded message with $err? tell the world about it!
583 attach_link($ctx, $ct, $p, $fn, $err);
586 my $l = $ctx->{-linkify} //= PublicInbox::Linkify->new;
587 foreach my $cur (@sections) {
589 flush_quote($rv, $l, \$cur);
591 flush_diff($rv, $ctx, \$cur);
594 $$rv .= $l->to_html($cur);
599 obfuscate_addrs($ibx, $$rv) if $ibx->{obfuscate};
602 sub _msg_page_prepare_obuf {
603 my ($hdr, $ctx, $nr) = @_;
604 my $over = $ctx->{-inbox}->over;
605 my $obfs_ibx = $ctx->{-obfs_ibx};
607 my $mids = mids_for_index($hdr);
611 "<pre>WARNING: multiple messages have this Message-ID\n</pre>";
613 $rv .= "<pre\nid=b>"; # anchor for body start
618 $ctx->{-upfx} = '../';
620 my @title; # (Subject[0], From[0])
621 for my $v ($hdr->header('From')) {
622 my @n = PublicInbox::Address::names($v);
624 $title[1] //= ascii_html(join(', ', @n));
626 obfuscate_addrs($obfs_ibx, $v);
627 obfuscate_addrs($obfs_ibx, $title[1]);
629 $rv .= "From: $v\n" if $v ne '';
631 foreach my $h (qw(To Cc)) {
632 for my $v ($hdr->header($h)) {
635 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
636 $rv .= "$h: $v\n" if $v ne '';
639 my @subj = $hdr->header('Subject');
641 my $v = ascii_html(shift @subj);
642 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
644 $rv .= $over ? qq(<a\nhref="#r"\nid=t>$v</a>\n) : "$v\n";
646 for $v (@subj) { # multi-Subject message :<
648 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx;
649 $rv .= "Subject: $v\n";
651 } else { # dummy anchor for thread skeleton at bottom of page
652 $rv .= qq(<a\nhref="#r"\nid=t></a>) if $over;
653 $title[0] = '(no subject)';
655 for my $v ($hdr->header('Date')) {
657 obfuscate_addrs($obfs_ibx, $v) if $obfs_ibx; # possible :P
660 $ctx->{-title_html} = join(' - ', @title);
661 if (scalar(@$mids) == 1) { # common case
662 my $mhtml = ascii_html($mids->[0]);
663 $rv .= "Message-ID: <$mhtml> ";
664 $rv .= "(<a\nhref=\"raw\">raw</a>)\n";
666 # X-Alt-Message-ID can happen if a message is injected from
667 # public-inbox-nntpd because of multiple Message-ID headers.
668 my $lnk = PublicInbox::Linkify->new;
670 for my $h (qw(Message-ID X-Alt-Message-ID)) {
671 $s .= "$h: $_\n" for ($hdr->header_raw($h));
673 $lnk->linkify_mids('..', \$s, 1);
676 $rv .= _parent_headers($hdr, $over);
682 qq(expand[<a\nhref="T/#u">flat</a>) .
683 qq(|<a\nhref="t/#u">nested</a>] ) .
684 qq(<a\nhref="t.mbox.gz">mbox.gz</a> ) .
685 qq(<a\nhref="t.atom">Atom feed</a>);
688 sub thread_skel ($$$) {
689 my ($skel, $ctx, $hdr) = @_;
690 my $mid = mids($hdr)->[0];
691 my $ibx = $ctx->{-inbox};
692 my ($nr, $msgs) = $ibx->over->get_thread($mid);
693 my $parent = in_reply_to($hdr);
694 $$skel .= "\n<b>Thread overview: </b>";
696 if (defined $parent) {
697 $$skel .= SKEL_EXPAND."\n ";
698 $$skel .= ghost_parent('../', $parent) . "\n";
700 $$skel .= '[no followups] '.SKEL_EXPAND."\n";
702 $ctx->{next_msg} = undef;
703 $ctx->{parent_msg} = $parent;
708 $$skel .= '+ messages / '.SKEL_EXPAND.qq! <a\nhref="#b">top</a>\n!;
710 # nb: mutt only shows the first Subject in the index pane
711 # when multiple Subject: headers are present, so we follow suit:
712 my $subj = $hdr->header('Subject') // '';
713 $subj = '(no subject)' if $subj eq '';
714 $ctx->{prev_subj} = [ split(/ /, subject_normalized($subj)) ];
716 $ctx->{prev_attr} = '';
717 $ctx->{prev_level} = 0;
718 $ctx->{skel} = $skel;
720 # reduce hash lookups in skel_dump
721 $ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
722 walk_thread(thread_results($ctx, $msgs), $ctx, \&skel_dump);
724 $ctx->{parent_msg} = $parent;
727 sub _parent_headers {
728 my ($hdr, $over) = @_;
730 my @irt = $hdr->header_raw('In-Reply-To');
733 my $lnk = PublicInbox::Linkify->new;
734 $rv .= "In-Reply-To: $_\n" for @irt;
735 $lnk->linkify_mids('..', \$rv);
737 $refs = references($hdr);
738 my $irt = pop @$refs;
740 my $html = ascii_html($irt);
741 my $href = mid_href($irt);
742 $rv .= "In-Reply-To: <";
743 $rv .= "<a\nhref=\"../$href/\">$html</a>>\n";
747 # do not display References: if search is present,
748 # we show the thread skeleton at the bottom, instead.
751 $refs //= references($hdr);
753 @$refs = map { linkify_ref_no_over($_) } @$refs;
754 $rv .= 'References: '. join("\n\t", @$refs) . "\n";
759 # returns a string buffer via ->getline
761 my ($ctx, $hdr) = @_;
762 my $ibx = $ctx->{-inbox};
764 my $skel = " <a\nhref=\"$upfx\">index</a>";
768 thread_skel(\$skel, $ctx, $hdr);
773 if (my $n = $ctx->{next_msg}) {
775 $next = "<a\nhref=\"$upfx$n/\"\nrel=next>next</a>";
778 my $par = $ctx->{parent_msg};
783 if (my $p = $ctx->{prev_msg}) {
784 $prev = mid_href($p);
785 if ($p && $par && $p eq $par) {
786 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
787 'rel=prev>prev parent</a>';
790 $prev = "<a\nhref=\"$upfx$prev/\"\n" .
792 $parent = " <a\nhref=\"$u\">parent</a>" if $u;
794 } elsif ($u) { # unlikely
795 $parent = " <a\nhref=\"$u\"\nrel=prev>parent</a>";
797 $rv .= "$next $prev$parent ";
799 $rv .= qq(<a\nhref="#R">reply</a>);
802 $rv .= msg_reply($ctx, $hdr);
805 sub linkify_ref_no_over {
807 my $href = mid_href($mid);
808 my $html = ascii_html($mid);
809 "<<a\nhref=\"../$href/\">$html</a>>";
814 'm' . id_compress($msgid, 1);
818 my ($upfx, $mid) = @_;
820 my $href = mid_href($mid);
821 my $html = ascii_html($mid);
822 qq{[parent not found: <<a\nhref="$upfx$href/">$html</a>>]};
827 $level ? INDENT x ($level - 1) : '';
831 my ($ctx, $level, $node, $idx) = @_;
832 ++$ctx->{root_idx} if $level == 0;
833 if ($node->{id} eq $ctx->{mid}) {
834 $ctx->{found_mid_at} = $ctx->{root_idx};
840 sub strict_loose_note ($) {
843 " -- strict thread matches above, loose matches on Subject: below --\n";
845 if ($nr > PublicInbox::Over::DEFAULT_LIMIT()) {
847 " -- use mbox.gz link to download all $nr messages --\n";
853 my ($ctx, $msgs) = @_;
854 require PublicInbox::SearchThread;
855 my $rootset = PublicInbox::SearchThread::thread($msgs, \&sort_ds, $ctx);
857 # FIXME: `tid' is broken on --reindex, so that needs to be fixed
858 # and preserved in the future. This bug is hidden by `sid' matches
859 # in get_thread, so we never noticed it until now. And even when
860 # reindexing is fixed, we'll keep this code until a SCHEMA_VERSION
861 # bump since reindexing is expensive and users may not do it
863 # loose threading could've returned too many results,
864 # put the root the message we care about at the top:
865 my $mid = $ctx->{mid};
866 if (defined($mid) && scalar(@$rootset) > 1) {
867 $ctx->{root_idx} = -1;
868 my $nr = scalar @$msgs;
869 walk_thread($rootset, $ctx, \&find_mid_root);
870 my $idx = $ctx->{found_mid_at};
871 if (defined($idx) && $idx != 0) {
872 my $tip = splice(@$rootset, $idx, 1);
873 @$rootset = reverse @$rootset;
874 unshift @$rootset, $tip;
875 $ctx->{sl_note} = strict_loose_note($nr);
883 require PublicInbox::ExtMsg;
884 PublicInbox::ExtMsg::ext_msg($ctx);
888 my ($prev_subj, $subj, $val) = @_;
890 my $omit = ''; # '"' denotes identical text omitted
891 my (@prev_pop, @curr_pop);
892 while (@$prev_subj && @$subj && $subj->[-1] eq $prev_subj->[-1]) {
893 push(@prev_pop, pop(@$prev_subj));
894 push(@curr_pop, pop(@$subj));
897 pop @$subj if @$subj && $subj->[-1] =~ /^re:\s*/i;
898 if (scalar(@curr_pop) == 1) {
900 push @$prev_subj, @prev_pop;
901 push @$subj, @curr_pop;
906 sub skel_dump { # walk_thread callback
907 my ($ctx, $level, $node) = @_;
908 my $smsg = $node->{smsg} or return _skel_ghost($ctx, $level, $node);
910 my $skel = $ctx->{skel};
911 my $cur = $ctx->{cur};
912 my $mid = $smsg->{mid};
914 if ($level == 0 && $ctx->{skel_dump_roots}++) {
915 $$skel .= delete($ctx->{sl_note}) || '';
918 my $f = ascii_html($smsg->{from_name});
919 my $obfs_ibx = $ctx->{-obfs_ibx};
920 obfuscate_addrs($obfs_ibx, $f) if $obfs_ibx;
922 my $d = fmt_ts($smsg->{ds});
923 my $unmatched; # if lazy-loaded by SearchThread::Msg::visible()
924 if (exists $ctx->{searchview}) {
925 if (defined(my $pct = $smsg->{pct})) {
926 $d .= (sprintf(' % 2u', $pct) . '%');
932 $d .= ' ' . indent_for($level) . th_pfx($level);
934 $ctx->{first_level} ||= $level;
936 if ($attr ne $ctx->{prev_attr} || $ctx->{prev_level} > $level) {
937 $ctx->{prev_attr} = $attr;
939 $ctx->{prev_level} = $level;
944 $$skel .= "<b>$d<a\nid=r\nhref=\"#t\">".
945 "$attr [this message]</a></b>\n";
948 $ctx->{prev_msg} = $mid;
951 $ctx->{next_msg} ||= $mid;
954 # Subject is never undef, this mail was loaded from
955 # our Xapian which would've resulted in '' if it were
956 # really missing (and Filter rejects empty subjects)
957 my @subj = split(/ /, subject_normalized($smsg->{subject}));
958 # remove common suffixes from the subject if it matches the previous,
959 # so we do not show redundant text at the end.
960 my $prev_subj = $ctx->{prev_subj} || [];
961 $ctx->{prev_subj} = [ @subj ];
962 my $omit = dedupe_subject($prev_subj, \@subj, '" ');
965 my $subj = join(' ', @subj);
966 $subj = ascii_html($subj);
967 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
968 $end = "$subj</a> $omit$f\n"
974 my $mapping = $unmatched ? undef : $ctx->{mapping};
976 my $map = $mapping->{$mid};
977 $id = id_compress($mid, 1);
979 $map->[0] = "$d<a\nhref=\"$m\">$end";
982 $m = $ctx->{-upfx}.mid_href($mid).'/';
984 $$skel .= $d . "<a\nhref=\"$m\"$id>" . $end;
989 my ($ctx, $level, $node) = @_;
991 my $mid = $node->{id};
992 my $d = ' [not found] ';
993 $d .= ' ' if exists $ctx->{searchview};
994 $d .= indent_for($level) . th_pfx($level);
995 my $upfx = $ctx->{-upfx};
996 my $href = $upfx . mid_href($mid) . '/';
997 my $html = ascii_html($mid);
999 my $mapping = $ctx->{mapping};
1000 my $map = $mapping->{$mid} if $mapping;
1002 my $id = id_compress($mid, 1);
1003 $map->[0] = $d . qq{<<a\nhref=#r$id>$html</a>>\n};
1004 $d .= qq{<<a\nhref="$href"\nid=r$id>$html</a>>\n};
1006 $d .= qq{<<a\nhref="$href">$html</a>>\n};
1008 ${$ctx->{skel}} .= $d;
1014 (eval { $a->topmost->{smsg}->{ds} } || 0) <=>
1015 (eval { $b->topmost->{smsg}->{ds} } || 0)
1019 # accumulate recent topics if search is supported
1020 # returns 200 if done, 404 if not
1021 sub acc_topic { # walk_thread callback
1022 my ($ctx, $level, $node) = @_;
1023 my $mid = $node->{id};
1024 my $smsg = $node->{smsg} // $ctx->{-inbox}->smsg_by_mid($mid);
1026 my $subj = subject_normalized($smsg->{subject});
1027 $subj = '(no subject)' if $subj eq '';
1028 my $ds = $smsg->{ds};
1029 if ($level == 0) { # new, top-level topic
1030 my $topic = [ $ds, 1, { $subj => $mid }, $subj ];
1031 $ctx->{-cur_topic} = $topic;
1032 push @{$ctx->{order}}, $topic;
1036 # continue existing topic
1037 my $topic = $ctx->{-cur_topic}; # should never be undef
1038 $topic->[0] = $ds if $ds > $topic->[0];
1039 $topic->[1]++; # bump N+ message counter
1040 my $seen = $topic->[2];
1041 if (scalar(@$topic) == 3) { # parent was a ghost
1042 push @$topic, $subj;
1043 } elsif (!defined($seen->{$subj})) {
1044 push @$topic, $level, $subj; # @extra messages
1046 $seen->{$subj} = $mid; # latest for subject
1047 } else { # ghost message
1048 return 1 if $level != 0; # ignore child ghosts
1049 my $topic = $ctx->{-cur_topic} = [ -666, 0, {} ];
1050 push @{$ctx->{order}}, $topic;
1057 my $order = delete $ctx->{order}; # [ ds, subj1, subj2, subj3, ... ]
1059 $ctx->{-html_tip} = '<pre>[No topics in range]</pre>';
1064 my $ibx = $ctx->{-inbox};
1065 my $obfs_ibx = $ibx->{obfuscate} ? $ibx : undef;
1067 # sort by recency, this allows new posts to "bump" old topics...
1068 foreach my $topic (sort { $b->[0] <=> $a->[0] } @$order) {
1069 my ($ds, $n, $seen, $top_subj, @extra) = @$topic;
1071 next unless defined $top_subj; # ghost topic
1072 my $mid = delete $seen->{$top_subj};
1073 my $href = mid_href($mid);
1074 my $prev_subj = [ split(/ /, $top_subj) ];
1075 $top_subj = ascii_html($top_subj);
1078 # $n isn't the total number of posts on the topic,
1079 # just the number of posts in the current results window
1083 $anchor = '#u'; # top of only message
1085 $n = " ($n+ messages)";
1086 $anchor = '#t'; # thread skeleton
1089 my $mbox = qq(<a\nhref="$href/t.mbox.gz">mbox.gz</a>);
1090 my $atom = qq(<a\nhref="$href/t.atom">Atom</a>);
1091 my $s = "<a\nhref=\"$href/T/$anchor\">$top_subj</a>\n" .
1092 " $ds UTC $n - $mbox / $atom\n";
1093 for (my $i = 0; $i < scalar(@extra); $i += 2) {
1094 my $level = $extra[$i];
1095 my $subj = $extra[$i + 1]; # already normalized
1096 $mid = delete $seen->{$subj};
1097 my @subj = split(/ /, $subj);
1098 my @next_prev = @subj; # full copy
1099 my $omit = dedupe_subject($prev_subj, \@subj, ' "');
1100 $prev_subj = \@next_prev;
1101 $subj = ascii_html($subj);
1102 obfuscate_addrs($obfs_ibx, $subj) if $obfs_ibx;
1103 $href = mid_href($mid);
1104 $s .= indent_for($level) . TCHILD;
1105 $s .= qq(<a\nhref="$href/T/#u">$subj</a>$omit\n);
1109 $ctx->{-html_tip} = '<pre>' . join("\n", @out) . '</pre>';
1113 # only for the t= query parameter passed to overview DB
1114 sub ts2str ($) { strftime('%Y%m%d%H%M%S', gmtime($_[0])) };
1117 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $_[0]);
1118 timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
1121 sub pagination_footer ($$) {
1122 my ($ctx, $latest) = @_;
1123 delete $ctx->{qp} or return;
1124 my $next = $ctx->{next_page} || '';
1125 my $prev = $ctx->{prev_page} || '';
1127 $next = $next ? "$next " : ' ';
1128 $prev .= qq! <a\nhref='$latest'>latest</a>!;
1130 "<hr><pre>page: $next$prev</pre>";
1133 sub index_nav { # callback for WwwStream
1134 my (undef, $ctx) = @_;
1135 pagination_footer($ctx, '.')
1138 sub paginate_recent ($$) {
1139 my ($ctx, $lim) = @_;
1140 my $t = $ctx->{qp}->{t} || '';
1141 my $opts = { limit => $lim };
1142 my ($after, $before);
1144 # Xapian uses '..' but '-' is perhaps friendier to URL linkifiers
1145 # if only $after exists "YYYYMMDD.." because "." could be skipped
1146 # if interpreted as an end-of-sentence
1147 $t =~ s/\A([0-9]{8,14})-// and $after = str2ts($1);
1148 $t =~ /\A([0-9]{8,14})\z/ and $before = str2ts($1);
1150 my $ibx = $ctx->{-inbox};
1151 my $msgs = $ibx->recent($opts, $after, $before);
1152 my $nr = scalar @$msgs;
1153 if ($nr < $lim && defined($after)) {
1154 $after = $before = undef;
1155 $msgs = $ibx->recent($opts);
1156 $nr = scalar @$msgs;
1158 my $more = $nr == $lim;
1159 my ($newest, $oldest);
1161 $newest = $msgs->[0]->{ts};
1162 $oldest = $msgs->[-1]->{ts};
1163 # if we only had $after, our SQL query in ->recent ordered
1164 if ($newest < $oldest) {
1165 ($oldest, $newest) = ($newest, $oldest);
1166 $more = 0 if defined($after) && $after < $oldest;
1169 if (defined($oldest) && $more) {
1170 my $s = ts2str($oldest);
1171 $ctx->{next_page} = qq!<a\nhref="?t=$s"\nrel=next>next</a>!;
1173 if (defined($newest) && (defined($before) || defined($after))) {
1174 my $s = ts2str($newest);
1175 $ctx->{prev_page} = qq!<a\nhref="?t=$s-"\nrel=prev>prev</a>!;
1182 my $msgs = paginate_recent($ctx, 200); # 200 is our window
1184 walk_thread(thread_results($ctx, $msgs), $ctx, \&acc_topic);
1186 PublicInbox::WwwStream->response($ctx, dump_topics($ctx), \&index_nav);
1189 sub thread_adj_level {
1190 my ($ctx, $level) = @_;
1192 my $max = $ctx->{cur_level};
1194 return ('', '') if $max == 0; # flat output
1196 # reset existing lists
1197 my $beg = $max > 1 ? ('</ul></li>' x ($max - 1)) : '';
1198 $ctx->{cur_level} = 0;
1200 } elsif ($level == $max) { # continue existing list
1202 } elsif ($level < $max) {
1203 my $beg = $max > 1 ? ('</ul></li>' x ($max - $level)) : '';
1204 $ctx->{cur_level} = $level;
1205 ("$beg<li>", '</li>');
1206 } else { # ($level > $max) # start a new level
1207 $ctx->{cur_level} = $level;
1208 my $beg = ($max ? '<li>' : '') . '<ul><li>';
1213 sub ghost_index_entry {
1214 my ($ctx, $level, $node) = @_;
1215 my ($beg, $end) = thread_adj_level($ctx, $level);
1216 $beg . '<pre>'. ghost_parent($ctx->{-upfx}, $node->{id})